MacBreak Weekly 1029 Transcript
Please be advised that this transcript is AI-generated and may not be word-for-word. Time codes refer to the approximate times in the ad-free version of the show.
Leo Laporte [00:00:00]:
It's time for MacBreak Weekly. Jason Snell has the week off. Congratulations to his kid graduating from college. But good news. John Gruber of Daring Fireball sits in along with Andy Ihnatko and Christina Warren. And of course, we are going to analyze a week later now what Apple announced with Siri and Apple's new AI. What it really means, where it comes from, how to use it. John's got a lot of experience. He's been using it all week. That's coming up next on MacBreak Weekly.
Leo Laporte [00:00:34]:
This is MacBreak Weekly, episode 1029, recorded Tuesday, June 16, 2026: Intimate Functionalities. It's time for MacBreak Weekly, the show. We cover the latest Apple news. And here we are one week out of the reality distortion field. So this is when the actual story begins. Christina Warren is here, developer advocate at GitHub.
Leo Laporte [00:01:04]:
Hello, Christina.
Christina Warren [00:01:06]:
Hello, Leo.
Leo Laporte [00:01:07]:
Christina was on TWiT, so she's getting a little overdose of me.
Christina Warren [00:01:10]:
I'm sorry, I was gonna say, it's not that I'm not getting an overdose of you. The audience is getting an overdose of me.
Leo Laporte [00:01:15]:
Oh, no, they're happy to see you. No, no, no, don't get that wrong. Also here from enotko.com. i'm happy to say that Mr. Andy Ihnatko. Hello, Andrew. Chief Editor, ihnatko.com.
Andy Ihnatko [00:01:26]:
What can I say? I nailed the interview. They said, good fit.
Leo Laporte [00:01:31]:
But we had to. We had.
Andy Ihnatko [00:01:32]:
I'm on probation for 60 days, so Jason Snell's out.
Leo Laporte [00:01:35]:
We had to search far and wide for somebody whose website's been going on longer than yours, Andy. And we. And we found him. John Gruber is here from Daring Fireball legend. Hello, John. Good to see you.
John Gruber [00:01:47]:
Hey!
Leo Laporte [00:01:48]:
How are the Yankees this year?
John Gruber [00:01:50]:
Excellent.
Leo Laporte [00:01:51]:
Okay, so you'll be in a good mood.
John Gruber [00:01:54]:
Yes.
Leo Laporte [00:01:55]:
Just checking. You're not a Knickerbockers fan though, are you?
John Gruber [00:02:01]:
No, but I'm happy to see them win. I don't really.
Leo Laporte [00:02:04]:
Celtics fan. For real. Right.
John Gruber [00:02:06]:
You know, I've evolved into. It's the one sport where I root for my local team, the Sixers.
Leo Laporte [00:02:12]:
Oh, that's right, you're in Philly. What am I thinking? Yeah, but it's.
John Gruber [00:02:19]:
How could you not root for the next year?
Andy Ihnatko [00:02:21]:
I don't know.
John Gruber [00:02:21]:
They're a local team.
Leo Laporte [00:02:22]:
Although our 22 year old came upstairs yesterday and said they're rioting in New York City. I said, no, they're not. That's the celebration? No, they burn buses.
Andy Ihnatko [00:02:31]:
Well, that's just usually a good thing to say. There's at least a 70% chance you're right that there's someone's writing about something somewhere.
Leo Laporte [00:02:38]:
I remember when the Eagles won the super bowl, they had to grease the light poles down.
John Gruber [00:02:43]:
That didn't stop them.
Leo Laporte [00:02:45]:
No, it didn't.
John Gruber [00:02:48]:
It's. It is literally the most useless job in the city of Philadelphia to grease the polls before the Phillies or the Eagles win a championship. Because the people who are going to climb them, climb them anyway. So what was the point of grease?
Andy Ihnatko [00:03:02]:
That's the point. And also they're gonna, by the time they get back down, they're gonna be covered with grease and therefore harder for the police to cuff. It's a very self defeating plan.
Leo Laporte [00:03:10]:
That's right. Yeah.
John Gruber [00:03:11]:
And the weird tradition, it comes from South Philly and the Italian American community and when they have like holidays where they do it, they have contests, they grease the polls, they train with greased poles.
Leo Laporte [00:03:22]:
They train. That's right. Very important. John, you were, you were at, you were in Cupertino last week. Jason Snell was. I guess he's still there. I don'. He's on vacation.
Leo Laporte [00:03:35]:
He had a very long time.
Andy Ihnatko [00:03:36]:
He'll never return. No, he'll never return.
Leo Laporte [00:03:38]:
He'll be back next week. But I'm so glad we could get John here. I watched the talk show, it was exciting. You had Neelay Patel back and Joanna Stern, who is not, I'm happy to say, a robot. And actually the three of you, or maybe it was you who said it, there is that reality distortion field when you're in the circle, the force field of what Apple's saying, kind of even without Steve Jobs, encompasses you. But now here we are a week later and we can kind of absorb and think. What are you, what's your take on Apple's AI announcement?
John Gruber [00:04:15]:
I think they nailed it. I've been using it primarily on a spare iPhone since Monday. I mean, I started installing it a week ago right after the keynote. And it's, you know, two years after they initially announced it. The stuff they said would work.
Leo Laporte [00:04:34]:
Works because you were a little, let's face it, a little bitter about the announcement two years ago.
John Gruber [00:04:41]:
Man, I wouldn't say that I was bitter. I tried not to.
Leo Laporte [00:04:44]:
Something.
John Gruber [00:04:45]:
Well, I said something is rotten.
Leo Laporte [00:04:48]:
Oh, maybe it was something else. Okay. I just assumed you meant that.
John Gruber [00:04:52]:
Yeah, but I think that piece holds up.
Leo Laporte [00:04:56]:
Oh, it does, it does.
John Gruber [00:05:01]:
You know, it's one of the. To me, one of the best things about Apple institutionally is that they and it does come from Steve Jobs. He wanted to show things when they were finished and he knew. It just seemed like intuitively that announce it. You're so excited about something when it feels like it's almost done, right? And everybody, I mean it doesn't matter, matter. Everybody knows this, right? The first 90% takes half the time and the other 10% takes the other 90% of the time. The end of finishing anything, a house project, a work project, anything. The end takes more time.
John Gruber [00:05:47]:
And so you're excited because you feel like you're 90% there or 70% there and you want to tell the world because you feel like, oh, the end's in sight, let's announce it now. That's just not the Apple way they announce things when they are ready to ship. Like, hey, you can buy this next Tuesday or you can buy this, you can pre order Friday or you can download it later today after this keynote. And so the way that they pre announced what they called Apple Intelligence two years ago.
Leo Laporte [00:06:16]:
Oh, sorry, I didn't clearly, I apologize. That was not commentary.
John Gruber [00:06:23]:
Clearly.
Andy Ihnatko [00:06:24]:
Mute the audience mics. But they are loving what you're laying down, man.
John Gruber [00:06:30]:
Clearly without the end in sight. I mean, and again, do I think they thought it was going to be canceled? No, of course not. They would have to be insane, right? They clearly thought there was some chance that they would be able to ship it on the promised schedule, but they obviously couldn't and they had to go back to the drawing board. And a lot of the explanation we saw last week at wwdc, the hardest to say on a podcast conference name known to man was explaining how they went back to the drawing board with all new and all new way to do their models. But the end result is what they've landed on really works and it really is what a very Apple like approach to on device AI.
Leo Laporte [00:07:15]:
I think that was our takeaway last week, right Christina. We, we felt like they. Anyway, to me it's, it's what the vast majority of people who haven't much of an experience with an AI or only with ChatGPTs chat bot will be using going forward and this will give them, I think a good sense of what AI can do. It's, you know, I mean anybody who's been using AI seriously is way be well beyond what Siri is going to be able to do. But it's still for the billion and a half people using iPhones, this is going to be a big thing. You're muted, Christina.
Christina Warren [00:07:49]:
Sorry. No, it's great to have a Good default. I mean, that's really what it comes down to is that it's the default.
Leo Laporte [00:07:55]:
Yeah.
Christina Warren [00:07:55]:
It's the fact that it's a default and it's the fact that I think we can all be honest about it and some of us have been since 2024. Like, the default has been pretty terrible. And so if it's not terrible, even if that's damning with praise, that's a really good thing. If the default experience. And I think especially when you get into things like how this will be able to work with doing things, with shortcuts and with other types of automations that only Apple can really let you do with a fully integrated, vertically integrated ecosystem that I think is going to be fantastic for a lot of folks. But yeah, you're right. I mean, I think there are many, many people who already use Claude and ChatGPT and Gemini, but there are lots of other people who don't or only use it sparingly. And the fact that it can be built into their device and, you know, hopefully work as it has been shown off to work is only a good thing.
Leo Laporte [00:08:48]:
Yeah. And there's no chance that Siri is so smart that the Commerce Department will ban it, is there? Oh, there's a little.
Andy Ihnatko [00:08:57]:
Not so long as Tim has more golden tchotchkes to manufacture.
Leo Laporte [00:09:01]:
Keep. Keep making those. Keep making those gold bars, Tim. You may be needing them. Yeah. But since we met last, it's kind of ironic right after the show Fable was released last Tuesday and by Friday, banned by the federal government. We'll talk more about that, of course, on Intelligent Machines on Wednesday. So did it.
Leo Laporte [00:09:25]:
Do you think that. Well, what do you think, Andy? Does this live up to what Apple promised two years ago?
Andy Ihnatko [00:09:31]:
I think we forget about what Apple promised two years ago.
Leo Laporte [00:09:33]:
Yeah, I think they'd like to forget about it.
Andy Ihnatko [00:09:35]:
I honestly think that, you know, while I was watching the keynote, I was sort of repeating the demonstrations that they were giving on my Pixel Phone through Gemini and a couple other AIs. What I loved about it was that, no, they didn't show, show off anything that I haven't been able to do daily for the past year or so. Okay. But that's not terribly impressive. What's impressive is how, well, it seems like a feature of the iPhone, a feature of the entire experience. What would have been a big, big fail is if they'd come back and said, here is a feature of the keyboard that, where it can predict some text for you. Here is a feature of the launcher where it can automatically take things down for you. Here's a new feature we've added to Calendar and now here's a separate app that's our intelligent assistant.
Andy Ihnatko [00:10:23]:
What did is they basically said there is an assistant that much like if you are lucky enough to rate an office with an assistant at a desk outside the door, they're always there out of sight when you call upon them. They'll stick their head in the door and fix things for you and get involved in what you need them to do. But the thing is it's always a presence even if you're not actually using it. And it seems like such a natural and integral way to enhance the entire iPhone experience that. I mean, I've got a piece that I didn't finish yesterday because I got thinking too hard about it about I used to think that AI was one of the great levelers because look, I can basically get whatever AI assistant I want on whatever phone I want. So lock in is no longer as big a problem. So if I'm using AI as one of my points of contact for a lot of the features that I'm doing, however, Apple, either intentionally or just because they like to make things that are beautiful and elegant, they've built something that is not in any way on the same level as an app that works alongside even an app that is integrated with the other apps on your phone. Gemini is great.
Andy Ihnatko [00:11:32]:
I love Gemini. I continue to pay 20 bucks a month forever for it, but it will never be as accessible and as part of the day to day, moment by moment experience on the iPhone as Siri is. So I'm really, really excited to find out if they can fill out the rest of the experience to make it as reliable and consistent as Gemini is. Gemini still has. In addition to all the stuff it can do, it's also the thing where I trust it to have an answer to whatever problem I throw at it at this point. Even if it's not perfect, I know how it works and I know how to get it to do what I want. And I'm pretty sure it's going to get me 90% there. It would be terrible if Siri AI is just good at taking down those calendar things.
Andy Ihnatko [00:12:15]:
And I can count off, okay, Here are these 11 things I'm going to throw to Siri. Anything else, I'm going to switch over to Claude, I'm going to switch over to Gemini. If they can actually mature this in the course of a year or two where I'm just using it over Gemini, my favorite agent, because it's just easier to use and it's more in my mind space when I'm doing something on my iPhone that would be a miracle. But I think it's a very, very attainable miracle for them after what they've
Leo Laporte [00:12:39]:
shown off last week you actually, I think one of the most interesting things is that they did say with app intents developers can also build this intelligence into their own apps. And of course that's a two way street because those apps will then feed information back to Siri which can then collate information from a variety of apps. But you point this out on Daring Fireball John that the private cloud, the highest end compute is actually going to be limited to people with a lot of money.
Andy Ihnatko [00:13:09]:
No.
Leo Laporte [00:13:09]:
Or vice versa.
John Gruber [00:13:11]:
No. Right. You can't buy your way out of it which is actually even more limited.
Leo Laporte [00:13:17]:
You have to have fewer than 2 million first time app downloads.
John Gruber [00:13:20]:
Right. And it's not just for an app that you want to add private cloud or Apple intelligence to. And I think I could be wrong but I'm pretty sure that it's like if an app is using Apple intelligence I don't, I could be wrong but I think that you ask the system to do something and then the system determines whether it's going to try to do it with an on device model or oh that's difficult. We're going to send that off to private cloud computing.
Leo Laporte [00:13:51]:
That's the impression I got that they
John Gruber [00:13:52]:
were orchestrating the call and that they don't really want individual apps to be making that decision. Let the system decide what's difficult for that so that you don't but basically to use that in a third party app in Apple intelligence you have to be part of the small business program which has its own million dollars in revenue. I forget what but somewhere around there and 90 some percent of developers do qualify for the small business program. But in addition to that, whatever the developer's most popular app is of all time, all time it has to be under 2 million initial downloads.
Leo Laporte [00:14:34]:
This is to what keep Amazon from using it or Snapchat or Meta or
John Gruber [00:14:38]:
what I, I hope that Apple that maybe they didn't think too much of. I, I guess, I guess the gist is I think and again Apple being Apple they're not explaining what the explanation is but I think the basic explanation is the, the Occam's razor applies that the most important thing from Apple's perspective is Siri AI. That's what most people are going to use and that's Apple's use case of private cloud compute. And then add to that the things where Apple itself is using Apple intelligence, like numbers and pages and Keynote with the creator Studio editions where they've added some AI features. Obviously Apple is not limited to apps. With only 2 million downloads, it's Apple. So Apple is I think worried and I think probably reasonably so that they're going to be under, you know, whatever they expect their usage of private cloud compute in their servers and Apple, some of the, some of the queries are still going to go to the ones running Apple silicon. But the big message was that the most complex models are the ones running Gemini on the server in Google owned facilities but running Apple software.
John Gruber [00:15:56]:
You know, it's a collaboration, blah blah, blah within chips.
Leo Laporte [00:15:59]:
But is it really Gemini? They don't, they kind of pretend it's not. Well, it's, they, they say it's Apple foundation models, train Gemini or something.
John Gruber [00:16:09]:
Yeah, something like that. They're not, at least it's not Gemini in terms of when you ask a question and in Siri and then you go to your Gemini.
Leo Laporte [00:16:20]:
Right.
John Gruber [00:16:20]:
It's not going to show up in your Gemini history.
Leo Laporte [00:16:22]:
Right.
John Gruber [00:16:23]:
It's, it's seemingly different. It really is. It's like starting with a. Go back far enough to the roots and.
Leo Laporte [00:16:30]:
But is it, I mean their own version. It could be white label Gemini or it could be something else, I don't know. Anyway, but for third parties you're not going to get access to that if you're that big. You're going to.
John Gruber [00:16:43]:
Yeah, you can't even pay for it. Right. I spoke to one developer who, or guy Rambo who tweeted about it on Mastodon that he's, he's got an old app from 2010 and it's sitting at like 1.6 million downloads, you know, so now what does he do? Does he pull the app from the app Store so that it doesn't cross 2 million and he's locked out of Apple and hopefully they'll just rethink this limitation. Basically they don't want big developers using it because they don't know if they can keep up and what their bill from Google is going to be.
Leo Laporte [00:17:15]:
Right, well that makes sense. It's just, it's economic, not anti competitive. That makes sense actually.
John Gruber [00:17:20]:
Yeah. And I think, you know, come next year or maybe halfway through the year if, if everything looks good and hey, once Siri AI has rolled out to the general iOS population in the fall when iOS 27 ships for real and it's out of beta and real people are using it around the world or wherever, whatever countries they initially roll it out in, if they can keep up with it and it doesn't fall down and the server bill isn't too big from Google, maybe they'll loosen the restrictions on third party apps. But I think it's sort of like one thing and one thing first, and that's Apple's built in use of Apple intelligence and Siri first. And let's make sure that works before they expand it to third party developers.
Leo Laporte [00:18:04]:
One of my favorite cryptographers, Matthew Green from Johns Hopkins, says private cloud isn't all that private. He says private inference isn't private enough. His issue is that it's going to inevitably leak information because when it calls out to the world, it's going to call out with your prompt and that's going to leak out. So he gives actually a scenario in here, your calendar, for instance. If you're going to try to make a reservation at a restaurant or actually with a number of participants, that information is going to have to leak out in order for it to work. So he's a cryptographer and so he always thinks of the adversarial situation. But I think this is, this is important to remember. You can say private cloud and we've seen this before with data anonymization and other things, but it isn't always as private as one would think.
Leo Laporte [00:19:03]:
Nevertheless, this isn't, isn't this what Apple's biggest pitch is? We may not have the smartest AI, but we have the most private AI. Yes. You agree, Christina?
Christina Warren [00:19:15]:
I mean, maybe. I mean I think that's definitely the marketing narrative. Right. Like, I think that's definitely what they're pushing and there is definitely some truth to it. Right. That Apple same. Look, we're not training on this data. We're not trying to use this to build something else.
Christina Warren [00:19:30]:
Although I would, I guess like to see some of the really fine print about how they're getting any sort of data to train anything I find.
Leo Laporte [00:19:40]:
Where's that data coming from?
Christina Warren [00:19:42]:
Right. I was going to say I find it hard to believe that all of it is synthetic. It might be one of those things that you opt in. Like if you opt in to let Siri improve or if you opt in to like give developers feedback, they might,
Leo Laporte [00:19:51]:
they did say it'll be auditable by third parties.
Christina Warren [00:19:53]:
Right, right. And so it might be one of those things where, you know, we're not taking your prompt history and specific things, but we might take general usage based things to cover behavior. I'm not sure at the same time. So I feel like some of it is, you know, a marketing narrative, but I think a successful one and there's some validity to that. At the same time, you know, I think Matthew Green is incredibly bright and I'm not saying that he is wrong, but I do wonder if that misses the point a little bit. Right. Because I don't think that there exists a way where we could have AI that would actually be useful that would be as privacy and would do all the thing privacy focused and would do all the things that would actually make it useful that anyone would actually want to use. Like there has to be a certain amount of give and take on that.
Christina Warren [00:20:40]:
And that's why I hope that there are ways for people who really are like, I don't want any of this, that you can just turn all of this off. But I think that this becomes kind of one of those implicit agreements that we all have to make. And this, this does get into ironically in a different way, but it kind of underscores some of the tension with the EU I think a little bit, which is that you have, you have to take the trade offs if you want to use this technology. And some of that means that yeah, some of the prompt information or some other data transfer will exist but that exact that happens anytime you interact with a third party API. It's not like this is a new thing.
Leo Laporte [00:21:15]:
So yeah, I mean Apple will be better than everybody else. They may not be perfect.
Christina Warren [00:21:18]:
Right, exactly. And well, maybe they'll be better, maybe they won't, I don't know. But I think that they definitely are using it as a selling point for themselves. But I just don't think that it's possible for us to have AI in this way that's going to be actually useful and fully localized and never talk to any sort of third party and have any ability for data to ever leak in any conceivable way.
Leo Laporte [00:21:42]:
It's just we were debating this on security now we had somebody email and say, well aren't these companies collecting everything? How can I do this privately? And I thought, I foolishly, I guess thought no, no, no. Unless you ask them to remember stuff, they're not collecting this. Well, I got a signal message from a guy who shall remain nameless who says speaking with insider knowledge from one of the frontier AI companies, they absolutely extract every bit of knowledge and insight from user chats. I said, well wait, wait, what if I don't turn on memory in the chat app? He says turning off memory means they aren't exposed to the user but they're still extracted by the provider for their use according to the terms and conditions. Yes. Memories are extracted and leveraged in first party CLI harnesses as well.
Christina Warren [00:22:24]:
Yeah. Unless you're an enterprise. Enterprise. It's a little bit different. There's business agreements.
Leo Laporte [00:22:28]:
They're sensitive. Right?
Christina Warren [00:22:30]:
Yeah.
Andy Ihnatko [00:22:30]:
Christina, I always mean to ask you, like when Google announced their private AI computer basically as a response to Apple's whole like fireworks show about it's absolutely engineering wise impossible for Apple or any other company to see any of the data that you're sending to us when, when we need to use an AI cloud model. If they seem to, they seem to be kind of equivalent in terms of what I would want as a user is they're getting, they're going, they seem to be going at it in different directions where Google is not necessarily writing an entirely new operating system that's privacy cloud AI based. But I could also, also given explained it. I couldn't see how data could get out or be observed by Google. Like have you ever, have you looked at those two? What are they kind of equivalent or
Christina Warren [00:23:17]:
is that they seem very similar. They seem very similar and there might be differences in the implementation and there might be differences in certain things they can do but the idea is definitely the same. But I think that the big thing there is that not everything that you're using associated with the Google product and that's the thing that everybody should know even if you're paying, if you're not a business user, if that's you know, thing isn't on. Google has always been very upfront about the fact that it uses your data to make things better. But I do think that in terms of the way that the, that Google is like defined private AI compute and how to build private AI stuff for, for some of the Gemini models in the cloud it is similar to what Apple is doing. But like the Gemini consumer apps. No, I mean that's, that's part of the terms of service. You know that this is the thing that's happening that yes, we can train on your data, we can train on your prompts.
Leo Laporte [00:24:06]:
Admiral Birdie says in our discord. Yeah, Google for enterprises definitely has a different set of rules. You could tell because it's terrible. There is something to be said for memory. I mean one of the things I do with my agent is I have like three different memory systems because I want it to remember everything but it's local as opposed to putting it up on the cloud. All right. I'LL stop being a negative Nelly. Let's be more positive about this.
Leo Laporte [00:24:31]:
So, John, you've played with Siri. What kinds of queries have you asked? What's your experience been?
John Gruber [00:24:38]:
It's funny because the hardest part about it honestly is remembering that I'm supposed to be asking Siri to be useful.
Leo Laporte [00:24:52]:
You aren't really used to that, are you?
Andy Ihnatko [00:24:53]:
Yeah, it's like your cousin after his second rehab and everyone's telling you that it took. He's responsible now, but I'm not used to asking him to pick up stuff at the market and giving him cash.
John Gruber [00:25:04]:
Yes, exactly. And he is in recovery and he's doing great and you really.
Andy Ihnatko [00:25:08]:
We're happy for him.
John Gruber [00:25:10]:
But you don't. It's, you know, you're always waiting, aren't
Leo Laporte [00:25:13]:
you, for that other shoe to drive,
John Gruber [00:25:14]:
you know, and it's already Saturday night. It is 8 o'. Clock. He's, you know, you know, and it's like, no, he's great. He drive, you know, it's. But it's like little things. Totally realistic. My barber is in a building where to get in the building you need a door code from the street.
John Gruber [00:25:34]:
And I can never friggin remember it. And so I literally was going to get a haircut on Friday, Saturday. And I just asked, hey, what did Tola tell me? That's his name, Tola. What did he tell me the door code is?
Leo Laporte [00:25:47]:
Wow.
John Gruber [00:25:47]:
And it took a couple seconds and then it surfaced a text message from, I don't know, a week or two ago or last time I got a haircut a couple weeks, but showed me the text message in imessage and read the code aloud.
Leo Laporte [00:26:02]:
That's very cool.
John Gruber [00:26:03]:
Yeah. And it's a real realistic scenario. And it's like I knew it was in my text messages and I could have gone in there and looked him up and scrolled back, but I was on the sidewalk walking through the city and just said it aloud to the iPhone. And you know, was it fast? No. But was it too slow? No, it was a totally reasonable amount of time to get the answer and it was correct.
Leo Laporte [00:26:29]:
Was it the same speed as they showed on stage? Roughly.
John Gruber [00:26:31]:
I mean, yeah, I would say so. Which was, you know, a little uncomfortably slow.
Leo Laporte [00:26:35]:
No, but I like that. I appreciated that they did that because they were just being honest about.
John Gruber [00:26:41]:
I loved it. And that was a huge part of my something is rotten in the state of Cupertino was that they clearly had edited the ones from two years ago. And well, as we pointed out, they
Leo Laporte [00:26:52]:
paid a Quarter of a billion dollar settlement for false advertising that might, and
John Gruber [00:26:55]:
probably more honestly probably more than that. Reputationally. Right. Like it's one of those things like what is your reputation worth? You can't really put a dollar on it. And I think whatever they paid in that class action lawsuit, they lost more than that in terms of their, the reputational hit to their, you know. And you know, that's, that's the type of thing, lots of little things that I have been going to chatgpt more for because web search results often just don't put the, the obvious answer first. What time do the Knicks play? You know, what time is the next game in the NBA Finals or something? I mean obviously it's next year at this point, but like last week there were, they were legit questions and it gave the right answer every time. And there were this sort of questions and again there's that button right on the side of the phone begging to be used to ask these sort of questions.
John Gruber [00:27:57]:
And it's so hard to retrain myself to like, hey, I'm, you know, doing this professionally to test it over the summer. But like, you know, last year it wasn't like I had a reason to try Siri after wwdc. There was no reason to try it again.
Leo Laporte [00:28:12]:
But that's the difference. I mean you can, you can ask questions.
John Gruber [00:28:14]:
So those type of questions. I've also been asking questions. Yeah, well, I just want to add the other thing I've been asking are questions that are answered in my Apple Notes. I'm a big user of Apple Notes. I've got like, I don't know, 2,000 of them, 2,147 asking questions that are stored in my Apple Notes and it gives me the right answer every time. I think that's Steve and that is something that like ChatGPT and that Claude can't do.
Leo Laporte [00:28:44]:
Yeah, well, yeah, exactly. You can ask ChatGPT when the Knicks game is. That's any, any, any AI can do that. That. But it doesn't know what's in your notes, it doesn't know what's in your messages. And that's the, I think that's the real value that Apple can add.
Andy Ihnatko [00:28:57]:
That's but, and that's what, that's what gets me about the possibility of lock in. And it's, it's true of, of Google services too. It's like, I don't like, I don't like.
Leo Laporte [00:29:06]:
Does Google do that though? Are they.
Andy Ihnatko [00:29:08]:
Well, I mean they've got the daily brief that they announced they Introduced last month and I've been finding it actually surprisingly useful. Like I didn't plan to start using it but the thing is like that daily, that daily notification, every time I tap it, it does surface something that I might have forgotten about or a message that I totally missed. I'm going to have dinner with nine of my friends on Friday. Possibly partly because like this message that got buried in 500 WWDC mails like it surface at forms. Oh, by the way, your friends are still like waiting for you to reply to this. But the thing is it works right now at least it works great on every single Google service that you have, including your chats with Gemini, your messages, Google messages, everything like that. Once you go outside there, outer mailboxes, if you're using Apple Calendar, if you're using Apple Messages, it's not going to see that and it's not going to get stuff from there. And it would bum me out if I liked Siri so much that I had to switch from my favorite note taking app, which is Notion, and I had to switch to Apple Notes because it's the one that works the best.
Andy Ihnatko [00:30:16]:
I'm sure there's going to be some sort of a bridge, some sort of
Leo Laporte [00:30:17]:
a plug in for notion. These $2 million plus companies, I'm sorry,
Andy Ihnatko [00:30:22]:
Joplin, I'm sorry Joplin, not Notion, but
Leo Laporte [00:30:23]:
I guess they don't need access to use the intents. That's just they don't get private cloud.
Andy Ihnatko [00:30:27]:
So they could, they could use the intents hopefully. But, but the thing is like, and the thing is I have that knowledge and I have that interest that there must be a plugin for Joplin for this. It's an open source community. But the thing is like, well, I'm sorry, just to finish my thought, the people who are just normal consumers, they can't really be expected to put in all that kind of effort and put that trust into a third party. So they're basically going to use what is wired up and working and great and polished and perfect. When you take the phone out of the box, whether it is Gemini on a Pixel phone or whether it is Siri AI on an iPhone. And in a perfect world that I know that can't exist, it would be just as easy to integrate these AI assistants with any tool that you have rather than it being naturally more knowledgeable about its own in house stuff than anything that's outside.
Leo Laporte [00:31:23]:
So do we know if third party apps data will be available to Siri? This was something the developer has to turn on Right, so. So the Joplin developer would have to have. It's an intent to share with Siri. How does that work? Do we know anybody?
Andy Ihnatko [00:31:40]:
I don't. I know how it works on Android.
Leo Laporte [00:31:43]:
I can see why Apple might not want to do that because then you're compelled to use notes and messages.
Andy Ihnatko [00:31:50]:
Well and also to be fair, there's a privacy and security angle to it where they can make sure they can know that the conduit between Siri and their own is they give themselves privileges that they can't. They can't responsibly give to random third party developers.
Leo Laporte [00:32:04]:
It's less useful though, isn't it if you use.
Andy Ihnatko [00:32:07]:
Absolutely.
Christina Warren [00:32:08]:
So there is an app intents framework and that basically, and I'm reading off their website right now, make content and actions discoverable by Apple Intelligence and support system experiences like Siri Spotlight shortcuts and widgets.
Leo Laporte [00:32:20]:
So you could open it up to
Christina Warren [00:32:21]:
Spotlight or just so you could open up to Spotlight and Apple Intelligence. Now how that will be specifically integrated into something like Siri AI I don't know. And how deeply that will go. Okay, here they have a whole thing. Apple Intelligence and Siri integrate your app with Apple Intelligence and bring it to Siri AI. So they do have. I like that they do have ways to do that, but that would mean the developers have to do that. And I'm sure that at least on, on the iOS side, I think macrosos might be a little bit more difficult.
Christina Warren [00:32:49]:
But to be completely candid with you, I don't think macOS is, is going to be a big Apple AI use case.
Leo Laporte [00:32:55]:
Really? I don't think so.
Christina Warren [00:32:56]:
No, because I think that they're. Why, why would you like Claude and ChatGPT and Gemini? Their desktop apps already work just as well. Like to me, default is not good enough on the desktop and I might be alone in that, but I feel like default can be good enough on the phone. On the desktop, I've already been able to do all the things that they're telling me Apple Intelligence can do for years. There's zero reason why I would want to have to jump through additional hoops assuming that my note taking app of choice or something didn't have an app intent, if that makes any sense.
Leo Laporte [00:33:29]:
This is something Google wanted to do for a long time. Remember they had those cards that the Google now cards, you would slide to the right on the pixel and then you'd have like, oh, your plane is on time and you're going to be at the airport in a half an hour. And Google have backed off on that because I think it has a certain amount of creepy. Have you played Andy with Dream Beans yet in Google?
Andy Ihnatko [00:33:53]:
I have not used Dream Beans yet.
Leo Laporte [00:33:54]:
Dream Beans is at Google. All right, I'm going to show you Dream Beans.
Andy Ihnatko [00:33:59]:
We had some fun with that topic a few weeks ago on the Google show.
Leo Laporte [00:34:02]:
This is a Google experiment and it looks into your deepest, darkest secrets and then produces these. You can show my screen, John. It produces these cartoons. Cartoons. So that's me and my wife. Pretty good likeness. And it's telling me that we could get fresh produce at the Eastside Farmer's Market. It's saying I should check my Arch Linux setup for now.
Leo Laporte [00:34:26]:
How does it know I'm using Arch Linux? I don't know, but it does for the atomic arch hijack. I already did try the rebuilt Siri AI app in the Apple iOS 27 developer beta. Look how content I am in my little sweater in my armchair cooking a 48 hour sous vide chuck roast for a midweek dinner. It knows things about me that look, it even knows I have an anova sous vide. It must be looking at my purchases. It's a little creepy. That's exactly what my gym looks like.
Andy Ihnatko [00:34:57]:
Well, this, well, this is, this is
John Gruber [00:34:59]:
like even two years ago.
Andy Ihnatko [00:35:00]:
I was when, when Apple gave us their first cut at Apple Intelligence. I'm saying. You're really saying that Apple has a unique advantage over other AI because they have access to your personal information.
Christina Warren [00:35:12]:
Right?
Andy Ihnatko [00:35:13]:
You are, you are drinking at a high school level. They are. Google is drinking at a Mardi Gras level when it comes to knowing, getting information about you.
Leo Laporte [00:35:22]:
I know I'm unusual because I don't mind this. I want it to know everything and I want it to give me that useful information. But I think a lot of people
Andy Ihnatko [00:35:30]:
would be very like, well, it depends. Like I, I've always said my per. My personal point of view, and it's just a personal point of view is that you enter into a contract, so to speak, with Google and I'm willing to give them exposure to all kinds of data about me and in return they give me services that are greatly, greatly, greatly enhanced by knowledge of what my life is and what my intentions are at any given moment. Other people may not like that deal and hope and if they can turn it down, then I think that overall that's an okay thing. The problem is that when there's this surveillance that's happening sunrise to sunset, even to people who do not want to enter into that bargain, we enter into that bargain with Apple as well. It's just that they don't at this point monetize the data that they're collecting very well. And also to this point before Siri AI, I don't think they do a really good job of enhancing my experience with their personal knowledge of me, but I will. So again that's why I love these features like the Daily Brief where I didn't really set anything up but it was able to remind me, oh by the way, you've also, you were working on this coding project last week.
Andy Ihnatko [00:36:42]:
You might want to look back at this Python library that you're looking at because that looked like that was going to be something that you were very, very interested in. I think that it does a very, very good job at that. I don't know that they always do this, as you say, in the least creepiest manner. One of the things I like about Google is that they are willing to do things and label them an experiment and put them in a place called Labs saying we know this is, we know this is out there. We just think this is cool and based on how people freak out or embrace this like Notebook LM started out as an experiment started, people love it and it revolutionized and it's great.
Christina Warren [00:37:22]:
Revolutionized, yeah, is one of the best things that in my opinion like Google put out at all. And then it obviously became like a real product. But yeah, that they do at least because you know the Google graveyard is real.
Andy Ihnatko [00:37:32]:
They're willing to, they're willing to swing in a miss in public in front of of a 30,000 paying fan.
Leo Laporte [00:37:36]:
There's also the issue of hallucinations. A German court has ruled that Google is liable for the AI overviews if they're wrong, that they can be sued for life. Now this is why Apple, one of the reasons why Apple quite wisely said no Siri for you, for instance, Dreambeans thinks for some reason I'm going to Rome in December. Now I would love to be going to Rome in December, but it says there's a beautiful gate just a short walk from my hotel that's clearly a hallucination. I don't know, I don't know where it got that but nevertheless, I don't know.
John Gruber [00:38:10]:
Check your credit card and make sure
Leo Laporte [00:38:12]:
maybe I am going to Rome.
Andy Ihnatko [00:38:14]:
Now you're gonna.
Leo Laporte [00:38:15]:
Honey, I've got a little surprise for you.
Andy Ihnatko [00:38:17]:
You're gonna see a few burned out and overturned police cars. That means you're heading the right direction.
Leo Laporte [00:38:20]:
That's where the carnival is flying over The Knicks celebration. Anyway, let's take a little break. We will have more. You're watching MacBreak Weekly. Great to have jon gruber of daringfireball.net I want you to tell that story. I wasn't aware of this whole issue of why it's net, even though you have.com, but. Hold the thought. We'll do that in a sec.
Leo Laporte [00:38:45]:
Christina Warren is also here from GitHub where she's developer relations and is a avid AI user. What local model are you using in your framework these days?
Christina Warren [00:38:55]:
God, I think I'm still using Quinn.
Leo Laporte [00:38:57]:
Yeah, I was using Qin until Yesterday when GLM 5.2 came out.
Christina Warren [00:39:01]:
Now I'm all, yeah, I haven't had a chance to play with that one yet. Is it good?
Leo Laporte [00:39:06]:
This is actually now, the real downside of having these agents is like, I keep trying all these new models. It's a really bad. It's a bad habit. But the agent remembers everything, regardless of the model. So it's really. The agent is.
Christina Warren [00:39:19]:
Well, that's the nice thing, right, is that if you're using Hermes or openclaw or whatever, like, that takes care of all the history and memory for you. And then you can just swap the model out.
Leo Laporte [00:39:25]:
GLM 5.2 so far, very impressive. Yes.
Christina Warren [00:39:28]:
Cool.
Leo Laporte [00:39:29]:
I'm very impressed. Also, Andy Ihnatko, who religiously writes all his posts on natco.com by himself and we're proud of him.
Andy Ihnatko [00:39:38]:
In my freezing Garret, just like Laboem,
Leo Laporte [00:39:40]:
only without the music, minus the tuberculosis. I was going to ask John about Daring Fireball for I've trained myself to type Net, but then you said you have.com?
John Gruber [00:39:54]:
what this is. So when I do a media appearance, CNN or something like that, if they, you know, and it's like, we're gonna be on the air in a minute, you sound good, you look good, or whatever. And they're like, daringfireball.com I think I'll let it go because I registered Both domains in 2002, maybe 2001, maybe the site started in 2002, but I might have had the domain a year before I launched, but I got both. But back then I had it in my head that.com was for companies and commerce on the Internet. And it wasn't what I was doing and I wasn't an org. So I didn't even register daringfireball.org because I'm not an organization. I'm on the Internet. I don't know when the domain name system started in the 90s.
John Gruber [00:40:48]:
I just thought .net was like .internet. that's the neutral main top level domain for a site on the Internet. I was going to stick with it. When I launched Daring Fireball, I had both domains and daringfireball.com has just redirected to daringfireball.net for 24 years and counting. It doesn't matter if you get it wrong, but if I could go back in time to my younger self, I'd say give it up. That net thing is, is not going to happen.
Leo Laporte [00:41:22]:
Old timers will remember this for a long time. I called this a netcast and even the opening it said netcasts you love. Finally, my marketing team a couple of years ago said can we please, please just call it a podcast. Nobody knows what the hell you're talking about, Leo. So I love it that you never gave in. This is the advantage you have. You don't have a marketing team.
Andy Ihnatko [00:41:44]:
Also John, I've always thought that it was like, you know how like every house has like the front door that enters into the living room and then there's the side door that enters into the kitchen and everybody who's in the family or friend of the family doesn't bother with the front door. They go around to the, to the kitchen door. I've always thought that your.net is the kitchen door. It's like people who are like part of the group, like don't. They don't even have to call ahead.
John Gruber [00:42:08]:
I love that my mother in law's house is exactly like that. And when somebody, it's like a holiday, it's like Easter Sunday and somebody rings or knocks on the front door. It's like, oh, who's that? Who's that? Who could that possibly be?
Andy Ihnatko [00:42:23]:
If someone's knocking on and rattling the screen door in the kitchen, you actually drop what you're doing and answer the door because it is someone you want to see. Doorbell process server salesman.
Leo Laporte [00:42:33]:
Yes, by Waterloo. In our Discord system when I was growing up, we only had six domains.comedu.mill.org and gov and we liked it.
Andy Ihnatko [00:42:44]:
Oh, you didn't have to type in addresses. Oh, fancy.
Leo Laporte [00:42:49]:
I have for a long time I've been Leo FM on my blog and my email is email. And I still constantly have to tell people, no, not.email.com not.fm.com people are all stuck with that. So I want to talk a little bit about what role Gemini plays, how much Gemini really is inside. Siri AI you went to that little panel they had after the keynote, John, where Rockwell and Federighi Frederighi. And they stood on, they sat in chairs and they did the demos. Mike did the same demo he did in the film, which I thought was. That's kind of a giveaway. It's a little bit of a canned demo.
Leo Laporte [00:43:34]:
But they were adamant that this is not Gemini.
John Gruber [00:43:39]:
Well, Federighi's rant was pretty good and the diagram was pretty good. But I think what at the problem Apple is facing is that Google made the decision marketing wise to use the name Gemini for everything AI related, or at least LLM related. And what was the name that Bard had?
Christina Warren [00:44:01]:
Bard. They had to.
Leo Laporte [00:44:05]:
Bard was doing some bad things.
John Gruber [00:44:08]:
And if they had made different names for the models from the product. And just for example, you know, like you could be using GPT03. I mean, I know that's old, but it was like that's the name of the model. But then what app are you using? Oh, the app is chatgpt right there. It's like effectively they were using the equivalent of GPT 4.5 but Google just calls it Gemini. Right? They're using the mod and they're also being very secretive. Perhaps just because they're Apple, perhaps because it's in the contract with Google that some of the technical details. But they started with the model and then collaborated with Google to build their own model on top of the base model from Google that isn't connected to the rest of the Gemini system.
John Gruber [00:45:12]:
If you have never signed up for a Google account, period, you don't even have a Gmail address, you can use Apple Intelligence and you'd never know you're integrating with something from Google. It is like saying that you're using Microsoft's C compiler, but now it's running on the Mac and it's compiling C code into a binary that you can run. They're using Google's models to build out their own models. And it's effectively, you don't need to know this except or there is no except. You could not know this and you can happily use the new Siri AI and never worry about it. It. But the proof basically is my one week experience using Siri AI is so positive and anything I can think of that's negative is like, that could be a little bit better. But there are no, there have been no.
John Gruber [00:46:09]:
Oh, that is so off the rails. And no modern LLM system would ever make a mistake. There have been no fluffs like that.
Andy Ihnatko [00:46:17]:
And I think that.
John Gruber [00:46:20]:
Yeah, I don't think so. I don't think anything that I would call a hallucination and a couple of times where I've compared some answers versus ChatGPT. If anything, Apple's answers seem to be more accurate. And I think that's basically it, that instead of doing their own research and paying the literally tens of billions of dollars to train a leading edge model, they've just piggybacked on top of Google to borrow one to use as their own. But it is their own.
Leo Laporte [00:46:51]:
There's a lot of pride that Apple has. I think it's a wounded pride in their models. Like we have our own models. Our models are very good. This isn't Gemini and yet it is Gemini. So that's the question. It's not a C compiler, I guess. Christina, we think it's distillation, right? That they trained, post trained the AFM models on Gemini.
Christina Warren [00:47:20]:
Yeah, I mean, look, that's what I think. And I mean John was, was, was in the room with them, so he can speak more to that. But that does seem to be exactly what.
Leo Laporte [00:47:30]:
And by the way, John's right, nobody cares about this. Except.
Christina Warren [00:47:32]:
No, but, but, but, but it is interesting. I mean, but one of the things I think that Federighi apparently said, this is, I'm quoting Macworld, who I guess was able to get this down, said, trained using proprietary data with reinforcement learning and refined using outputs from Gemini Frontier models.
Leo Laporte [00:47:47]:
That's distillation, right?
Christina Warren [00:47:48]:
That's distillation. That's what this is. And that's what I'd assume this was last week. Now, that doesn't mean that the distillation doesn't mean that it operates in a way that might be better in some ways, it could be worse in some ways. Right? Like we've already seen reports from people saying that if you want to get images created of certain historical figures or other things, that the guardrails might be higher than they are if you were to just use the Gemini directly through Google. And as, as, as John also pointed out, and this is a very valid point, and I can say this because I used to have to work on these things and explain the differences, which was sometimes difficult. The Gemini that you get from the Gemini consumer app or from like Gemini Enterprise or whatever is not necessarily the same as if you get it directly from using the Gemini API calls. They're similar, but there are modifications that you make to kind of the base model.
Christina Warren [00:48:39]:
And that again does become difficult to understand as a consumer, although you don't really need to. It's worse, I think, more if you're Google because you call everything Gemini and then It's a different experience, slightly different model on every surface that you're on. And that's, that's confusing. Whereas Apple is just going to have, you know, kind of their distillation on top of the core, you know, Gemini models and I assume that they'll, you know, continue to do reinforcement and, and distillation over time.
Leo Laporte [00:49:07]:
Well, that's my question is this may not be a long term relationship with Gemini. I mean they could after a few years say thank you, bye bye.
Andy Ihnatko [00:49:15]:
You know, they could, I'm sorry, just to interject, that's, that's why this is a, this is an interesting conversation to have because when the, when the partnership, excuse me, when the business relationship was first announced, one of the things that was on the table was okay, so are they, is Gemini going to be part of Apple intelligence and that some, while they, to buy them some time while they build out and get Apple intelligent, their own Apple intelligence models working and then at some point they will pop out the Gemini, pop back in the, the Apple intelligence? Or is it the sort of thing where it's always going to be Apple and Apple's own models but they're getting tuning from Gemini, they're getting training from Gemini, they're getting experience from Gemini so that gradually the dependence on Gemini is going to be less and less and less. But it will never be a case where you're not using an Apple actual model. It's a billion dollars a year plus whatever compute they're going to be buying from. I can't imagine that beyond Apple's revulsion of relying on outside technology. They're going to want to keep writing those checks year after year after year indefinitely.
Christina Warren [00:50:21]:
Well, I mean, but it would cost them so much more to build it themselves.
Leo Laporte [00:50:24]:
Well, exactly.
Christina Warren [00:50:26]:
It would literally cost them more. I was going to say they don't have to. Where are you going to get the, the data centers? Where are you going to get the chips? Where are you going to get the custom silicon for training? Where. How are you going to figure out, how you figure out what data you can use to train the data set? Because then you have to scrape everything which they've already done with their other frontier models or attempts at a frontier model before. But, but I don't know. I think a billion dollars a year is pretty freaking cheap.
Leo Laporte [00:50:50]:
It's really cheap. Google's paying a billion a month to grok for capacity.
Christina Warren [00:50:54]:
So it's, they're paying Apple 20 billion a year for search defaults.
Leo Laporte [00:50:59]:
Right. I wonder if there's a check that goes one way and a check that goes the other way or they just give them a discount, I don't know. That's for the accountants to figure out.
John Gruber [00:51:07]:
Let me jump in here because I've been looking into this and I haven't written about it yet, yet, but I started using this over the last week. It, it popped into my head, good journalistic instinct that I have, like how do we know they're paying $1 billion a year for this? And I started researching it and it really all comes down to Mark Gurman. It's one reporter at one place. And it was, and he was right that they were, he was very much right. It's totally, totally was the first to report it. And it was exactly right that Apple was collaborating with Google on exactly the thing we're talking about, like a sort of white label version of Gemini. He was exactly right about that. And then he put out the $1 billion number and everything you've seen about it since more or less comes back to that point.
John Gruber [00:51:59]:
I don't think it's possible that they negotiated a flat fee. I was thinking about this. It makes no sense because if they negotiated a flat fee, Apple could let users do go crazy. And then Google is like, hey, you guys are using $40 billion of compute a year. And Apple's like, yeah, good deal for $1 billion. Right. It's got to be usage based. It absolutely.
Leo Laporte [00:52:22]:
It makes any sense. If you're going to go into the cloud, you're going to, there'll be limits unless you have an iCloud plus account. Right.
John Gruber [00:52:29]:
So I wouldn't be surprised if from Apple's perspective they were like, we would like to cap our spend at a billion dollars for the next year, you know, and that that was Apple's intention and that maybe that's what Gurman's source or source is told him, that they're looking to spend a billion dollars. And Google was like, well if you use this much, that would cost about a billion. You know, let's say that about a billion. And Apple's doing the math and thinking like users and how much they're going to use. And they're like, okay, that would be about a billion. And, and that's why they put those caps in that we were talking about earlier that for example, third party apps can't just use Apple intelligence and private cloud compute because that would cost Apple more. Well, so I don't think they're spending much. I'll bet that they're hoping to spend a billion dollars a year and to build the capabilities that the Apple intelligence that you can get today in the developer betas, anybody out there with an Apple developer account can install on their phone or iPad.
John Gruber [00:53:32]:
Mac today to get there on their own would have cost hundreds of billions of dollars and maybe not have worked. Right?
Leo Laporte [00:53:40]:
Right.
John Gruber [00:53:40]:
Who knows?
Leo Laporte [00:53:41]:
Well, they may have already spent that. Maybe that's what the problem was that
John Gruber [00:53:44]:
they tried to get on their own. I can't possibly believe that this is just a billion dollar deal. It's used.
Leo Laporte [00:53:51]:
That's a very good point.
Andy Ihnatko [00:53:52]:
I agree. If that number had been thrown out, I've always assumed that that was, was basically fees for turning the wrenches and putting in the pipe. I mean Google, Google's AI or I think it's Google, they could simply have
Leo Laporte [00:54:07]:
been a billion dollars to train on Gemini to post training.
Andy Ihnatko [00:54:10]:
Well, Google's AI cloud has a backlog of $460 billion as of like last quarter. Like that's how, that's how big. Like that. And again, that's just the backlog. That's not like actual.
Leo Laporte [00:54:21]:
You're borrowing $80 billion. So.
Andy Ihnatko [00:54:24]:
Yeah, so, so it's, it's, it always, it always came to. It's going to be weird when Apple's going to get a honeymoon period for the next year or two because a small fraction of the iPhones that are out there are going to be capable of like running everything that can cost Apple a whole lot of money. The rubber is going to meet the road in two or three years from now when most of the iPhones out there are modern enough that they are actually making a big use of Siri AI. And on top of all that, that in two or three years Siri AI, God, God willing, is going to be useful enough that people are going to be relying it on it day in, day out. And that's. I'm thinking about this in a broad sense that is Apple going to be happy having a contractor provide that widget to them indefinitely or do they have a financial strategy for Apple at this point? We are going to be able to manage this. We're not going to just simply be writing checks of however much we have to pay every single year to a provider to do this for us. We are going to.
Andy Ihnatko [00:55:31]:
This is, this is.
Christina Warren [00:55:32]:
I mean they have their iCloud. Yeah. They've never built their own data centers for iCloud.
Andy Ihnatko [00:55:36]:
Right.
Leo Laporte [00:55:37]:
Why is Apple so cagey? I mean this is. All of this speculation comes out of the simple fact that Apple's being extremely cagey about it for Instance, in their machine learning research white paper, they do talk about their core being a 3 billion parameter model. That makes sense. It's an on device model. The core advanced is an MOE with 20 billion parameters with 1 to 4 active at any given time. But then that's the last bit of information you get. Then they just say, our server side workhorse is AFM3 cloud and then there's FM3 cloud image and then there's AFM3 cloud probes, but they don't tell you anything about its capabilities. If they were a little less cagey, we wouldn't be doing all this speculation.
Leo Laporte [00:56:21]:
Is it because they think real people don't care and it's just us geeks that care? I mean, there's reason to speculate. For instance, what about safety? John, did you try any unsafe. Did you try to make an atom bomb?
John Gruber [00:56:35]:
No, I should have before I appeared on the show.
Leo Laporte [00:56:38]:
You should try some unsafe things. Get some nudes, you know, just do some, just. I'm curious. I would imagine Apple has stricter safety restrictions than Gemini.
John Gruber [00:56:48]:
Oh, I think so. I've tried. I forget what I tried. I wish I had it off the top.
Leo Laporte [00:56:52]:
Did you ask it how to get pizza toppings to stay on?
John Gruber [00:56:56]:
No, I did not ask that one either. I'll bet it gets it. I'll bet none of them are going to get that one wrong again.
Andy Ihnatko [00:57:02]:
Although someone did give it the car wash question and it did pass.
Leo Laporte [00:57:05]:
It did pass the car. It would get passed it in a weird way, but it passed it.
Andy Ihnatko [00:57:09]:
So actually in a condescending way, which I loved, which is. Well, I don't. Yeah, you could walk it, but I would think that you'd want to drive your car there if you're going to be washing your car, honestly, if your head weren't screwed on.
John Gruber [00:57:23]:
Leo, we have a great example and it happens to be between the same two companies, Apple and Google. And it's one of the benefits to Apple institutionally of the fact that their leadership is so long stayed at the company. Right. That these people don't. You know, once you become a senior leader at Apple, that tends to be where you ride out your career. So the people who are there remember when this happened. But when the iPhone launched, Steve Jobs introduced it and they're like, we have a great Maps app and with our friends at Google. Right.
John Gruber [00:57:54]:
And then how did that work out? Right, that didn't hurt.
Leo Laporte [00:57:59]:
As quickly as they could, they moved everything on. They took out as much Google off of the iPhone as they could as quickly.
John Gruber [00:58:04]:
But eventually, long story short, what happened was Google got Apple between a rock and a hard place where the maps that they had access to were bitmap Macs. They were images that when you scaled, they got pixelated and then you'd have to download smaller, more fine grained ones and the future of maps and what we all know now are vector maps like PDF that scale and the big one. And you tell people this and it's like telling people that the original iPhone didn't shoot video. And they're like yes it did. And you're like no, it had a camera but it did not even shoot video. And they're like well how did you shoot video? And you're like let me tell you
Leo Laporte [00:58:44]:
about, you had a flip cam of course.
John Gruber [00:58:49]:
But people don't remember that maps didn't have turn by turn directions. You didn't say here's where I want to go. And then your phone told you where to go, it just drew it on a map. Like you had a paper map with a pattern path mapped out. And obviously that was the future and it was the present. If you were using Google Maps on Android phones and Apple didn't have it, they didn't have a contractual right to it. And Google was asking Apple for more access to data and users location and the ability to sign into a Google account at the system level that Apple didn't want to give them. And Google thought well we've got them, you know, we'll play chicken with them because we kind of know where their own internal maps are at, which is bad.
John Gruber [00:59:33]:
So they're not going to switch to that and they'll comply with what we're asking. And instead Apple was like well we'll flip the switch and go with the maps that are not good because we got it, we've got to, we've got to make this switch. So I think, I think Apple, literally the actual leaders at the company were there when that happened, happened. They remembered, you know. So I think that they are moving ahead with a plan B that doesn't involve any help from Google and Gemini whenever this contract expires in a handful of years. So that if they need a plan B, they are already working on it.
Leo Laporte [01:00:08]:
But it feels like they're a little embarrassed to have to do this. Yeah.
John Gruber [01:00:12]:
But I also think it raises the question is why, why did Google agree to help them this way?
Leo Laporte [01:00:18]:
Yeah, what does Google get out of it? That's a really good question.
Andy Ihnatko [01:00:21]:
Well, because if they're basically saying that the largest tech company in the world needed AI services. Who did they come to? Of course they came to Google. And also the experience that they're going to get in maturing Gemini through this, it is a win, win, win. I'm sure it's not a loss leader for Gemini but it was just a win, win, win for both companies always.
Leo Laporte [01:00:43]:
I think Apple had to go with Google for a variety of reasons. Remember John Gennadrea was still there for one thing.
Andy Ihnatko [01:00:49]:
Right.
Leo Laporte [01:00:50]:
But, but also you can go to OpenAI but they don't, they don't really have the capacity. You know Google can both give you cloud capacity and a model. You go to Amazon. They have the capacity but they don't have the model.
John Gruber [01:01:02]:
Right.
Christina Warren [01:01:02]:
You'd have to go through them to go to to get Claude and I mean and I have no idea how the financials work. I imagine that, that the Google cloud division was very happy with whatever the result was for this and extending their usage.
Leo Laporte [01:01:17]:
There is an analog to the Maps story because Image Playground was so God awful, so terrible.
Andy Ihnatko [01:01:25]:
I still don't know.
Leo Laporte [01:01:26]:
And now it pretty much looks like Nano Banana. So there isn't
Andy Ihnatko [01:01:33]:
like we said last week, it's like if you're looking for Gemini again it's like did Barry Bonds take take steroids at a certain time? No. I think there's a reason why he's suddenly 300 pounds of muscle when he's smashing a baseball 800.
Leo Laporte [01:01:45]:
There was no evolution of Image Playground.
Andy Ihnatko [01:01:48]:
It just, I'm guessing that Gemini is a help as a help helpful handyman in that one.
Leo Laporte [01:01:52]:
Yeah.
John Gruber [01:01:54]:
But I think Andy, I think you're, you're right. I think Google agreed because I do think it's a feather in their cap. I'm sure that they've made sure that the finances work out in their favor. I don't think they're making a huge amount of money money because if it's even vaguely in the ballpark of a billion dollars, that's not that much money to Google especially if you're giving Apple 20. But I think the feather in the cap of hey, when Apple needed help and wanted the best models, who'd they go to? They went to Gemini. That's worth more than a billion dollars. That's worth more than the money to Google and let's face it, I think it's a way to, to kneecap OpenAI and anthropic that hey, let's not let them get this position where Apple is going to give them. They can build up this multi legged stool of revenue, actual revenue that they can Count on let's make a deal so that we get that position and they don't.
John Gruber [01:02:53]:
Because Google is the company that already has serious revenue, enormous profits that are all growing in their existing business. Why let their upstart competitors get a deal like this? And the prestige of working with Apple that this is who Apple went to
Leo Laporte [01:03:16]:
because they're a public company.
Christina Warren [01:03:17]:
But I think that goes a long way. Yeah.
Leo Laporte [01:03:19]:
Why is that?
Christina Warren [01:03:20]:
Because I think that for a lot of regulatory reasons, for a lot of other, specially when you're talking about dealing with things internationally and other stuff like I don't know, know if you want to make this sort of deal with a startup to, I mean look, let's, let's look at exactly what happened with Fable last week, right where the model is pulled. Now imagine that you were in a situation where you had based your foundation model on something that was then pulled that you're paying this money for. And the government now says that this can't happen for so many reasons that you can't even get into. I can't conceive of Google every being in a position that Anthropic was in last week and that they're still trying to kind of negotiate their way out of. I can't ever see Google being in that position. And not that I can see OpenAI being in that position as much, but Google, they're a public company. They have dedicated people who are, you know, lobbyists to the government. They have, you know, many, many, many more times employees.
Christina Warren [01:04:16]:
They've been around for, you know, almost 30 years. Like I think it's just, just if businesses, if you're going to make that sort of, you know, investment, even if you don't plan on it lasting forever, I don't know if you go with a startup. I just don't.
Leo Laporte [01:04:31]:
It does argue for Apple getting off Gemini as soon as possible though. They want to own their own destiny and they don't want to take any chances.
Andy Ihnatko [01:04:38]:
But there's, but there's a fact that ties into all of this, all the things we've talked about for the past five minutes. I mean there are a lot of companies that can make us make a satellite. There's only one or two companies that can, can give you the rocket that'll take that satellite where it needs to go. And Google is one of those two companies right now.
Leo Laporte [01:04:53]:
So John, you got a little cat fight with Mark.
John Gruber [01:04:56]:
All right, well
Leo Laporte [01:04:59]:
he, according to knowledgeable people, he said that they would have plug ins for other models. He turned out to be right. Right. Apple didn't say they sent a keynote. No, no. You don't think that's still a possibility? I thought we saw that in the course.
John Gruber [01:05:12]:
Oh, it's still a possibility, but it didn't happen at WWDC and it was as an iOS 27 feature. And the only extension that's there is ChatGPT, which has been there for two years. Well, it's been there since iOS 18 two years ago, and it is the case. And they didn't, number one, they didn't talk about that extension at all in the keynote. They didn't say anything like, hey, you can still use ChatGPT through Siri, even though you can.
Leo Laporte [01:05:44]:
That's a good point. They didn't.
John Gruber [01:05:46]:
And what Gurman had reported was that they were going to build out, you know, on that extension system, have a vetted list, presumably Claude and Gemini. And Gemini, you know, you would think would be maybe the first next partner because they've collaborated with Google with the Gemini models for Apple's foundation models. Right. So why not let Gemini in as an extension? Attention, that didn't happen. The architecture, I think that's what Gurman is still hanging his hat on, that the architecture for those plugins is still there and the ChatGPT plugin is still there, but they didn't talk about it. And the ChatGPT integration doesn't get any of the access to your Spotlight database. And when you talk to Siri AI, what happened before the new Siri AI? So for the last two years, iOS 18 and iOS 26, if you turned on the optional ChatGPT integration in Apple Intelligence and you asked Siri a difficult question, which for Siri was a lot of questions, it would automatically determine, oh, I can't answer that. I'm going to say, you know, and by default it would say, I can't answer that.
John Gruber [01:07:02]:
Would you like me to ask ChatGPT yes or no? And you'd say yes or you can turn on a preference that said automatically go to ChatGPT if you need to. And I turned that on. And every time I've used Siri over the last two years, almost everything I've asked it, it would go to ChatGPT because guess what? The built in Siri couldn't answer, needed a lot of help. That no longer happens. If you turn on the ChatGPT integration when you interact with Siri, no matter what you ask it, it will never go to ChatGPT automatically. To get ChatGPT, you have to ask, ask ChatGPT who won the Super bowl in 1969, you have to say ask ChatGPT. Or if you're using the new app, you have to go down to the bottom at the chat prompt where it says Ask Siri. And then you can tap that and it brings up a menu where you can change from Ask Siri to ask ChatGPT.
John Gruber [01:07:59]:
And then you can, and then the prompt will say ask ChatGPT and you can type your query there and it will ask ChatGPT. But it, it's almost to me like, what is the point of it being there? It doesn't integrate with any of the app intents data that we were talking about, which is what makes Siri AI different. It doesn't have access. So, and I've tested it, Apple said that it doesn't work. I've said ask ChatGPT the exact same question. What did Tola tell me the door code to his building is? And chatgpt says, I don't have access to your private data. I can't answer that. And if you ask other questions like ask ChatGPT when was the last time Ben Thompson texted me, ChatGPT will know that it can't answer and will say asking Siri.
John Gruber [01:08:51]:
And then it switches back to Siri and Siri gives you the answer. So it actually works, works the other way where for two years, if you asked Siri a question it couldn't answer and you had permitted it, it would ask ChatGPT instead. Now when you're using the Siri app, If you ask ChatGPT some questions that relate to your personal information, it'll ask Siri and switch you to Siri. That's fascinating. So what exactly is the point of these extensions at this point? If you Want to use ChatGPT, use the app. If you want, want to use Gemini, use the Gemini app. It's good. And then it integrates with the rest of the Gemini system on the web and on other devices and other platforms and all that.
John Gruber [01:09:34]:
It doesn't really make any sense to me. And my guess, and I asked off the record, I asked on the record and they were like, we're not going to talk about that. You know how Apple is. But what I asked is, is the ChatGPT integration that is still there, is that because of the contract that you're contractually obligated because it was a three year deal or maybe a four year deal or a five year deal, who knows? And is that why chatgpt a couple months ago was like, hey, we're thinking about filing a breach of contract Lawsuits with Apple. Yeah. Is that related to this? And they're like, we're not going to talk about that. But that's what my guess is. My guess is the only reason ChatGPT is even there as an option is that three years ago when they made this deal it was at least a three year contract and who knows if it's still there next year it might be four.
John Gruber [01:10:22]:
But I would say there's almost no reason to use this integration at this point.
Leo Laporte [01:10:27]:
Yeah. Do you think companies like Notion who have their own AI will still use those app intents to give Siri access to your Notion database? Or are they going to say no, no, we want to be a silo. That's going to be a big question is how many third party apps.
Christina Warren [01:10:43]:
Yeah.
Leo Laporte [01:10:44]:
Integrate that?
Christina Warren [01:10:45]:
That is going to be the question. And I don't know. I think it's probably going to be on an app by app basis and depends on how much an app like Notion maybe not because I don't know how often, I mean obviously a lot of people use it for note taking things but they've already pivoted into enterprise and kind of a, you know, I don't know if they, if their primary goal is like oh, we want to, you know, keep people creating personal notes on their phone. I'm not sure. Whereas I hope that Gmail, I mean you already could, if you added a Gmail account just in the messages app, you could still access things that way. But I would hope that Gmail and some of the other mail providers would add access to that. It's interesting when you look in the app intents developer portal like page on the developer portal the examples they give are how to use it with a streaming, with the music service, how to integrate it in with a messaging app, how to integrate it in with. There was, there were a couple of other examples but it was, it was interesting to see the examples that they chose because I think that's telling of okay, we know that there are, these are the types of providers that people will expect to get things from and this is the one part of it that whether you're all in on kind of the Google ecosystem or the Apple ecosystem or heck, I guess even the Samsung ecosystem is will you actually be able to do everything that you want to do from all of your devices or are you still going to have to rely on using multiple AIs? And unfortunately I think that the dream of just being able to have like I think at least on a phone, I think on a desktop, it's much different.
Christina Warren [01:12:18]:
But on a phone I think that we're, we're probably going to be stuck at least for a little bit of time, if not forever having to, you know, realize that we still live in data silos and, and of course there will inevitably be a phone from OpenAI or from Anthropic or someone else. Right. And, and they will then be like please give it, please connect to us all your data. That is sort of the one interesting wrinkle here which I think Apple's a little bit unique on as being an outlier is that you can already connect, you know, your Microsoft 365 account, your, your Google Docs account, your Gmail account, a lot of other third party accounts, many times your notion account to services like ChatGPT and a Claude they already have those abilities to plug in. Like that's how Claude Cowork works. Right. And, and so I think that it. What I am sort of curious about and I, I'm sure the answer is no.
Christina Warren [01:13:12]:
But like would Apple ever allow you to connect, you know, your, your Apple Mail or your iCloud not saying your messages. I'm not saying everything else. I'm not trying to get crazy to one of those services to be able
Leo Laporte [01:13:24]:
to surface those things you'll never see. I don't think.
Christina Warren [01:13:26]:
Oh, I don't think you ever will. I don't think you ever will. But, but I'm just pointing out that that is one of the things that puts them a little bit behind is that the way that all this works so great is that everything is in the Apple ecosystem but the second you step out even a little bit, it starts to fall apart. And if you're already using multiple things anyway, I do think that that lessons that's Apple.
Andy Ihnatko [01:13:47]:
But I love, I love the mechanism that John just explained. You can explain it to me further if it's a more fundamental thing. But the idea of. Remember the day when if you had a word processor all your documents could only be used by that word processor. You could not use it with not even a different word processor on the Mac, let alone on another platform. We're in the risk of having that same problem problem with all these AIs because I use, I use two or three different AIs for two or three different purposes. Gemini is my general purpose one. It's my personal assistant.
Andy Ihnatko [01:14:21]:
So it has all the subtle information about me. Claude knows about my, my programming projects and things like that. I would love it if I had, if I switched back to iPhone or if I wanted to use Siri AI on my iPad. I could make a request of Gemini. Gemini knows that, okay, I don't have access to that information. Let me ask that.
Christina Warren [01:14:41]:
I can pass it off and it can ask the password. That was cool.
Andy Ihnatko [01:14:44]:
Siri tells me that there was an imessage about this and because basically I've told, I've had them, I got the pizza and donuts in the break room, they had a meet and greet. So they trust each other. That seems like such a natural way for these things to get along with each other. Because like I suggested before, I don't begrudge Apple for saying that. Look, we're not going to let just any AI or any app have the same level of access that we're going to give our own products because we, we have, we put more security and more demands on our own software than any we would than we can ask of any developer. But the idea of having, look, I trust this Gemini person. Don't let Gemini. Don't let.
Andy Ihnatko [01:15:22]:
If Gemini wants to delete everything that's in my docs directory, don't let them. But basically, if they're going to make reasonable requests for information that only use Siri has, that's kind of a dream.
John Gruber [01:15:35]:
Yeah. And to be clear, what I was describing is how it works when you're in the Siri app and switch to ask ChatGPT in there so those things don't show up. When I go back to the ChatGPT app, which is my main app, there's no record of the interactions that were handed off to Siri. But I had the same thought as you, Andy, that this could be the future. And the way, the vague way that I could see Apple implementation, this would be like the way that they protect your photo library. Now where if you're in any third party app and you want to access a photo from your photo library, the system pops up a picker and you can say, I only want to give this app these five recent images. Check, check, check, check, check, share these. And then the app sees those images.
John Gruber [01:16:26]:
So like, like I'm imagining something like that. But for anything Siri can surface. So you're in the Gemini app and you're asking a question that would query device private information and it would say, oh, let me ask Siri. And then Siri would pop up a system picker that would be like, here's the answer. Do you want to share this with Gemini? You see the answer, it's in a system picker. And if you say no, then Gemini just didn't get it. Just like if it brings up the system photo picker and you just decide, yeah, I changed my mind, I'm not going to send it any images. You were looking at them in the picker, that gives you the permission.
John Gruber [01:17:08]:
But if you say no, then the app doesn't see any of your pictures, doesn't know which ones just came up that you were looking at. It could be the exact same with answers from Siri and you're like, you know, then you could see, oh, Siri is going to share those text messages. Oh, those are fine. Those aren't really private. Sure, share those with Gemini and then Gemini would get that information. I think that would be fabulous. And that's the sort of interoperability that really does hark back to applescript right where it's like, oh, it's not just willy nilly and an app pokes about and deconstructs the file format for a word perfect for dos, deconstructs it on its own.
Andy Ihnatko [01:17:51]:
Here's a dictionary. Open it up. Yeah, it's inside there.
John Gruber [01:17:53]:
Yep, here's a dictionary that tells you what structured information you can expect. And to go back to Leo's question from a couple minutes ago, I think Notion in particular is a really interesting example because Apple called them out in the keynote for switching to SwiftUI for their native interface that they're going to switch out some cross platform, probably web based stuff that they've been doing for their user Interface and using SwiftUI for the iPad and iPhone and maybe the Mac too. I actually don't know but it could be because SwiftUI works on them all. But Apple calling Notion out, there weren't very many. There never are many third party companies that get a call out at wwdc. Notion got a particularly high profile one. So I wouldn't be surprised if they're also saying to them, hey, while you're switching to switching SwiftUI, how about you do the app intents thing? And I think the other thing that Christina didn't mention earlier is and I don't know what the difference is, but there's also app schema and the schema is like the dictionary that defines the structure of the information you're providing. And so like the schema would be notion.
John Gruber [01:19:08]:
I think defining. Here's the dictionary that describes the a notion note and the app intent is the actual API that would be the flow of information to outside. So I wouldn't be surprised if notion does this because I also think that what people would ask of Siri AI of Their information in their notion database, whether it's personal stuff or work stuff, isn't at all duplicating the AI. That notion has built into the app itself for complex workflows and stuff like that. It's one of these ways that like everything is becoming a computer, right? Your watch is a computer. My headphones are two little computers right now in my ear. Most of the light switches in my house now are little computers. Everything is going to have AI to some degree.
John Gruber [01:20:06]:
There's going to be little bits of it. There might be one main one that you use for like, oh, when I'm like spending hours working on an AI project, like a big programming project, this is my AI system. I use codecs or whatever you use. But there will be little pieces of AI everywhere. And I think Siri isn't little, but it is sort of superficial. It's just like, let's just cover the basics and kind of fulfill the promise of Siri from 2010. I mean that's basically what Siri 2027 is.
Leo Laporte [01:20:39]:
That begs the question is will they do an agent? Because that's basically what an agent would do, is represent you to all of the other AIs and data systems out there in the world. Siri could do that.
John Gruber [01:20:53]:
I don't know if that's going back to that tech talk. They definitely poo pooed it for now. Federighi at that tech talk was like, you guys are talking about agents. And I think it was a great answer. Where it was sort of, we're doing one thing first, right? And that's just getting this stuff to work. And we think agents, you know, have a bright future but that's ahead of us.
Leo Laporte [01:21:14]:
Right?
John Gruber [01:21:14]:
And I think that's a, I think that is such a humble answer from Apple. And yeah, people who listen to MacBreak Weekly, people who are on MacBreak Weekly are all in on agents and doing these agentic things with, with AI. But most people out there aren't.
Christina Warren [01:21:32]:
Sorry, go on Andy.
Andy Ihnatko [01:21:33]:
I'll say very quickly. One of the most exciting things about Google I o about features on Android was just the ability of, oh, you want a widget to do a certain task, describe it, it will give you that widget and it will work on a phone, it'll work on a watch, it'll work on the desktop that hopefully we can entice you into buying at the end of the year. Things move forward with when the person who's buying their phones, the Verizon store and the AT and T store, you know, the people off the street when they understand how to do things just simply by describing it. Where I don't have to, where you have a tool like Gemini Sparks and I do not have to buy a Mac Mini and figure out how to run OpenClaw on it is Gemini's agent,
Leo Laporte [01:22:14]:
I might point out.
Andy Ihnatko [01:22:17]:
And so that's why, like Apple, that's why I so agree with you, John. That was such a perfect thing. They were a little bit snarky at a couple of points during the main keynote, but that, but that was exactly the right answer.
Leo Laporte [01:22:29]:
That.
Andy Ihnatko [01:22:29]:
Look, we're not there yet. We're, we're, we're, we got, we could, we can find you Boston on a map with our Siri and we're very happy that it can do.
Leo Laporte [01:22:35]:
There's a risk, though. I mean, they're still behind and all they announced really is catch up to what's happening.
Andy Ihnatko [01:22:41]:
But they still got time.
Christina Warren [01:22:42]:
They do have time, but I would say no. All I was going to say is, is when we talk about agents and I think that's a fine answer and it's a, it's a true answer. Right. You should be honest about it. We're not there yet. I think though, if you look at the third party app developers who already have in many cases a contentious relationship with Apple because Apple goes out of their way to not make things easy on them, right? Like, let's just be honest, many of them have already invested time, like Notion, for instance. That's a great point, John. I didn't know that.
Christina Warren [01:23:10]:
I didn't recall that they were even shouted out during the keynote. But that's, that's great. It's great that they put some Swift UI Things and the iOS and iPad app and yeah, maybe that'll even come to the Mac. We'll see. But like, but Notion already has MCP servers and they already have agent frameworks and they already have agents that are integrated right now. Now, if they're going to work, whether it's through a schema or app intent or whatever, if they're having to do additional work, I'm not saying they won't, but I am saying is that this is not the same universe that we were five years ago. Even where app developers are going to be jumping at their feet to jump through whatever hoops Apple wants of them to go into, you know, to have their systems integrated, you have to make it compelling to them somehow. And that's the part that I think, you know, we can look to Vision Pro as, you know, proof that just because Apple puts out a new system doesn't mean everyone is going to immediately it.
Leo Laporte [01:24:04]:
They may not come.
Christina Warren [01:24:06]:
Exactly. And so especially when there are already known, you know, frameworks and systems for how things work and like, Apple's going to make you do it just a little bit differently. Not saying companies won't do it. Many of them will, and I hope that they do. But we can't pretend like there's not an opportunity cost and there's not even real, you know, maintenance costs for those companies to support those things when you don't know. Okay, well, why am I bothering with this when I already have an agent framework and I, I can already do these things? Why? Just so somebody could. It would be better for their, for Apple's overall ecosystem, for me, for me to be able to surface, you know, my data from notion. But what's in it for notion?
Leo Laporte [01:24:41]:
Right.
John Gruber [01:24:41]:
Yeah.
Leo Laporte [01:24:41]:
I got to take a break. I wish I'm way behind. This conversation is too interesting. But we will continue on. I mean, this is very interesting. It's great to have John Gruber in the house first time on MacBreak Weekly. I don't know how we've managed to miss you for so long, but we're so glad to have you. We quote you pretty much every week from during Fireball.net Christina Warren from GitHub Andy Anatko from Anatko.com Boy, I want to get more into that energetic thing, but I think we have to move.
Andy Ihnatko [01:25:12]:
We have another three hours on that.
Leo Laporte [01:25:14]:
Yeah, yeah.
Andy Ihnatko [01:25:14]:
I mean, well, it's only getting more and more interesting, which is the fun and the tragedy of this show sometimes.
Leo Laporte [01:25:19]:
Yeah, well, and this is, you know, I'm pretty bullish about what Apple's done because it does make all this easy and straightforward for a normal user. But I'm also very aware that more and more people like Christina and me are using agents and doing a lot of stuff that Apple probably won't ever do because they cross these silo barriers. And fortunately, at least right now, people are making MCP servers, they're making APIs and SDKs, are making it possible, possible to do this so, so that we can create agents that know everything about us privately and can integrate with the outside world securely. And that's a very powerful proposition that Apple probably won't ever.
Andy Ihnatko [01:26:02]:
Yeah, well, I don't. I'll, I'll, I'll, I'll. My opinion is just that all Apple has to do is make sure that they. We don't. It doesn't wind up in a situation where when, when Google, when Google Photos first got its AI gener AI regenerative editing tools and I could say oh well I didn't this dog in the background, I didn't. What is like urinating as the tree and kind of ruining this beautiful picture of my mom and my sisters. Circle it, save, erase it erases and fills it with and every time I would show that to somebody it'd be iPhone usually say why doesn't my phone do that? And that's the sort of thing that Apple does not want to ever happen. That's the sort of thing that Google does not want to ever happen.
Andy Ihnatko [01:26:44]:
For someone to see someone, someone one of their friends or family members doing something really slick, really useful that would, they can immediately recognize this will, this will simplify my life. It will solve problems or create opportunities for me. Why doesn't my phone do that? And so, so long as Apple can stay ahead of that problem, I think they're okay. And again they're, they're, hopefully they're, they're being humble enough to realize that they didn't say that oh, we think that agents are the wrong way to go and their security nightmares and we are, we're. It's part of a reflection of how smart and how responsible we are that we're not talking about agents. I'm very, very relieved that they're saying yeah it's a great thing we're not there yet. I do think they'll get there so long as they get there at a good pace with reliable feature that people are expecting at the time when people are finally expecting to get this whenever they take the wrapper off a new
Leo Laporte [01:27:32]:
phone Apple doesn't want and this is why they're not in the eu. Interoperability, they really, they really like that ecosystem lock in. The issue with the EU is the Digital Markets act, the DMA which says things should be interoperable. And Apple is of course never going to be interrupt. We just talked about it. They're not going to interoperate Siri AI with anthropic. Yeah. Does that mean they'll never be in Europe or is EU going to give in?
Andy Ihnatko [01:28:03]:
I don't, I don't think on this, on this issue, I don't think Apple's going to give in because this is, this is. There are areas in which I do think Apple's just being dumb, okay. By not lying. A third, by not allowing a non Apple watch to have access to notifications and other simple things that are so simple and that are in full control the user saying, I trust this, I trust Garmin, I trust Pixel, I trust Samsung. Please, Apple, please let them have access to actual actionable, actionable notifications. That's Apple being dumb and being isolationist and trying to protect their monopoly or whatever. However, when it comes to something as profound as the pipeline of communication of all of your personal information through, through AIs, they, they are making a very good point that this is an area in which, yeah, it is unfair, it actually is unfair to these other companies. But you can't be.
Andy Ihnatko [01:28:53]:
This is a line they don't. I don't want to cross when it comes to privacy and I don't. I wish that they would come up with a way that, that would satisfy both the need for interoperability as well as the privacy and security of their, of the data for their users. However, in this case, I'm hoping that they're going to err on the side of privacy and security for that.
Leo Laporte [01:29:12]:
And John, you point out that losing EU is not the worst thing in the world. They're only about 7% of Apple's worldwide revenue.
John Gruber [01:29:18]:
So maybe, you know, it's for App Store. Yeah, but I think that that's probably a good measure overall because why would the App Store be lower than their overall revenue? I mean, it's possible that Europeans buy high end Mac studios that they never connect to the App Store at a higher rate than elsewhere. I mean they're very decorative, they look
Andy Ihnatko [01:29:42]:
beautiful on and off.
Leo Laporte [01:29:43]:
But you believe that Apple could leave the, leave Syria out of the E forever?
John Gruber [01:29:48]:
I do. And I think that was the gist of Apple's rather strident news on the keynote day. And as I emphasize, their public statements regarding the DMA and regulation around the world have always been rather diplomatic. It's such a perfect word for the tone they strike in public. Public, which I think is the proper tone to take. And then off the record, when they have briefings for the media about this, they don't use curse words, they don't cuss about it. But you can see them thinking about the cuss words. Off the record, there's a very full
Andy Ihnatko [01:30:33]:
square jar somewhere on Apple campus.
Leo Laporte [01:30:34]:
They do hold out hope though. That newsroom piece did say Siri delayed. They didn't say denied well, but it
John Gruber [01:30:42]:
would take the gist that I got from it and from. They had a very, I think there were a couple of instances of it, but I was in a briefing last week at Apple park the day after the keynote with a bunch of other people from the media and the gist that they laid out there is if the European Commission doesn't change its mind on this, I don't think Apple, I don't think this version of Apple Intelligence and Siri AI will ever come to the eu. I said that's a much smaller feature. But iPhone mirroring is now two years old and still isn't there.
Leo Laporte [01:31:25]:
I said it's because Apple wants to preserve their lock in. But they say it's a security issue. You could say that with if you on mirroring.
Christina Warren [01:31:32]:
I think it's both.
Leo Laporte [01:31:33]:
Yeah.
Andy Ihnatko [01:31:33]:
Can I, can I also say that Apple, although I'm on Apple's side here, Apple does have the responsibility to not just say trust us, bro. They have to bring forward. Here are the technical reasons why we will allow third parties to look at this white paper we've written. Don't trust us on this. We are not just being, we're not just trying to force people. We're not giving the answer of we'd really love your mom to buy an iPhone and replace that place their Android phone. There really are actual real reasons that we have argued inside our own, within, within our own institutions.
Leo Laporte [01:32:01]:
They went through the trouble of designing something they called the Trusted System Agent to appease the eu. And the EU according to Apple, did not agree to any of Apple's proposals. However, given the clear dangers to EU users and regulators failure to acknowledge these risks, there's currently no timeline for Siri AI's availability in the EU.
Andy Ihnatko [01:32:22]:
So you gotta admit it was cheeky for them to ask for an 18 month extension. It was like, okay, it's totally cheeky.
John Gruber [01:32:28]:
And that has gotten a lot of attention. And I think, you know, and I have people saying that, well that's clearly contrary, that's outside the law of the dma. They can't offer an extension like that. And I get it. But as I understand it, and again this is odd background briefing, I wish that they would publicize all of this, but my understanding is because at a first level Apple asked and I think they have a case that the DMA doesn't say, that this is against the dma, that they could ship this version of Apple Intelligence and Siri AI in the EU and have the European Commission deem it as not being against the interoperability requirements of the dma. There's nothing, the DMA is very vaguely written to say the least, where there's certainly nothing in there that mentions AI or LLMs in particular. And this is where Apple and Google are in complete alignment, where Google just went Ahead and shipped the integrated Gemini in Android in the eu and now they're facing an inquiry or whatever they call it, where the EU said oh, that's not allowed. So Google took the do it first and ask forgiveness tactic and Apple took the let's ask before we ship it and was told no, that would not be compliant.
John Gruber [01:33:56]:
But for example, and I asked people at Apple this and at first they had some answers, but when I asked and said what's no, explain this, they kind of started thinking and were like, yeah, I don't know which is how is the current Siri? The current Siri, the one that you get today in iOS 26, how is that compliant with the DMA? You can't replace that with a third party. There's no other option to get the side button on your iPhone to go to a different voice assistant. The only difference between old Siri and new Siri is that new Siri works, but in terms of what it tries to do, it's the exact same thing. So I don't see how old Siri is compliant with the DME either. I think Apple and Google have a good case and Apple signed. I forget what they call it in the eu, but it's like we've asked for comments and Apple submitted their own comments in support of Google in the eu that Apple is saying to the EU what Google is doing with Gemini and Android should be permitted under the dma. Not that we want you to change the DMA or add a specific extension to it or an amendment to allow this under the DMA as written law of the land in the eu. This should not violate it.
John Gruber [01:35:16]:
That's the first ask and clearly the European Commission disagreed. They're saying that Gemini's existing integration in Android is non compliant and Siri AI as announced last week at WWDC would be non compliant in the eu. Second, Apple went to them with the Trusted System Agent proposal. Not working code, but they sent engineers. You know, it wasn't just marketing people, it was engineers who had designed the outline of it, sent them to Brussels to present this six months ago, which is highly we all know, and Apple emphasized this in the briefing. Highly unusual for Apple to tell anybody anywhere in the world what they're doing six months ahead of time. But they went and said, here's our plans for AI and Siri. Here is a proposal A, we think it should be compliant as is, but if it's not, here is our next plan which would be to build this Trusted System Agent.
John Gruber [01:36:15]:
Unfortunately, I don't the briefing did not include any kind of technical description of what that actually is or what it would be like. But apparently it's pretty, pretty big proposal. And Apple estimated that it would take 18 months to build, but wanted the European Commission to say if you built something like that, it would probably be compliant or it would be compliant and then Apple would go ahead and commission the engineering work to build it. And what they, the 18 month period that they were asking is, and if you allow the Trusted System Agent and how about you give us an extension and let us ship what we have now while we're building it? That was like the last thing on the list. And I don't think Apple really thought they were going to say yes to that. I think Apple was hoping they would either just say yes, new Siri is compliant just like old Siri was, except it works. B if they said no to that, that they would say, okay, this Trusted System Agent sounds like a good plan. If you could build something.
John Gruber [01:37:17]:
And here's our suggestions to the outline. Go ahead and build it and however long it takes, you know, hurry up. But you know, when that's ready, you can, you can ship in the, in the eu. What happened is the EU said to Apple with that proposal, we don't judge proposals. Go build it and then come to us with code and we'll tell you. And with this one, Apple is saying it is such a major project what off the record, you know, or on background Apple, Apple described it as, in their estimation would be more engineering work on Apple's side than all of the other things they've built for DMA compliance to date combined. The browser engine kit, the alternate App Store support, the alternate payment support, all the other stuff that they've already built for DMA compliance. This project would be more than all of that combined.
John Gruber [01:38:10]:
And they don't want, want to do that and then find out after they've done it that it still isn't compliant. That's why what they wanted was for the EU to say build that and it'll be compliant and they wouldn't do it. And so that's where things stand where app. That's why Apple can't even say, we think 18 months from now maybe we could have something because they're not even sure that they're going to go ahead and build it because.
Andy Ihnatko [01:38:34]:
Okay, that's good to know.
Leo Laporte [01:38:36]:
Why build it. Yeah. So this is, it's interesting because as you say, the DMA is so kind of vaguely written that it's up to the regulators to interpret it Apple calls the EU regulators interpretation extreme. They say Apple would have to give any virtual assistant direct access to users private data and the ability to directly control other installed applications. And I agree with Apple, that is a non starter. That's clearly an issue. I wouldn't want Apple to do that either whether I was in the EU or here that, that no third party AI should have access to all of that information.
Andy Ihnatko [01:39:15]:
And this is where I'm doubly sorry that Apple plays that card so frequently that oh well, we can't, we can't allow, we can't allow. We can't allow third party, third party developers to link outside to their own websites because that would be a compromise of app of customer privacy and security. And we will absolutely, we're going to take a stand against that.
Leo Laporte [01:39:37]:
Yeah, they've overpowered that bridge.
Andy Ihnatko [01:39:39]:
And so when they, when it actually, when it actually is a valid argument. That's why I say okay, well now's the time where you. Here's the.
Leo Laporte [01:39:45]:
It is finally in this case it's completely. It's completely. And I'm a big interoperability fan, I'm big open systems fan. I But in this case I completely understand why Apple doesn't want to do that.
Andy Ihnatko [01:39:56]:
I wouldn't say both things are true. It's unfair to, it's unfair to third parties and it's necessary for security sometimes. Both of those things are absolutely true.
Leo Laporte [01:40:03]:
Yeah. Let's move on to some other topics besides Siri AI. There's a little battle going on. I'm wondering if John has any inside information about whether the folding phone that has not yet been announced will be shipped this year or later. Some, some rumors say it will have to be in 2027. One rumor mill adamantly says no, no, it's coming out this year. What do you know about this, Jo John? Anything?
John Gruber [01:40:30]:
I know nothing. Other people I've gotten out of that game of trying to figure out what hardware's coming.
Leo Laporte [01:40:38]:
I don't do it either.
John Gruber [01:40:39]:
Yeah, but just so not talking to anybody at Apple but just reading the rumors, it's the people who seem to get it right. I think it's about as close to a lock as possible that it's coming. And, and I think it's just the sheer scale of iPhones and even iPhones that aren't popular compared to the others. Like let's say this year the iPhone air, which clearly was not a smash hit. They've sold so many iPhone airs that I don't think there was any way to keep its Existence under wraps a year ago at this time. And I think again, let's just assume and I think it's probably a fair assumption that this iPhone that folds, which I think they'll probably call iPhone ultra, but who knows. The folding iPhone is probably going to be the most expensive iPhone because the folding mechanisms, displays are always among or the most expensive components. Folding screens are all more expensive than non folding screens and Apple.
Andy Ihnatko [01:41:50]:
Can I say a sidebar? Motorola, which is the king of budget of incredibly good phones that are incredibly cheap, they had, they introduced their first folding phone and it was $100 cheaper than a Samsung foldable. That's how expensive that component is that even Motorola cannot track fifteen hundred dollars.
John Gruber [01:42:08]:
Right. Whose brand is sort of a high quality phone at a lower price. So let's say it's super expensive. It's still, you know, it doesn't gonna sell in high quantities. Maybe they don't think they're gonna make many. It's gonna be hard to get for Christmas, whatever. The rumors are so strong and the clues at WWDC were like, hey, you can, you can make extra widescreen apps in the iPhone mirroring app. You know, just in case, you know, if you're wondering what these apps do,
Andy Ihnatko [01:42:37]:
you want to run the simulator and see what what your iPhone app would look like if it were suddenly and unexpectedly made wider.
Leo Laporte [01:42:44]:
They spent some time in the Xcode session talking about responsive displays as well. It's pretty clear where it's headed. They're not, I'm amazed they have been able to say it with a straight face without any little winks or nudges, but it's pretty clear what's gonna happen. Good. Well, I hope that that's the case. Apparently Samsung is actually starting to manufacture the touchscreens for the new MacBook Ultra, which implies that it might even be available sooner than later, which would be good news.
Andy Ihnatko [01:43:15]:
Yeah, I think there was two weeks ago that there was some supplier, supply chain reporter said that Samsung was targeting a 90% yield rate and they achieved that. And so that meant that it was viable for this contract to go forward or something like that.
Leo Laporte [01:43:27]:
Yeah, yeah, it's.
Andy Ihnatko [01:43:28]:
And, and also people who are using. I don't know if you've used this, John, but who's using Continuity on the iPad? I never, I can never get these individual features. But basically Sidecar, Sidecar, Sidecar, Sidecar, Sidecar. Sidecar now responds to touch on the iPad, which I, which it might be significant. It might not because I thought that why wouldn't it do that from there in 2000 you could buy like a touchscreen on that you could add on so that you could actually touch as a USB mouse pointer. But yeah, that's. There's a lot of snow on the ground and a lot of footprints in that snow.
Leo Laporte [01:44:06]:
Good. All right, that's pretty much it on the rumors as far as I can tell here. So maybe we should do the long awaited much wanted Vision Pro segment.
John Gruber [01:44:18]:
Here we go.
Christina Warren [01:44:19]:
What do you see? What do you know? It's time to talk about Sean.
Leo Laporte [01:44:22]:
You don't have to do it. We wouldn't have Vision Pro segment if it weren't for Andy and Ako who found the one Vision Pro story today.
Andy Ihnatko [01:44:33]:
Apple pig with a truffle baby.
Leo Laporte [01:44:36]:
You snurfled it out below that tree over there. Apple helped Vision Pro helped Disney re engineer a classic Epcot ride. Apparently Apple Disney Epcot ride. Which one is that? What's your best? Sorin Soin is fun.
Christina Warren [01:44:51]:
Soren Soarin is a great ride. That's what they use it for. Soren is actually, actually freaking excellent.
Leo Laporte [01:44:55]:
So that's the one where you sit in these seats and they bring. It's like hang gliding. It was originally over California, but I think Soarin now is everywhere.
Andy Ihnatko [01:45:03]:
Soarin across America.
Leo Laporte [01:45:05]:
Yeah.
Andy Ihnatko [01:45:06]:
Anniversary of what? The United States.
Leo Laporte [01:45:08]:
And they have Putty as your pilot and you go flying. So that's good. Are we gonna get Soarin on the Vision Pro? Is the question.
Christina Warren [01:45:17]:
See that, that's what I've always thought. Like that would be a killer feature of the Vision Pro, to be honest. Would either be to like, like the Avatar ride at. At Animal Kingdom or. Or Soarin. Like these are the sorts of things where when I first used the Vision Pro, I went, okay, I don't want one of these things to sit on my face all the time. But this would be fantastic. Either yeah on a ride or to recreate a ride.
Christina Warren [01:45:40]:
Like I could see in both scenarios, like if you could get it durable enough, have them actually, you know, have individuals be able to use them on the ride itself because the screens are much better than what you're typically getting on those experiences. Or yeah, like let's recreate Soarin at home like that.
Andy Ihnatko [01:45:55]:
Or even just the right content. Just. Just yesterday some I was watching the.
Christina Warren [01:46:00]:
The.
Andy Ihnatko [01:46:01]:
The Simpsons motion ride. We're basically the Simpsons. The Simpsons crew basically wrote an entire episode of the Simpsons, starting with the videos you watch while you're waiting in line. And then when you get into the 3D environment, like you're. You're in the car. It's a continuation with all these jokes that are going through. And it's like, it's free. It's already there.
Andy Ihnatko [01:46:21]:
Just again, don't, don't, don't make. I don't have to, like, have the whole thriller. I want to watch when I was watching this, where suddenly, like, hey, right through my mouth. I want to see. I want to see what that would be like to ride that car through there.
Leo Laporte [01:46:34]:
It's funny because that one made me nauseous. But then I went across the hall to the flying Broomsticks at the Harry Potter thing. And I love that both of them, you put on a helmet. But the Simpsons are an older one. You're sitting in a little 3D.
Christina Warren [01:46:50]:
Yeah. The Simpsons one is like 20 years old now at this point, which is really, really.
Leo Laporte [01:46:54]:
You know, it's not really as bad as you think, because they say you can put your backpack on the shelf in the cart here and it will be safe. But on the, on the, on the Hogwarts ride, you were sitting on a saddle and you're moving around. It's. It's really, it's much. And I did not get nauseated.
Andy Ihnatko [01:47:10]:
I just think it's a shame when you have all this content that was built so beautifully, so seriously sometimes by Francis Ford Coppola, by George Lucas, by the Muppet, by Jim Henson. And the thing is, like, if they decide to shut down the ride, one of the in jokes in the Simpsons ride is that it replaced the Back to the Future. And the opening scene is basically how Dr. Frink is responding, responsible for the destruction. And it's like, the thing is, like, it's. I think that anything that's been created with such care, it shouldn't be allowed to die. If there's a way to preserve it, I agree. And if there's a way to simply take that stuff and say, look, five bucks.
Andy Ihnatko [01:47:47]:
Or just, you know what? We're looking the other way just like that. Just. Just like Apple does with a whole bunch of like, old, old, old roms. Like, you know what? We're not going to litigate this. We can't give you permission. But we're. We. We are not opposed to people running.
Andy Ihnatko [01:47:59]:
Apple works on another system. And I hope that turns out it
John Gruber [01:48:04]:
really is the sort of thing those. These rides are preservable in an immersive way that like, just. All right, we'll send a guy through with a camera is not. And the Simpsons one, I've been on that many Times it is. It is oddly immersive for a thing that just kind of dips you up and down a little. But it's like a little. But for the most part, though, it really would transfer very well. You would get no motion, but you would get something that a flat, rectangular version wouldn't.
Leo Laporte [01:48:36]:
Yeah, yeah. Are you a Vision Pro fan, John? I don't know.
John Gruber [01:48:41]:
I am in an odd place where I'm a booster of the future prospects of the platform and I barely use the one that I have over here. Like to a degree that every. Every single time I want to use it, the battery is empty. And then I think I'm going to get another plug. I'm going to put. I have a whole socket of. Or a whole wall of outlets over here to this side of the desk. And I'm going to have a dedicated charger just for my Vision Pro.
John Gruber [01:49:13]:
And then eventually I'll have some other gadget that need to charge and I'll be like, oh, I'll use that one I'm using for the Vision Pro. And then my Vision Pro expires. So, no, but I will say this too. The other thing I do, and I've done it, I think, three years in a row is my WWDC live talk show. We live stream.
Leo Laporte [01:49:30]:
Oh, yeah, with
John Gruber [01:49:33]:
sandwich. And then we have a higher quality. It might be coming out later today, actually, where you can go to the sandwich theater app and watch it in 3D. And I really thought like three years ago when we first did it, I was like, well, this is like one of the first events live streamed immersively by next year. I didn't think that this product was going to be a hit. I was like, but there will be more events, including probably from Apple, that are live streamed for the people who do have one. Nope, still just me.
Andy Ihnatko [01:50:05]:
Yeah. What surprised me is that they didn't. I know they can't necessarily launch a YouTube for immersive experiences, but I thought that, look, you've got this thing with all these cameras on it. I'm sure that people will want to make their own 3D experiences, even if it's just that, hey, look, here's what was like at Ozfest this year. And I've got 10 minutes worth of video and I'll be able to post it and share it with somebody. That would have been exciting. It wouldn't have justified $3,500, but $408 of that. Yeah, that might have justified it if it had taken off.
Leo Laporte [01:50:33]:
You do get a lot of credit for Doing the talk show in a Vision Pro with Adam Lysigore. I think that, that, that for two and a half hours.
John Gruber [01:50:42]:
Yes.
Leo Laporte [01:50:42]:
That's above and beyond.
John Gruber [01:50:45]:
That was the longest podcast I've ever recorded. Maybe not by the clock.
Christina Warren [01:50:49]:
If I saw. If I felt it.
John Gruber [01:50:51]:
Yeah.
Andy Ihnatko [01:50:52]:
How long until the lines around your forehead?
Leo Laporte [01:50:54]:
When I was gonna say like a whole.
John Gruber [01:50:56]:
I think like last week. I don't know. It was a very long time.
Leo Laporte [01:51:00]:
You can watch that on YouTube as well as this week's talk. The talk show. And then if you have a video Vision Pro, you need the sandwich TV app. Right. To watch the talk show in your sandwich theater. Just look for theater. That's it. Yep.
Leo Laporte [01:51:16]:
And that. Well, we got some extra content in there. Is your Vision Pro.
Christina Warren [01:51:21]:
Now, you know, we're done talking the Vision Pro.
Leo Laporte [01:51:27]:
You're watching MacBreak Weekly with. It's wonderful to have John Gruber, the legendary John Gruber from daringfireball.net, Andy Ihnatko and Christina Warren. A couple of other stories. The UK government, I thought I actually was very impressed with Apple's announcement of their child protection features. As you mentioned, the parental. What do they call it? Parental features, parental controls. Which I didn't know, but you told me or you mentioned on the talk show. Are actually Apple's developers thinking about their parents?
John Gruber [01:52:03]:
Yes, originally
Leo Laporte [01:52:06]:
this gives it a whole new meaning. But I was very impressed. I was especially impressed with the defaults. I thought that that was exactly the right way to take parental controls and to I think maybe end bypass and around the governmental restrictions that are coming not just in Australia and in many US states, but now, now in the uk. I don't know if it's going to be enough. The UK has announced they're going to ban social media for under 16s. Actually, Nilay said he didn't think this was going to be enough to keep the government out of the iPhone business. I really liked what Apple did and I think it's much better to let the parents control it than governments.
Andy Ihnatko [01:52:54]:
Yeah, 100%. I can't imagine how. Of all the things, the fact that YouTube's included inside that ban, you're basically imagine us at in the mid-80s and saying, by the way, NBC and ABC networks do not exist. You will not be able to access either of those two networks because the amount of content that kids get through YouTube is like, I don't know how they're going to enforce that. That has to be a first step towards. Oh, and by the way, VPNs are also banned. Banned within the UK I was gonna
Christina Warren [01:53:22]:
say you don't enforce this because any, any the kids are going to figure out ways around it. And I mean, especially for things like YouTube, I mean, maybe, okay, you try to make it so onerous that, that nobody, you know, the companies have to put in completely, in my opinion, draconian tests to try to find out users age, which are completely defeating the point of any sort of privacy thing. And now these companies are collecting information that you really don't want them to have because it's not like they have secure servers or that they don't resell that data or anything else. I don't know. I mean, I think that all this stuff is just deeply, deeply depressing.
Leo Laporte [01:53:59]:
Yeah. Keir Starmer announced it'll happen as soon as next year. Early next year will ban Snapchat, TikTok, YouTube, Instagram, Facebook and X. Messaging apps like WhatsApp and Signal will not be banned. And romantic companion chatbots will be banned. If you're under 18, you can't keep us apart.
Andy Ihnatko [01:54:22]:
Me and Sheila are in love. It's fake to you, but it's real
Leo Laporte [01:54:25]:
to us as Will. And I don't know what this means. I'll have to ask a Brit. Intimate functionalities will also be banned for those under 18. Yeah, this is, is that domain available?
Andy Ihnatko [01:54:39]:
Intimate functionality?
Leo Laporte [01:54:40]:
Intimate functionalities. No way. This is why the birth rate is dropping in Britain, I just want to say. So Apple obviously would like to get ahead of this. Is what they're doing enough? Probably not, huh, Neil? I believe that governments are just going to go right ahead and do it.
John Gruber [01:55:03]:
I keep turning back to the comparison. I think I said this on stage too, but to the government's approach to end to end encryption where they'll say they don't understand it technically and they have a good use case where okay, what if there are two pedophiles exchanging csam? Why can't you give law enforcement just the good guys a backdoor to their end to end encryption? And, and people explain to them that defeats the definition of end to end encryption. There's no way to have a backdoor just for the good guys. You can't do it. And what the government bureaucrats who don't understand it here is what they hear I think is that's a lot of work that you don't want to do. It's hard. We'll just mandate it and then you can use your nerd power and make it happen.
Leo Laporte [01:56:02]:
Yeah, we call that the nerd harder.
John Gruber [01:56:04]:
Yes, exactly. And I really do Think. And with end to end encryption, it's a little bit simpler to understand because you understand there's an end to end encrypted chat between me and Andy is just between us. And if we add Christina and we have a group chat, we understand that these communications are only going between our three devices. And you can kind of understand that. And adding a third, another party, law enforcement or just bad guys would be breaking the circle of trust. Parental controls is a little more nebulous because some of these things are like, oh, I just don't want my kids spending time playing Roblox or something.
Leo Laporte [01:56:46]:
But Apple has the functionality, the ability. Apple knows, Google knows how old you are, they have the functionality built in. Apple has the age intent where people can say, what age group is this user in? And it would all be implemented through the gatekeepers.
John Gruber [01:57:01]:
So some of this is possible, some of it's possible, but I think some of it comes down to the bureaucrats and the politicians wanting to say we'll just make them nerd harder and have parents be able to say, oh great, thanks to this law that makes Apple and Google and Meta and everybody else nerd harder. Now my kid isn't doing anything inappropriate or spending time in places I don't want them spending time on their devices. And that's magical thinking and it's just not going to happen. But I think some of these controls do seem better and I think some of them. The knock is that a lot of what Apple announced last week isn't actually new. It's actually been there. But like Joanna said on stage with me, like it just didn't work. I mean, and so it's like, okay, the features that we've had now, they actually work.
John Gruber [01:57:49]:
So if that's the difference, maybe that makes, you know, it just doesn't absolve parents though from needing to take some significant amount of responsibility for what their kids are doing on devices.
Leo Laporte [01:58:04]:
But that's who should be taking responsibility in my opinion. Right?
John Gruber [01:58:08]:
And I think you can mandate that, you know, you could say, hey, if you set up your kid's device as a child device, then they have to get permission from the parent to say use a vpn. And if the parent says no, I don't even know what a VPN is. No, no. And that's it.
Leo Laporte [01:58:23]:
I like the functionality where the kid would say, can I go to this site? And the parent gets a pop up on their phone. That's new, isn't it? That seems like a feature. Yeah.
John Gruber [01:58:32]:
And if it's not, it was so buried that nobody could ever. It never happened.
Christina Warren [01:58:36]:
Right. Which was the issue before. I think the secondary issue was that you had had third party companies who were trying to build these better tools before, like what Apple introduced, which I agree with Leo, I think is great, that existed and then they were very limited in how much.
Leo Laporte [01:58:50]:
Well, there were privacy issues. There were also.
Christina Warren [01:58:53]:
Which makes total sense. Right, but, but, but like this MDM and I mean, well, you would have to do that. Yeah, but like there were a number of different, you know, companies who were trying to do those sorts of things and you couldn't really do that unless you, you know.
Leo Laporte [01:59:04]:
No, Apple should do it, Google should do it. Yeah, that's, that's who should do it. And you know, people say, well the kid will figure out a way around it. Oh yeah, the kids aren't going to figure out a way around a social media ban in the uk. Yeah, the kids will figure out a way around it, sure.
John Gruber [01:59:18]:
Yeah. It's just like with alcohol and tobacco sales. Right. Does a 21 year old age limit mean that nobody in any of the United States or United Kingdom or any of these countries is drinking alcohol or smoking cigarettes before they turn 21? No. But do the laws greatly decrease how, how often that happens? Yes. Well, that's a win. And if you're waiting for perfection, no kids accessing anything inappropriate or spending more time than their parents think, doing X, Y or Z, you're going to be waiting forever.
Leo Laporte [01:59:51]:
Yeah.
Christina Warren [01:59:51]:
I think the difference though with, with like alcohol laws for instance, versus what they're doing in the UK is that there's not a database held every time I go to the look of store and buy a beer and then I have to give them my ID and now it's stored on their server someplace. But now because I'm an adult, if I'm in the uk, I now have to verify and give my ID to somebody who I don't trust at all. I don't trust anyone in the, in the UK government. I don't trust any one of those companies to have a copy of my identification on their servers. I sure don't. So thanks a lot guys. You've now created like an even worse honeypot for everybody who is of age to get hacked and have their identity stolen. Like congratulations.
Leo Laporte [02:00:30]:
My fear with this fable thing is that at some point you're going to have to proof of citizenship to use the top AI models. That's not going to be a good thing. That is not going to be a good thing. I'll tell you what is a good thing for Apple tv? Fox, before we move on. Oh, sorry.
Andy Ihnatko [02:00:48]:
I have an update I did check. Intimate Functionalities.com is not available because I just bought it.
Leo Laporte [02:00:54]:
Congratulations.
Christina Warren [02:00:55]:
Congratulations, Andy.
Leo Laporte [02:00:57]:
Fantastic. Did you get Net as well? Because you know, you're not a commercial enterprise. Okay, very good. Fox has purchased Roku, the streaming service, for $25 billion.
Christina Warren [02:01:12]:
Wow.
Leo Laporte [02:01:13]:
I immediately saw a number of people asking on Reddit, what's the next best streaming device? Actually they were asking on our Twitter forums at Twitter community, what else should I get?
Christina Warren [02:01:25]:
Apple tv.
Leo Laporte [02:01:26]:
Apple tv. There really is only one. Roku was the low price leader though. In fact, they.
Christina Warren [02:01:31]:
Yeah, they had more than Fire Stick.
Leo Laporte [02:01:34]:
Yeah, yeah. I think they said Roku has 100 million global households. That's.
Christina Warren [02:01:38]:
Yeah, they have more.
Andy Ihnatko [02:01:39]:
But they bought the channel as well. Didn't they
Leo Laporte [02:01:43]:
get the whole thing?
Christina Warren [02:01:47]:
This is just gonna be. Look, on Twitter we talked about this and I think then the rumor was Paramount and I was like, that will be the worst thing ever. This is slightly better than Paramount, but only slightly. Like now it's just gonna be the Tubi tv. Right? Like, congratulations.
Leo Laporte [02:02:01]:
Yep. Fox plans to keep Tubi and the Roku channel separate. Fox also hopes to cut $400 million from the annual costs of Roku. So if you work for Roku, you know what that means.
Andy Ihnatko [02:02:15]:
I hope that doesn't mean a drop in the quality that I expect from roku.com free streaming.
Leo Laporte [02:02:21]:
Yeah.
Andy Ihnatko [02:02:23]:
Love Island, Canada. If I don't get that, the season
Leo Laporte [02:02:26]:
eight of that Roku has a 25% market share according to the Wall Street Journal. That makes it number one. Number two, Samsung's Tizen.
Christina Warren [02:02:35]:
Yeah, that makes sense.
Leo Laporte [02:02:36]:
All those Samsung TVs. 23% of the streaming market. Who stre. You know, I have a lot of Samsung TVs. I never want to use their streaming interface. Oh Lord.
Christina Warren [02:02:45]:
There are a lot of people who don't know. Well, my dad is one of those people. I've bought him so many. He had a Roku, he's had a Fire stick. He's had multiple Apple TVs. He uses the freaking Comcast box. He won't even use the built in webg. He has a very, very nice LG TV.
Christina Warren [02:03:04]:
I would much rather him use the WebOS app apps that are built into that than the ones on the Xfinity box. But no, that's what he uses and I can't do anything. I bought him two Apple TVs. He has them in two rooms. I can't get anywhere.
Leo Laporte [02:03:17]:
So it's Too Complicated, Christina. It's Too Complicated by Waterloo, by the way, just purchased Intermittent Functionalities. Edu. Now there's an opportunity. There's a business opportunity. I think that's good.
Andy Ihnatko [02:03:30]:
I welcome the competition. And I will crush you.
Leo Laporte [02:03:33]:
I will crush, crush you. I will crush you. All right. I think, I mean, there's a lot more we could talk about, but I think we have pretty well.
Andy Ihnatko [02:03:42]:
One, one last thing just to mention. Congratulations, Apple. They. They won the Tony Award for Schmigadoon, which means that they completed their egot.
Leo Laporte [02:03:50]:
Apples and egots.
Christina Warren [02:03:52]:
Congratulations, Apple. Thanks for canceling TV show and winning the Tony. That's great.
John Gruber [02:03:57]:
And it definitely seemed like the hardest one for Apple to pick up up. Right?
Leo Laporte [02:04:01]:
I mean, the Tony Award. Yeah, yeah.
Andy Ihnatko [02:04:03]:
If you're a stand up comedian, you can do a one man show. I mean, you got options. But for Apple. So it's gonna be.
Leo Laporte [02:04:08]:
Let me get this clear. So they had a TV show called Schminga Doon. They canceled it. It went to Broadway.
Andy Ihnatko [02:04:13]:
Yeah.
Christina Warren [02:04:14]:
Yes.
Leo Laporte [02:04:14]:
And Apple produced the Broadway version.
Christina Warren [02:04:17]:
No, but, no, they owned the, they
Andy Ihnatko [02:04:18]:
owned the rights, but they, but they're listed as one of the producers because that's the way the Tonys work. There's like 800 producers like attached to a thing. So.
Christina Warren [02:04:25]:
Yeah, they were not on stage. They were not on stage.
Leo Laporte [02:04:28]:
In fact, John Turnis didn't jump up.
Christina Warren [02:04:30]:
No, no, no, no. In fact, one of the producers purposely called out and said, thank you, Apple, for canceling us.
Leo Laporte [02:04:35]:
Yes, I did see that.
Andy Ihnatko [02:04:37]:
They weren't, they weren't as. They weren't ashamed enough. They weren't so ashamed of the show that they didn't like actually put something on the Apple newsroom like proudly saying, hey, Apple's won a Tony Award. I don't blame them. If I won a Tony Award under any circumstances whatsoever, I would be pretending as though that was a wonderful thing. Well, because it's cool. It's cool. Object.
Leo Laporte [02:04:56]:
There was one more story. The rumor is now that the new iPhone will have at least the Pro will have 12 gigs of RAM for Apple. AI. That actually is a story that is going to develop is what you can run the new Siri on, what you can run the higher end stuff on which phones work and so forth. Yeah, 12 gigs means probably much more expensive, I would imagine. Yeah.
Andy Ihnatko [02:05:22]:
After the keynote I was going through, I do not have a single important computer, including my phone and my watch that is not four or five years old. And ready for an update because I've been waiting for a really Good reason update. And so it's like as my decision making this year because I'll definitely be buying a new MacBook, like in the next month or two, another new desktop.
Leo Laporte [02:05:45]:
Wait a minute. You're not going to wait for the M6 with the touch and the.
Andy Ihnatko [02:05:48]:
I'm okay. I do not have MacBook Ultra money. I do have, like, I do have M5 money. I do, I do have. Wait until October just in case they pull up a surprise money. But the thing, whereas usually, like, I have a metric that I've followed like for the past 10 years, this year it is like, what is the most amount of system RAM that I can afford and still get storage? That will work for me. Because that's going to me, in my mind, that's going to be the difference between getting another good five years out of this and getting three and a half years in which you get everything. And then a year and a half where it's like, I really wish I could get that agentic stuff working.
Andy Ihnatko [02:06:25]:
But unfortunately it requires 30 at least
Leo Laporte [02:06:27]:
24 functionalities.com money starts to roll in, you might have more options.
Andy Ihnatko [02:06:32]:
You know what? I am going to be that optimistic. M5 Pro Ultra.
John Gruber [02:06:38]:
I think it's fun. I do think it's funny, Leo, that for years and years when people, you know, because Apple doesn't like to talk about RAM and iPhones, they don't like to tell you. And it's like, then it's like. And every, every year when they come out, it's like somebody will ask in the press and they'll say, oh, you know, we don't talk about that. And then every year they'll say, like, but, you know, as soon as we get it, we'll be able to know. So why don't you just tell us? And they're like, well, you go ahead and do that, and then we go ahead and do that. And every year it's, you know, not as much RAM as people who watch shows like this were hoping, hoping for. And for years it was always, come on, look at Apple's margins.
John Gruber [02:07:13]:
They could easily put four more gigs of RAM or however many more they could have. And, and now this year, the rumor is they're going to increase it. And everybody's like, how are they going to do that? These phones must cost $3,000, right? It's totally flipped from, come on, why are you being so stingy with ram? RAM makes everything better. And now it's like, oh, my God, 12 gigs of RAM in a phone. I don't Believe it.
Christina Warren [02:07:41]:
Well, it's funny because I think the OnePlus CEO said that now the cost of RAM is more than half the cost of bill of materials for phones, which is ridiculous, right? So it's absurd and it's, we don't know when this will abide. Now my only question is I feel like Apple, probably from everything I've read they had their brand prices locked in annually, right? They don't do it quarter to quarter or month to month like other suppliers do, but that had to have gone away. So I hope that they locked in like a year ago at the lower rates because otherwise I think I am worried that we are all going to be in for a little bit of shell shock when new phones come out.
John Gruber [02:08:21]:
I don't think they're going to raise prices because I think they see the prices as so inextricably part of the brand. So I don't think so. So I think that at least for the phone. But who knows, maybe this explains why there don't have M5, Mac Minis and Mac Studios out yet. You know that they're like, well we've got this much RAM and it's all going in.
Christina Warren [02:08:40]:
Well. Yes, well, maybe, maybe. Let me be more specific. I think that you're probably right in terms of like the base iPhone and maybe like the iPhone Pro. But I feel like this gives them an opportunity to have an excuse to have the Ultra and the MacBook Ultra priced even higher than they would be otherwise because they're going to need to make up the margin on the higher volume products.
Leo Laporte [02:09:00]:
That's what Ultra means. Ultra expensive means absolutely for what it's worth.
Andy Ihnatko [02:09:06]:
Take the word of analysts for whatever it's worth, which is not rumor but not necessarily gospel truth. But I think a month ago there was a report from a couple of different analysts who are saying that Apple is pretty. They're not going to be unaffected by the ramp situation. However, they feel as though Apple has enough cover that they are going to pursue a policy of we can put more pressure on particularly Chinese premium smartphones by keeping the prices exactly the same, even if they have to eat a little bit of margin to do it. In the long run it'll be much, much better for their share.
Leo Laporte [02:09:37]:
I think that makes sense and I
John Gruber [02:09:38]:
also think, and again it's a whole separate discussion, but I think that the line on Apple services is like a ruler draws it right. It is like there's no seasonal up and down and I think they're confident that they're going to keep keep Making more on services gives them the ability to, okay, we'll take a little less margin on the phones because of this RAM thing and we'll be fine because we're not. Our overall margin isn't even going to dip because we're going to keep increasing it on the services.
Leo Laporte [02:10:09]:
Do you think they'll put AI, additional AI costs into services beyond iCloud Plus?
John Gruber [02:10:15]:
No, no. And certainly not anytime soon. I think right now they're really just, please trust us. It's worth trying. Use it. You know, they don't want to ask you for money. I think integrating any kind of paid tier into the existing iCloud plus tiers is obviously the way to go. I think Apple is a long way off from charging even a dollar a month more for, for Siri.
John Gruber [02:10:40]:
I mean, it's. They have so much credibility to earn back and I hope, I honestly hope their humility. They showed last week that maybe they make some commercials that are funny, you know, like, hey, Siri is actually good now.
Leo Laporte [02:10:54]:
You know, kind of admit it a little bit. Yeah, yeah.
John Gruber [02:10:58]:
And I think that could actually get people to be like, oh, maybe I should try it. If they're, if even they're cracking jokes
Leo Laporte [02:11:03]:
about it, maybe they get John Hodgeman play old Siri.
Andy Ihnatko [02:11:06]:
You know, they're on the back foot where they didn't do it by saying, our users love Siri and we're happy to say we've made it even better.
Leo Laporte [02:11:16]:
Yes.
Andy Ihnatko [02:11:18]:
You know what? I don't care if I drew the short stick. I'm not saying that even on.
John Gruber [02:11:22]:
Right. That's a macro button in pages when they write the script of the keynote. Our users love this. Now they're going to love this more. I think that they were like, Andy.
Christina Warren [02:11:32]:
Yeah.
John Gruber [02:11:32]:
They were like, how do we write this? I don't know. And they're like, we better call in the a team of writers.
Leo Laporte [02:11:39]:
Well, speaking of a teams, it's great to have John Gruber on the show. Wonderful to finally meet you. And of course, Staring Fireball is a must read. We talk about your pieces all the time. Really good journalism and we really appreciate it. Christina Warren also here from GitHub, where she is. Did you mourn Fable? Did you get to play with Fable at all?
Christina Warren [02:12:00]:
I did get to play with Fable a little bit and it was great. It was great for the two and a half days that I got to use it. Rip.
Leo Laporte [02:12:07]:
I'm bitter. I'm bitter. I actually started a major project.
Andy Ihnatko [02:12:12]:
Oh, no.
Leo Laporte [02:12:14]:
Well, well, you know, I wasn't going to pay the token cost for it. So I had till June 22 to use it on my max subscription. So I thought, Well, I got 10 days. I am going to rewrite our entire twit ad sales system in those 10 days. And I got pretty far on it. They generated a questionnaire for the stakeholders and it was understood the schema, it understood the old software. It had everything. And then they pulled the plug.
Leo Laporte [02:12:41]:
It was right in the middle of. It said, this model's no longer available. And I was. I thought it was me. I thought, did my credit card not go through? Anyway, you still have the good copilot, Christina. You're good, you're golden. You're fine.
Christina Warren [02:12:55]:
I still have copilot with all the, you know, I have all the other anthropic models and all the OpenAI models and Gemini 3.1 Pro. But, yeah, I mean, Fable was. Fable was really fun for, like.
Leo Laporte [02:13:06]:
It was pretty good.
Christina Warren [02:13:07]:
Like, like, like the 60 hours we had it or whatever it was. It was. It was awesome.
Leo Laporte [02:13:11]:
Yeah. Yeah.
Andy Ihnatko [02:13:12]:
It's. It's like Crash Davis and Bull Durham talking about the two days he was. Two weeks he was in the major leagues. Just, you know what all the hotels are room service. You hit white balls during batting practice. You don't carry your own bags either.
Leo Laporte [02:13:26]:
How's the new website going, Andy? In Akko I H N a T K O dot com.
Andy Ihnatko [02:13:31]:
It's going terrific. I want to thank everybody who's been signing up and visiting. I will say that I basically had the idea that, look, if the audience is this big by the end of the summer, I will be very, very happy. And it was just like one week for that audience. And so I'm very, very grateful. So much so that it's like, you know what? I think I'll cancel what I was supposed to be doing today and tomorrow and make those extra writing days because I'm very, very excited. Again, thank you very, very much. I was not assuming that people would be interested in and me like, doing the blog again for real.
Leo Laporte [02:14:05]:
And that'd be all.
Andy Ihnatko [02:14:06]:
Very sure.
Leo Laporte [02:14:06]:
Yeah.
Christina Warren [02:14:07]:
Yeah.
Leo Laporte [02:14:07]:
Very happy to see that. I H N a T K o. You're watching this week in Mac MacBreak Weekly. And we're so glad you're here, you club members. We appreciate your support.
Andy Ihnatko [02:14:23]:
We're all the grandchildren and great grandchildren of immigrants who would rise for. Who had come to us in our sleep and throttle as we took. We turned down $20 million for any reason after all, they did.
Leo Laporte [02:14:34]:
That's true.
Andy Ihnatko [02:14:34]:
In the, in the, in the Italian Alps, which is what my grandfather used to do.
Leo Laporte [02:14:38]:
A very good point. A very. They would watch. They would want us to take the money, wouldn't they?
Andy Ihnatko [02:14:42]:
Yes.
John Gruber [02:14:42]:
Yes.
Leo Laporte [02:14:43]:
Okay. Thank you, Andy, for that. I get. You get to start the picks of the week.
Andy Ihnatko [02:14:48]:
Okay, well, this is. This is a really, really fun feature in a really, really good app, Google Earth, you all know about it, it's basically the digital globe that you get for free. And the app version of it, the desktop version of it, it has a flight simulator. So you can actually use all this wonderful data and fly through cities and fly from Bangor, Maine to Fenway park or whatever. They've just added it to the web version of the app as well. And it's not like these super intense flight simulators where you will of course crash for the first three months. You even try to do anything because everything is scientific and aerodynamic. It is more like the original, like Microsoft flight simulator of like 1.2.
Leo Laporte [02:15:31]:
How do I launch it? How do I get it going?
Andy Ihnatko [02:15:33]:
Go to wherever you want to start off from. Go under tools and you will see a new option for Flight Simulator.
Leo Laporte [02:15:41]:
Oh, yeah.
Andy Ihnatko [02:15:42]:
And instead of a complicated flight yoke, it's just the arrow keys to control, like up, down and your rudders. And then there's a.
Leo Laporte [02:15:50]:
This is probably not.
Andy Ihnatko [02:15:52]:
The mouse is.
Leo Laporte [02:15:53]:
I think I just did a barrel roll.
Andy Ihnatko [02:15:54]:
The mouse is. If you try to use it with a trackpad of the mouse, I find it like, inoperable. There is actually, actually, again, to keep the tone of the show, I had to ask Gemini, are there keyboard controls for this? So, yeah, page up, page down for increase and decrease thrust, up arrow and down arrow for pitch, up pitch down, left arrow for roll, left, right, out for Robank. And yeah, it's fun. Especially now in my part, my little coastal colonial fishing village. It's rather flat, it's rather simulated 3D. However, once you go to a good city, it's like, you know what? I'm going to fly all the way to Fenway Park. And I don't think it'll let you land because again, it's not a simulator.
Andy Ihnatko [02:16:37]:
It's a simulator in the weakest sense, in the sense that it's fun. It's more like an arcade simulator. But it is. Is freaking wild, man, to just basically be controlling, not controlling like a point of view flyover with control with on screen, heads up display and just saying, here's what it would be like to basically buzz Boston harbor and here's what it would be like to try to.
Leo Laporte [02:16:59]:
And unfortunately, I'm not good at this.
Andy Ihnatko [02:17:01]:
Exactly.
Leo Laporte [02:17:02]:
Help me.
Andy Ihnatko [02:17:03]:
But I actually was getting better and better at it because unfortunately I can't across this at like 10pm, which is
Leo Laporte [02:17:10]:
like, I'm gonna auger into Doylestown right now. Wait a minute. Watch out. Here we go. Sorry.
John Gruber [02:17:17]:
It's a shame Spirit Airlines went under Leo because I think you've got spirit pilot skills.
Andy Ihnatko [02:17:25]:
The big problem is that it only supports one. It only simulates one jet and it's a Boeing, so.
Leo Laporte [02:17:30]:
Oh, good. That's okay. As long as it's not a 737 Max, I'm okay.
Christina Warren [02:17:34]:
Exactly.
Leo Laporte [02:17:35]:
Flight Simulator. A new reason to go to Google Earth, which I actually haven't launched in a long, long time. You know, they have AI in it now.
Andy Ihnatko [02:17:43]:
It really is one of the most fun apps ever. It really is like when you, when you went to your grandparents house and they had like the big atlas and a whole bunch of box of maps and you should be able to just imagine where you're going and say, oh my God, the terrain in like Peru is so weird. I spent so much time just basically browsing Google Earth and it's just so much fun.
Christina Warren [02:18:03]:
Fun.
Leo Laporte [02:18:03]:
It's fun. Yeah. Let's go to our guest John Gruber for his pick of the week. John?
John Gruber [02:18:09]:
My pick of the week is from my friend Adam Lissegore at Sandwich, but it's his new app that he's made with AI, which I think is super fascinating because he's not a developer. It's called Hovercraft and the URL is. It's Sandwich Vision Hovercraft. Just search the web for Sandwich Hovercraft. But basically what it is, it's an app for your Mac that creates a virtual camera. So you still need a real camera, but when you're in an app like Zoom or FaceTime or any other app like that, you can pick Hovercraft as your camera. And then on screen you can show slides. You just drag an image.
John Gruber [02:18:51]:
If you just have a photo, just drag it into the Hovercraft window on your Mac and then it becomes a rectangle in your square. And then you can use Vision Pro, like gestures to like move it around. And Adam is doing amazing things really quick. He added support for I wrote down the acronym because I always get the letters wrong. USDZ, which is the open standard for 3D objects, you can drag a USDZ file into hovercraft and then in your camera you get the 3G object and you can use your hands like a visual pro. To like twist it around and move it back and forth. Really, really cool. You can finger paint on the screen like Picasso.
John Gruber [02:19:35]:
There's a mode where you can make a drawing on screen all really, really cool and fun. If you don't want to use your hands on the camera, you can just use your mouse or your trackpad or something like that. But the hand stuff is very fun. It's 19 bucks for one Mac. 29 bucks gets you a license. One time fee, not a subscription. 29 bucks gets you a license. License for two Macs.
John Gruber [02:19:54]:
Go to Sandwich Vision Hovercraft. Very, very cool app. And it's one of those things it may or may not. It's like I feel like Apple should Sherlock it because it's like why don't all the cameras let you drag images in and drag them around the screen?
Leo Laporte [02:20:10]:
Yep, very nice hovercraft. I'm installing it. Soon you will see things floating over my shoulder. Christina, your pick of the week.
Christina Warren [02:20:22]:
First of all, I love hovercraft. I'm looking at as somebody who has to present a lot on screen. I'm going to use this in some of my future meetings. My pick is as parachute apps or parachute Backup.
Leo Laporte [02:20:34]:
There is a certain commonality to all the picks here. We got Flight Simulator, we got Hovercraft. Now you're jumping out of the plane.
Christina Warren [02:20:41]:
This one's great. It's recently purchased by a new owner. The same guys who bought Kaleidoscope, which is the best diff tool on the Mac, they bought this, which is a backup iCloud, an iPhoto backup app for Mac and for iOS and it's a one time purchase. And I back up a lot of things to iCloud and I just kind of trust it. But that's not really a great strategy to be honest because things do get corrupted, things do get broken. You don't know where things are going to go. And iCloud and iPhoto is a as a black hole, which is great, that's by design, but it's also annoying if you're somebody who would like to have reliable backups. And so this is basically just a utility that'll basically let you back up your iCloud documents or your photo libraries, including your shared photo libraries, directly to a NAS or another folder or another hard drive or whatever the case may be.
Christina Warren [02:21:34]:
There are a few different modes how you can like set it up. Basically you can have it mirror and download like the full version each time. An incremental backup, which would be more similar to how something like Time Machine works or have it be in sync so that it's exactly what you have in one is the same as what's on the other. It doesn't do two way sync. So if you delete something on your local drive, it's not going to delete it on your iPad side, which is nice. But it's well designed and it's a one time purchase, not a subscription. And so yeah, parachute.
Leo Laporte [02:22:09]:
Very nice. Some excellent apps. We appreciate the picks and we appreciate you. John Gruber, thank you for joining us this week. You didn't have to. Thank you for having your helmet to be on the show. So this shouldn't have been a horrible two and a half hours anyway. Yeah, I don't need to tell everybody daringfireball.net but maybe I do need to tell you.com also works.
John Gruber [02:22:31]:
Yes.
Leo Laporte [02:22:32]:
Great to have you John. Thank you for joining us.
John Gruber [02:22:34]:
Oh, it was a lot of fun.
Leo Laporte [02:22:36]:
I'm not going to say fly eagles, fly, but you know, I will climb up a greased flagpole for you anytime soon.
John Gruber [02:22:41]:
Yes, exactly.
Leo Laporte [02:22:44]:
Christina. Thank you so much. You'll find her@GitHub.com Andy Ihnatko's new website is I H N A T K. Oh, we do Mac break weekly every Tuesday, 11am Pacific, 2pm Eastern. That's 1800 UTC. You can watch Just in the Club, Twit Discord, of course, but also on YouTube, Twitch, X.com, Facebook, LinkedIn and Kick. We stream on all of those after the fact on demand versions of the show available at the website Twit TV MBW. There's a YouTube channel dedicated to MacBreak Weekly.
Leo Laporte [02:23:14]:
Great way to share clips with friends and family, help spread the word about MacBreak Weekly. And of course best way to get it is to subscribe in your favorite podcast app. That way you'll get it automatically the minute we're done. Thanks for being here everybody. We'll see you next week. And now it is my sad and solemn duty to tell you to get back to work because break time is over. Bye Bye.