0:00:00Daniel, it's good to hear your voice.
0:00:01Good to see you.
0:00:02Good to hear yours.
0:00:03Great to be back on the show.
0:00:04You were born in California, right?
0:00:07You're you're a native native San Jose.
0:00:09Not San Jose.
0:00:10Never say it.
0:00:11Never say it.
0:00:11How dare you, sir?
0:00:15No, it's what I'm drawn.
0:00:17I'm Santa Cruz.
0:00:19I actually was born in Southern California and my parents were
0:00:23row roaming van life hippies for awhile.
0:00:27and they eventually settled in Santa Cruz, California.
0:00:31So I grew up in that town.
0:00:33- At this point in your life though,
0:00:35do you think you're used to the East Coast weather?
0:00:39- Oh yeah, yeah, I'm good at the weather.
0:00:41I am so used to the East Coast weather
0:00:44that I put a lot of East Coasters to shame
0:00:47because for, it's just like the Californian, right?
0:00:51To come into town and be like, listen,
0:00:53I'm doing the East Coast better than you.
0:00:54But for instance, I run every other day.
0:00:58You know, it's like my regimen.
0:00:59And I get a lot of interesting looks
0:01:03when it's zero degrees blizzard.
0:01:06And I'm out there with like two pair of long underwear
0:01:10and a ball of clava.
0:01:12Tell you what, I never learned that in California.
0:01:17- Our mutual friend and your relative neighbor, Paul Kavasis,
0:01:21is a year long runner as well.
0:01:23And I've, I have seen pictures of him, selfies where there's like frost on his
0:01:29beard.
0:01:29Oh yeah.
0:01:30I've had like the icicles coming off my eyebrows.
0:01:33It's like, look like a cold, what's the cold miser or whatever.
0:01:36But yeah, it's, it's kind of fun.
0:01:38It's kind of fun when, when it's so severe that people kind of look at you like
0:01:43you're crazy.
0:01:43I guess I've never really, that's the Californian in me.
0:01:46Not, not shying away from people thinking you're a little cuckoo.
0:01:49I brought it up with Katke too.
0:01:51Cause I don't know.
0:01:52somehow I've lived my whole life other than two years where I lived up there your way in
0:01:57Massachusetts when I was working at bare bones. I've lived in the Philadelphia area, greater area
0:02:02from birth till now. So I'm used to East Coast weather. But the older I get, the less used to it
0:02:09I get. Yeah, I don't know. It's on my mind more you I would have thought by the time I hit 50.
0:02:14I'd be used to it. And it's like you did like my dad doesn't seem to talk about the weather. It's
0:02:19He's been through 84 years and he's used to it.
0:02:23Whereas me, it's like when it gets dark now in the fall, I'm just shocked by it.
0:02:28I'm like, I can't believe it.
0:02:30And now in the spring, when it starts getting nicer and we've got all this wonderful daylight,
0:02:34I'm like overjoyed.
0:02:35I'll tell you what, it's one thing I really appreciate about the East Coast compared to
0:02:41California.
0:02:42It's the very thing that a lot of people hate about it is the severity of the variation
0:02:47because you don't get that kind of variation typically in California, especially the part
0:02:53Northern California where I grew up. But, oh, so one of the first things I noticed moving out here
0:02:58was like, you really know when a season changes. And it's like the seasons change in California,
0:03:05but it's like a 10 degree difference from one extreme to the other.
0:03:11Right. Well, and like on the highway, whichever way you want to drive between the city and the
0:03:18valley up there, you know what season it is by whether the mountains are burned out or green.
0:03:23Well, that's true. Well, yeah, well, that's true. Although I will add, I'll hasten to add that
0:03:28typically they're always burned out. Like, typically, it's not seasonal. It's just like,
0:03:34there's never, it's kind of funny that grass, particularly the grass, it's like, designed to
0:03:41stay like yellow and it's like only when I right now as everyone knows probably
0:03:46California has had this but very unprecedented amount of rain California
0:03:51is probably just gleaming green right now and it's so unusual but anyway the
0:03:57thing I like coming out here with the with the variation and I keep noticing
0:04:00it every year is I forget how much I like this is this is lucky as a
0:04:05transplant to New England, it's lucky that I legitimately like every season.
0:04:11And that's good because I notice every time a new season comes around I'm like,
0:04:17"Oh right, I'm tired of it being fall. I want it to get cold." And then
0:04:22a couple months of that I'm like, "Damn it, it's too cold. I'm ready for some
0:04:26spring." And I guess like you were saying, if your dad just sort of
0:04:32got used to it and he's not really doesn't really notice or care or whatever. I'm noticing it always
0:04:38constantly. But I'm also sort of like always surprised by how it makes me feel like every
0:04:46time fall comes around. I'm like, Oh, yeah, this is what falls like. Well, my dad notices I should
0:04:52say because he's outdoors. One of me he walks the golf course still at age 84. And so like,
0:04:57when it warms up, and it's good to play golf, he definitely know, but he just doesn't complain
0:05:00about it is my take. Whereas I and never became like a snowbird who him and him and my mom don't
0:05:06go to Florida or something like that. Right. Which is what I'm thinking. Where I'm going is I think I
0:05:10got to find a way to go somewhere. Probably not Florida, the way Florida is going, but somewhere
0:05:14warm by the time I'm old, because I can't take it anymore. But I want to move back by March,
0:05:19because I want to be here for for when it gets nicer. I'm, I'm just in a good mood is all I'm
0:05:25saying. That's good. Yes. Well, I think the trick is if you don't want to go to Florida, I think the
0:05:30trick that from the East Coast what people seem to have figured out are the islands the various
0:05:35like kind of I don't know I don't know a lot about those cultures but there's so many Puerto Rico
0:05:42right yeah the U.S. right yeah we vacationed there last year it's really just gorgeous just
0:05:48beautiful beautiful place yeah maybe that might be the answer speaking of good news I don't know
0:05:52is it good news we can start with some news today the day we're recording Wednesday Apple announced
0:05:58the and speaking of Northern California, WWDC 2023, I would I would say, the most predictable
0:06:06date on the calendar, June 5, tonight, right? Otherwise, if it wasn't June 5, tonight,
0:06:12that almost certainly would have been the next week. Yeah, well, particularly now that,
0:06:16as you noted, it's become more and more well, I don't know, you noted that last year did turn out
0:06:23to be the template for mostly online with the in-person event. But the fact that the in-person
0:06:30event is at Apple HQ, Apple Park, some may call it, that means they don't, like we used to all
0:06:39just guess because we were so concerned about how to get hotel rooms and all that. We used to guess,
0:06:43but we knew there were constraints like what other conferences are already booked and what hotels
0:06:50already have full reservations. But now for a company that does like to repeat the same
0:06:58playbook again and again, like why would they ever pick a different date if they control
0:07:03everything about it? Well, you mean a different location? Well, I mean, like, well, no date,
0:07:08because like, we used to guess at like, well, right, right. Second week, third week of June,
0:07:13and like, sometimes you gotcha. We didn't know how much of it might be like internal problems,
0:07:18like they wanted to finish something up or, but we mostly I think, assumed that it was like, well,
0:07:23there's going to be one of these two weeks. And turns out, like Oracle's having a big conference.
0:07:27But now there's nothing like that to impose anything on them as far as I know.
0:07:33Right. Right. Both both me like as a public pundit writing a daring fireball and our whole
0:07:39crew community, I was I would even expand it to of people who wanted to go we used to play all these
0:07:45games, looking at the Moscone schedule and seeing, "Oh, well, here's the American Association of
0:07:52Bicycle Mechanics is having a thing at Moscone West the second week of June. So that's out. It's
0:07:58got to be the first week, right?" And there'd sometimes be like private event or something like
0:08:03totally generic, but no name for what it was on the Moscone West calendar. And we would just
0:08:10think that's got to be it. But you know that Apple always I mean, one of the reasons they
0:08:14every other major developer conference or major tech conference I know of or conference period,
0:08:21like whether it is the American Association of bicycle mechanics or whatever. They always
0:08:27that they pick their dates years in advance, right? It's like I remember Mac World Expo,
0:08:32it's like as you'd be leaving, there'd be they'd be handing out flyers, right? And they'd have
0:08:37sign up forms. You could sign up for next year's Mac world expo. They'd be like, see you next year
0:08:42and here's January 5th or whatever the date was. And they'd be trying to get you to buy tickets and
0:08:48sign your, when I was at bare bones and we had a booth that they definitely were offering steep
0:08:54discounts or significant discounts to commit to the next year already. Apple's is, WWDC is the
0:09:01only conference, major conference I've ever known of that announces like two and a half months out.
0:09:06sometimes closer, right? And it's because Apple is Apple and they want to, you know, just in case,
0:09:13you never know what if something's delayed. There was that one year way, way back where they had it
0:09:19in August instead of June because something wasn't ready. Yeah, it's hard to even imagine that now it
0:09:26has become so predictable. Right. But it's funny that you mentioned like there would be like a
0:09:30reservation on the conference center for an unnamed guest. And it's like every other company
0:09:37or organization in the world practically, unless it was going to be like a CIA meetup,
0:09:42would be like, "Why wouldn't we want to hide the fact that we are a vibrant company that can have
0:09:47public shows?" It's like another opportunity to get your name out there.
0:09:52Right. Predictable public shows, right? For any other company, not scheduling it until weeks in
0:10:00advance would make it seem like you were flying by the skinnier pants or what's the what's the phrase
0:10:05CD your pants CD your pants all right i guess so yeah not the skinnish skinnish
0:10:09skin of your teeth which doesn't get any of your teeth all right what a terrible mixed metaphor i
0:10:15don't i don't i guess if you have leather pants though you might have skin out of your pants but
0:10:19otherwise i don't know i don't know about you i don't i don't i'm i'm low on leather pants at the
0:10:25at the moment. Yeah, only chaps here. I do think though, and I hate to complain,
0:10:31I think last year's was great. It felt like last year's WWDC, after the keynote was over,
0:10:40even before the State of the Union, it already felt like, yeah, this is the way that they're
0:10:45going to keep doing it. It was, I don't know if they're going to be able to put more people
0:10:49on the lawn. It is intriguing as an event, what's the word, organizationally, that Apple Park was
0:10:58never designed with this, right? The Steve Jobs Theater was designed to host the type of things
0:11:04that they do host there, but the lawn was not meant to host 1500 people for this. And I've
0:11:11taught, I talked to people who were there and they were like, yeah, definitely not that it was just
0:11:15sort of serendipity and the advantage of the fact that they bought so much square acreage of that
0:11:26area to build the ring building and have it surrounded by so much green space that they
0:11:31have the room to do that. But it wasn't meant for it by design or planning, yet attending it
0:11:40in that area felt very smooth. You would have thought that it was part of the original spec
0:11:46for the campus. Like, maybe we'll do this. Well, that's the nice thing about open spaces
0:11:50is you can use them for that kind of, I mean, the main challenge, as I understand it, was
0:11:56the projection of video, which apparently they did a brilliant job of. Yeah, it was. It really was.
0:12:01It was unbelievable. But for the rest of it, it's just kind of like an opportunity to mill about,
0:12:07right? Milling about. That's what open spaces are good for. And when I think back to,
0:12:15I think you know this, but for probably the past 10 years or so, even when WWDC was happening in
0:12:22person, myself along with other friends of ours, many of us didn't buy tickets for the conference.
0:12:30and the one aspect of the conference I missed was the Apple Bash, which is kind of what they're best
0:12:40suited now to host. So I don't know, it makes sense, yes, it wasn't designed for this purpose,
0:12:49and especially for seeding a bunch of people to watch a keynote event, but I hope that they
0:12:55lean into the advantage they have of it being suitable for that big milling around space,
0:13:01because that's a really nice event. That's the part WWDC that sort of invites everybody to just
0:13:08kick back and all meet in one place, not have any pretense of an educational or classroom environment.
0:13:17That's my idea of a good time.
0:13:20It is weird and it does sort of—and again, there were hints last year, not just because as an
0:13:26attendee in the media and talking to other media people and talking to the people who I met. There
0:13:33are people listening to the show right now who I shook hands with and took selfies with, because
0:13:38they got the lucky tickets in the special event draw to be there. And I look forward to that
0:13:42again, too. But everybody who thought, Yeah, this is really great. I also did talk to people from
0:13:50Apple. And there were, again, their Apple people and nobody, even someone in a position to actually
0:13:56state it authoritatively, yes, we're going to do that this way every year. They at least
0:14:01gave a like a look on their face that was like, Yeah, this Yeah, we're really happy with the way
0:14:08this has gone and it's almost certainly the way it's going to go. And I know the other thing that
0:14:13they did say, honest, I'll just say it, even just talking to Phil Schiller, who runs events,
0:14:18they're over the moon, happy with the post COVID online delivery of the developer sessions.
0:14:28There was no question that they're going to pre-tape those probably starting like now,
0:14:33right? I'm guessing now is the time when everybody inside Apple is submitting the session ideas,
0:14:39but that the Apple is so much happier with the way that they turn out and so getting so much better
0:14:48feedback from the, I guess, millions, right? Millions of developers around the world who
0:14:53watch them that it's just, for lack of a better word, pedagogically a vastly superior format.
0:15:00the presenters who are all practicing Apple engineers and designers and not, therefore,
0:15:08not professional speakers, no longer have to deal with the, what do you want to call it? I mean,
0:15:13stage fright. I mean, stage fright, it's fear. It's just the natural anxiety that almost almost
0:15:20everybody has of public speaking, whether you actually are further down the stage fright
0:15:25spectrum or further up on the comfortable spectrum. It still is something that people
0:15:30who work full time at Apple doing real engineering and design work just don't practice and therefore
0:15:35aren't that great at even though over the years, the many, many decades of WWDC with in person
0:15:41presentations, they were always very good because they rehearsed right and they had professional
0:15:45coaching from absolute experts to help them. So the coaching and the practice and rehearsals,
0:15:52it was always very good, but it's just not their game, right? And so being able to do it in the
0:15:57comfort of without an audience and being able to do multiple takes and not having to worry about
0:16:04this is the one live one where I better not screw it up or my demo better not go wrong. All that
0:16:09pressure is gone. And the other huge advantage of it, in my opinion, and you must have a stronger
0:16:15opinion of it because you consume more of the sessions as a working developer, it lets all
0:16:21sessions be the length they need to be. Some of the sessions are 15 minutes because it's 15
0:16:25minutes of material, whereas the old way, if it was 15 minutes of material, you had to pad it out
0:16:30to, I guess, a full half hour or an hour. I forget if they were all hour-long slots or what, but...
0:16:35Jared: I think they were at least a half hour.
0:16:36Pete: Right.
0:16:36Jared. I found that extremely tedious, especially as somebody who just doesn't
0:16:42particularly enjoy classroom environments anyway, in the best of cases. So, yeah, when you got the
0:16:49sense that, "Oh, boy, this is just being dragged out to like, it's just the worst." But yeah, the
0:16:54The other thing is, all those advantages you said, and another one that occurs to me is,
0:17:00when you said they might be working on them right now, not only that, they might be finishing
0:17:05them right now, which is something you would never do in the past.
0:17:10You wouldn't have somebody preparing and practicing and getting to the point of being able to
0:17:16present it now because they might forget it by the time the actual conference comes around.
0:17:21You know what I mean?
0:17:22It was kind of like everybody had to rehearse towards a specific deadline and now they can
0:17:27stagger it.
0:17:28I bet that's really nice for...
0:17:30I remember now 20 years ago when I last worked at Apple, I remember it was like a season
0:17:36of WWDC prep where...
0:17:39I never gave a talk at WWDC, but coworkers of mine who were doing them, it's like the
0:17:45season would come around and suddenly they were just like off limits.
0:17:49They're busy.
0:17:50They're doing stuff.
0:17:51to the people who help them present. So that must be nice too, because it's not like I imagine it's
0:17:57not so much like a no sleep till WWDC type of culture anymore. Yeah, I guess I don't think so.
0:18:04So I think it's all for the better. I will I can't let an episode of the show go by without
0:18:10complaining about something though. And I will say I'll go back to my only complaint about this new
0:18:15format. And I really do think this is it going. This isn't just year two. It's not just Oh,
0:18:21the tail end of COVID and COVID concerns, and they still want to do it this way to have it outdoors.
0:18:28I mean, even though that is an advantage, right? Other than the unlikely but not impossible chance
0:18:36that Monday the 5th is a rainy day, which I really—and I remember asking last year,
0:18:41what was Plan B if it rained? And Plan B if it rained was, well, everybody's going to be wet.
0:18:47I mean, there is no way to put a tarp over 1000 or 1500 or however many people there were. The big
0:18:56problem with the location is hotels, because Cupertino has very few hotels. The one I stayed in,
0:19:04which I picked because of its proximity, because I knew I was going to be there for a couple days
0:19:08because the media stuff tends to span two days. I was doing my live show at Apple. I think we
0:19:16we did it on Wednesday last year or I forget if it was Wednesday or Tuesday. Might have
0:19:23been Tuesday, but whatever. I knew I'd be there at least through that and I knew I'd
0:19:28be going back and forth from my hotel to Apple Park multiple times. So why don't I stay,
0:19:34oh, here's, I think it was, I don't want to throw a hotel chain under the bus, but
0:19:41national hotel chain is literally right across the street. But when you're looking at Apple Maps or
0:19:49Google Maps or any of these maps, you see the ring, the rings there on the map, and from a
0:19:54global perspective, from Philadelphia, a continental perspective, boy does that hotel look close to
0:20:01Apple Park. But the truth is it's all the way around Apple Park from the end to the visitor
0:20:08center. So it's like a mile and a half probably. I think it is
0:20:14honest to God. And so I am a walker. I mean, I'm as I record
0:20:18with you today is from a slack group. I'm in my my car. My car
0:20:23hasn't been able to start for weeks because we went weeks
0:20:26without starting it in the winter. And I still haven't
0:20:29needed to drive anywhere. So I'm waiting to get the battery jump
0:20:33started until I actually know I need the car. So that I mean,
0:20:36is how we have we own one car haven't driven it in two months and literally doesn't even start at
0:20:41the moment. I like to walk I really do I that's the most exercise I get is is walking throughout
0:20:48my daily life but walking a mile and a half to get to the keynote and then anything I might want to
0:20:54go back to my hotel I ended up taking Ubers and lifts all the time. And it was not a great hotel.
0:21:00And so once you're getting an Uber or a lift anyway, it's like five minutes more if you just
0:21:05expand the realm five minutes, there's way more hotel options, but none of them are great. You
0:21:11know what I mean? It's like, some, some people I think was Marco there last year. I think he might
0:21:16have been I know. Yeah, Marco was there. I think he had one of those Mark famous Marco, Tim Cook,
0:21:22close encounters. Yes, yes, exactly. Right, right, right. With the blue with the with the the the
0:21:29that blue, what do they call it? What's the name of the MacBook color midnight midnight?
0:21:34Right. Marco had been waiting and waiting and waiting and waiting for hands-on time with the one
0:21:40midnight MacBook that was on display. And by the time it was his turn, somebody tapped up on his
0:21:46shoulder and said, "Tim's going to take a photo here, so you need to back away." And it happens
0:21:50to him every year. Right. And now I remember, because I had press briefings while the State
0:21:58of the Union was going on, and Marco and our, again, mutual friend of the show, David_Smith,
0:22:04I ran into them at the cafe at the visitor center and they were kind enough to give me
0:22:09the Marco and underscore summary of the State of the Union, which unsurprisingly was GPT
0:22:16quality as a summary, as we'll get to later in the show. Very high quality summary of
0:22:23the show. Yeah, they were there. I think they stayed at, or at least Marco maybe, I know
0:22:27I had some friends who stayed all the way back in San Jose at the whatever the name,
0:22:33new name of the Fairmont is? Oh yeah, it's like a Hilton or something I think. Yeah,
0:22:39they renamed it, redid the lobby completely because I was there, I think I stayed there
0:22:44then for the iPhone event later in the year because I knew not to stay in Cupertino. I
0:22:49don't know. It's like that's the one thing, we used to bitch and moan so much when it
0:22:53was in San Francisco at Moscone about the hotel prices, but it's like we didn't know
0:22:59how good we had it where we could bitch and moan about the price of hotels, which again
0:23:03is laughable compared to the hotel prices now. Although maybe with San Francisco being
0:23:07dead, as everybody says, maybe San Francisco hotels are cheaper. But I have no reason to
0:23:12go to San Francisco anymore. But anyway, I didn't realize how lucky we had it. Every
0:23:18single hotel was what, a block and a half away from Moscone?
0:23:21Steven: Yeah, it was really perfect. You had every kind of hotel too from pretty high,
0:23:27just pretty high end, really high end, I would never consider going to, down to pretty high
0:23:34end, down to run of the mill, down to total rat trap. You could stay in a turn of the
0:23:43century 1905, survived the great fire hotel with a shared bathroom, and then if you couldn't
0:23:51afford that, you could go stay at the hostel. It's like they had everything.
0:23:55Yeah, everything. But it's so close. And then so you just didn't have to worry like, "Hey,
0:24:00do I have everything I need in my bag or my pockets?" I would just leave my laptop. I
0:24:05just leave with a notebook and know that if I needed my laptop, I was a block away or
0:24:10two blocks away. Now, when it's miles and miles and an Uber ride away. And it's, again,
0:24:16it's fun to go and I'm so glad to have been back after COVID last year. But it's like
0:24:20to go back. It's like you have to go wait in a special line for the ride shares. And
0:24:26it makes sense, right? They can't just have Ubers and Lifts pulling up higgledy-piggledy
0:24:31on the sidewalk all around Apple Park. So, there's like a dedicated go over there to
0:24:36Building 11 or something like that. And in the parking lot, there's a queue for the
0:24:40ride shares, but it's not great. And it doesn't make sense to rent a car because
0:24:46the parking situation, they really don't, they don't have parking for everybody. They
0:24:50don't, they did not build the Apple park visitor center with the idea that they'd be again,
0:24:54that's the one side effect of not designing it with this. They don't have parking for
0:24:581100 people or however many people show up.
0:25:01I really liked it. One thing I really liked about the one year I went to San Jose in the
0:25:07modern era, San Jose WWDC, because I had skipped it maybe sort of out of like snobby protest
0:25:13for a couple years because I was like, "This isn't San Francisco." But then I was like,
0:25:17"Oh, everyone's there. I better go." And I really did enjoy. So I stayed at a hotel.
0:25:22I remember it was me and Craig Hockenberry were both at the hotel about a mile away from
0:25:28the scene. But it was that time, I don't know if they still have them. I know they sort
0:25:34of had their time and they've lost popularity, but it's that time when the little rental
0:25:39scooters were everywhere.
0:25:40like oh yeah yeah yeah no what were they called not lift what were they these were
0:25:44a lot i lie my limes lime yeah they might still be there i don't know but that was fun yeah
0:25:50well so for instance if you had lime scooters outside your hotel around the corner from apple
0:25:56park it would have made everything a lot easier probably you would just be like whatever i'll zip
0:26:00over there and sit back but yeah it's a problem it's not that whole area i mean i know that area
0:26:06so well because I worked at infinite loop for seven years, I know that area better than
0:26:15most areas. And it's a really, really boring area that has some pretty good restaurants. That's how
0:26:23I should summarize it. And so that's at least something. It actually has pretty good, I mean,
0:26:28you have to travel a little bit, but it has, in its way, weirdly, that area probably has better
0:26:35restaurant service than downtown San Jose, which is kind of weird.
0:26:40Well, and I would say everything has been I know,
0:26:43this is a lot. I'm just throwing you a lot ball here. Right. But it is it's I guess what you're
0:26:50saying no San Jose as a city of its size has an unusually sparse downtown restaurant scene. And
0:26:57the sort of coupe greater Cupertino driving area like five minutes from Apple Park has surprisingly
0:27:04good restaurants. And I agree with that. And I just always, because I have friends who work at
0:27:09Apple, just defer to them. And they'll be like, "Oh, we should go here." And I'm like, "You tell
0:27:13me." And then every place that, you know, whether it's like a stupid mom and pop sandwich place for
0:27:19lunch or a pretty nice sit down place for at night, it is, there's tons of good places to eat.
0:27:25I will say this, coinciding before we segue, I feel like it's this sort of,
0:27:31It relates to my aforementioned talk about the season change, right? When you have seasons,
0:27:36to me, it internalizes the annual schedule. And I do think, I think it's safe to say
0:27:45that you don't have to be a scientist to state with authority that humans evolved
0:27:53to sort of grok the count, the annual calendar, right? And it's like, oh, it's the holiday season,
0:28:01it's Thanksgiving, and then after Thanksgiving, all of a sudden it's Christmas and New Year's,
0:28:05and it's dark, and it feels cold, but you don't even know it's not cold yet because it's not even
0:28:10really winter, and then springtime rolls around, etc. And I've always thought Apple in particular
0:28:17really does work on an annual schedule. And it's only gotten more regularized under Tim Cook.
0:28:24And I think, again, in the way that an effective leader instills their personality into the
0:28:32organization, that the Tim Cook-ification of the last decade of Apple is that they're
0:28:38even more reliable, right? That it's even more certain that WWDC is going to happen
0:28:46exactly when you think it is. And more, it seems to me, more and more of their products are,
0:28:53hardware products are getting more predictable in their updates. I mean,
0:28:59would be a whole tangent for the show to talk about like, why the 24 inch iMac hasn't been
0:29:06updated with the M2. But you know, it seems to me that in the Apple Silicon era, though,
0:29:11Macs are going to be getting updated on a more regular basis, like iPads at least,
0:29:16even though it doesn't seem like anything other than the phone and the watch are on it precise.
0:29:22Every single September, that's when the new ones come out. But WWDC is,
0:29:28I would say, clearly the two tentpole annual things in Apple schedule. WWDC in June,
0:29:37the iPhone/watch/if there's something else special to announce, do it with the iPhone
0:29:44September event. And the thing about WWDC, and I would say in the Tim Cook era,
0:29:51especially the last five, six years, has really become a sort of… It's not just, "Here's
0:29:58a bunch of new developer APIs, and here's in the morning, we'll show you some new features that
0:30:05are coming out in the fall, it's sort of, "Here's what we're doing for the next 12 months."
0:30:11Right? We say Apple is very secretive, and they are, and they hopefully won't leak what is going
0:30:18to be announced at WWDC. But effectively, WWDC has become a way for them to leak stuff that won't be
0:30:25coming until next March or April. Right? Because look at this week with the—this is my segue,
0:30:32my very long meandering segue. This is the week where Apple updated, I think, literally every
0:30:39operating system they have, right, with a sort of feature update, Mac OS, iOS, iPad OS, watchOS,
0:30:46tvOS, the HomePod OS, something else. Oh, the studio display even got a firmware update,
0:30:53right? It's like all at once. It's Michael Corleone taking pictures today.
0:30:57Also, you might even count this as another platform getting an OS update, the pretty
0:31:02substantial Safari update. Yeah, yeah, I would say so. I would say that WebKit
0:31:08counts as an Apple platform. They don't call it WebKit OS, but web browsers, the web rendering
0:31:14engine effectively, it certainly is the complexity and size of an operating system, right? I don't
0:31:20know that it's completely fair, but I would guess WebKit overall is a larger software project than
0:31:27watch OS yes I would think so but who knows but yeah at least it's fathomable
0:31:33to me I mean I mean they of course both of them of course lean on all of the
0:31:38infrastructure of other OS projects but right right it's it's that significant
0:31:44and it is it's been a major update for WebKit and they including a couple of
0:31:49again this could be a whole segment of the show that we probably don't have
0:31:52time for, but some mobile web kit features that web developers have been clamoring for
0:31:58for a long time, like the notifications and which is a big deal for like the what I call
0:32:05PWAs, progressive web apps, but it basically what when you save a web app to be an app
0:32:10on your home screen and making them more feasible to act like you expect an app to be able to
0:32:18act by giving you notifications and stuff.
0:32:20Right. And I don't, I don't know that much about that world, but I, I have had the feeling,
0:32:25the impression for years and years that to summarize most web developers have been grumpy
0:32:31about the speed with which Apple has embraced those kinds of features. So I don't know.
0:32:37Well, I think there's two, there's two sort of slants to the criticism. I think, and again,
0:32:44maybe I'm biased as a Safari user and as a native app proponent and as somebody who does not want to
0:32:53see and I guess most, my bias is most covered, colored by my deep antipathy for the web
0:33:05amplification of desktop apps, right? Just to, I know that not all written with Electron, but
0:33:11that's just use Electron as a catch-all umbrella, the electronification of desktop software
0:33:17development. I don't want to see that happen to mobile too, or at least want to see it happen as
0:33:22slowly as possible. And so there, in some sense, I'm rooting against Chrome and Chromium taking
0:33:30over the mobile market the way they have the desktop market. And so there's one contingent
0:33:37of web developers who just want that to happen, right? That they, because they're all in, the way
0:33:42they develop software is to just use all of the bleeding edge Chrome features and they really see
0:33:48right once run everywhere as nothing but a positive, or that the positives greatly outweigh
0:33:56the negatives in the overall scheme of things, want that to happen. And I just disagree. I'm
0:34:02not saying they're wrong. I'm just saying I disagree with them, right? There's a difference,
0:34:05right? I see their point, but I don't think that's good for the platform experience.
0:34:11But on the other hand, I think there's more pragmatic, less zealotrous web developers who
0:34:16just are frustrated by the pace that Apple has adopted some of these technologies that to me
0:34:23aren't about. The conspiracy theory such that it is that Apple as a company holds back WebKit
0:34:31on mobile simply to ensure that the App Store remains the only feasible way to do X, Y, and Z.
0:34:39**Matt Stauffer** Yeah. That seems false on the face of it because they could do such a worse job
0:34:45than they do. If they really wanted to hamper mobile advancement of web apps,
0:34:52then we would not be seeing even a quarter of the enhancements to WebKit over the years.
0:34:59Right, and it's sort of an insult to the WebKit team. And I can't say I'm like tight buds with
0:35:05any of them, but I do know several of the people on the WebKit team and they've been,
0:35:08and a lot of them have been there for a very long time. And they're very smart and very proud and
0:35:13very, very proud of both their work and WebKit's position in the world. And it's frankly insulting
0:35:21to argue that they just go along with, "Hey boss, we would like to do X, Y, and Z." And they're like,
0:35:28"Nope, because we want to protect the App Store," and that they just take it. And they're like, "Okay."
0:35:31Right. Yeah, they have different priorities than the people who are clamoring for X, Y, and Z
0:35:38features. And just so often the case when... I think WebKit embodies that in the context of
0:35:46the web, they embody that Apple kind of idea of being willing to say no in order to focus on the
0:35:56things you think are more important. And I deeply respect the WebKit team as well. I think it's I
0:36:03know, I, for the record, I'm also a Safari user, but I get it, there are some shortcomings, but I
0:36:09think they focus a lot on things that that matter to me anyway. So I, I think I'm gonna piss some
0:36:16people off. Well, I keep meaning to redo it, and I will, but I'll hear I the talk show is where I
0:36:22I spoil stuff that might be forthcoming on Daring Fireball.
0:36:26And often what I spoil here never comes out anyway.
0:36:30But a couple of years ago,
0:36:32I think it was back around the time or before,
0:36:36remember when Apple had the round table where they said,
0:36:40okay, this trashcan style Mac Pro was a mistake,
0:36:44but we're not giving up on a pro market,
0:36:46but we don't have anything to announce right now,
0:36:48except for the fact that we have been working
0:36:51on a pro-oriented iMac, which they didn't say
0:36:54would be called the iMac Pro, but which was the iMac Pro.
0:36:58That was a couple years, it was like 2016, 2017,
0:37:00something like that.
0:37:01So it's been a while since I did this,
0:37:03but what I did is I, I think it was pretty clever,
0:37:07is I wrote like an Apple script to go through.
0:37:09I picked like 25 popular websites.
0:37:13I forget how I picked them, but they felt like a fair thing.
0:37:16And I wrote an Apple script that would open them up in tabs
0:37:19and do some, use like the GUI scripting
0:37:22to do fake scrolling, like to sort of simulate a user
0:37:27actually reading the web page.
0:37:29Open it up, wait 10 seconds, scroll down,
0:37:33wait 10 seconds, scroll down, open a new tab,
0:37:37do this with a new website, do the same thing.
0:37:41Keep 10, 15, 20 tabs open in the background,
0:37:45then close them all, do it again,
0:37:47and just keep going until the battery's out
0:37:51and see how long the battery lasts in Safari.
0:37:54And then ran the same Apple script against Chrome
0:37:56against the same websites.
0:37:58And I thought there's other,
0:38:02I'm proud of myself because it seems like something,
0:38:04you know, like Macworld Labs used to do back in the day
0:38:07and have these test suites.
0:38:08But I thought it was a pretty good simulation of like,
0:38:10hey, with nothing but a web browser running,
0:38:12what's the difference between Chrome and Safari?
0:38:15And I forget the answer, but it was significant, right?
0:38:17It was like hours on a MacBook at the time.
0:38:20It was hours of difference overall.
0:38:22So I keep meaning to redo it and see what,
0:38:25here in 2023, what's the difference?
0:38:27I don't know if Chrome's closed the gate at all
0:38:30or closed the gap.
0:38:31I might've, you know, every once in a while,
0:38:33I notice in the Chrome release notes
0:38:35that they say something about battery life improvements,
0:38:38but I don't think there's no doubt in my mind,
0:38:41and again, it's the order of your priorities.
0:38:45Of course the Chrome team at Google wants battery life to be good, right? It's just
0:38:50not as high a priority for them as it is for WebKit, right? It's both teams value that.
0:38:56And I'm sure in the same way, WebKit values adding new features for web developers. But
0:39:01there it's a little bit of a lower priority for WebKit to stay on the bleeding edge of
0:39:08web developer APIs and more of a priority, if not the highest priority, to do everything
0:39:14a web browser has to do the most power-efficient way possible. Whereas at Google, maybe the
0:39:21APIs and staying on the bleeding edge and pushing the platform forward as a developer
0:39:25thing is the highest priority, and being energy-efficient is second or third. But I wouldn't be surprised
0:39:32if that's a difference. And I just know that that's a big deal at WebKit. And for
0:39:38web developers who only care about us, we built this whole web app, and if only Safari
0:39:43supported this API, too, we could take out all of this extra code that we have to do just to run on
0:39:49the iPhone." It just completely doesn't bother them at all that there's something like power
0:39:55efficiency, which consumes massive amounts of the WebKit team's attention and time. And they're
0:40:01like, "What do I care about that?" Whereas for them at WebKit, they're like part of Apple,
0:40:06and it's like part of the, "Hey, how long does the battery on your iPhone last?" Right? And if WebKit
0:40:12is less efficient, that's a lower number. And that makes
0:40:15Apple as a whole look bad. Yeah. Any any features from this
0:40:20week's OS updates that jump out at you?
0:40:23As not really, I think I haven't really absorbed it yet. I felt
0:40:27like there was something I noticed even in the Mac OS
0:40:31update, but it's just not jumping to my mind right now.
0:40:35Yeah, it's a couple tell me something that what's your
0:40:38favorite thing? And if you have it, and then I'll tell you, if
0:40:41that reminds me that that's my favorite thing too. Well, I don't know it's my favorite. It's actually
0:40:46because I'm sort of ignorant of the whole field, but it seems like the biggest one is that Apple
0:40:51Music classical is tied to this. Right? So if you want to use this new long rumored, or I guess they
0:40:58announced it, I guess it wasn't like it popped out of nowhere. They somehow announced it months ago
0:41:03that there would be a new entirely set standalone app just for classical music in Apple Music,
0:41:09which is fascinating, right? That it's not just a genre, but is considered,
0:41:16for aficionados, fans of classical music, so different that it deserves its own standalone
0:41:22app, but you need to be on iOS 16.4 or Mac OS 13.3 to get it. I don't really have any strong
0:41:30opinions about it because I'm not a classical music person. And even though I do enjoy it,
0:41:34who doesn't like classical music, even if you're not a fan? I mean, it's kind of
0:41:38of, yeah.
0:41:39- Well, probably a lot of people would say
0:41:41they don't enjoy it, but then you get some familiar
0:41:45recognizable riff in front of them, and they're like,
0:41:47"Oh yeah, I like that song."
0:41:48But I also am not an aficionado,
0:41:51but what I like about this is it intrigues me
0:41:55to think that it might help me become more
0:41:57of an aficionado.
0:41:58Like it sort of, I think it's addressing,
0:42:03and I don't understand all of the reasons
0:42:06for it being a separate app,
0:42:07I kind of got the gist of it. In a sense, classical music is... I'm hoping that the app
0:42:18sort of addresses a problem that I have when I am inspired to explore classical music. It's like,
0:42:27"Well, which performer do you listen to?" If I want to go listen to some Chopin, which I do happen to
0:42:35to like, even if I'm not an aficionado, how do I find out whether the first performance
0:42:40that pops up?
0:42:42Because here's in summary, all of the classical music that we listen to our covers, right?
0:42:48Yeah, yeah.
0:42:49So it's like if you had a situation where it was like, okay, let's listen to the Beatles.
0:42:55But the only caveat is they're all covers, then how do you find out which version of
0:43:01help to listen to?
0:43:03And I could use some guidance in that respect.
0:43:06I think maybe that sort of gets at the uniqueness of the problem that Apple's trying to address
0:43:12here is that it's not the same as other genres because other genres' performing artists are
0:43:22the highest level of categorization.
0:43:26And in the context of classical music, the performing artist isn't the highest level
0:43:32classification. It's the author, the writer, the composer.
0:43:36Right. Beethoven or a whole category, Baroque or something. I'm sort of
0:43:41speaking off the top of my head here. But yeah, our friend Jessie Char is a trained
0:43:46musician. I believe she plays the cello? Cello sounds right to me.
0:43:51Cello sounds right to me. Yes, he's a cellist. Is that how you pronounce it? Cellist. But she had a
0:43:56a good thread on Twitter weeks ago, like when Apple said, "Hey, you can pre-order this
0:44:02app," which you didn't have to pay for it, but I guess that way you'd automatically
0:44:05download it from the App Store when it was available. She had a good thread on Twitter.
0:44:10I'll make a note here and try to put it in the show notes, just sort of explaining the
0:44:14way that the simple artist album song categorization doesn't really work for classical music fans
0:44:21and therefore needs a new app. So there's a major feature.
0:44:24I'm looking forward to that. I feel like this is kind of a release that has a bunch of little things.
0:44:30Yeah. So it's, it's hoping for the best. Yeah, it sort of seems like that the schedule is,
0:44:36the annual schedule is they announce all this stuff at WWDC and they give us developer betas
0:44:43and that some of this stuff is in there, right? Because there's some stuff that's been percolating,
0:44:47all these things have been percolating on teams greater than a year, right? Like when a major new
0:44:52feature shows up in any of these OSs. It's not because they started it a couple months before,
0:44:59it's like months or 24 months or longer. Some team's been toiling away, didn't make the cut
0:45:04last year, like, "Ah, that's not ready this year." But the team keeps working. And then here's the
0:45:09stuff that's ready for this year's OS updates. Some of this stuff actually is ready. Those
0:45:15developer betas are hit and miss in June in terms of what's ready and what's not. September is when
0:45:21the iPhone comes out and iOS and then the Mac OS. Last year they delayed iPad OS 2 until October.
0:45:29But in the annual sense, September, October, what's the hell is the difference, right?
0:45:33Everybody's impatient for iPad OS at the end of September because iOS came out. But they come out,
0:45:40they're a little buggy at least. Maybe I'm being euphemistic. I don't know. Some years they're more
0:45:46buggy than others. iOS 13, for whatever reason, really sticks out at me as sort of a really rough
0:45:52launch. That was the one I think they had the emergency, almost a hotfix, 13.1,
0:45:59was ready by the same day the iPhones actually came out. It was really pretty bad.
0:46:04Jared: Yeah, this past year seemed pretty smooth, if I recall correctly.
0:46:08Pete: Yeah. The people out there, and I'm in a weird spot because, again, I am getting older.
0:46:14we're all getting older, but I'm at the age where I'm more crotchety than I used to be.
0:46:18And my inclinations are to be conservative with, especially my Mac, right? Because that's where I
0:46:26do my most work. And so I don't jump on the new version of macOS on my main work Mac until I feel
0:46:34like it's stable. But I'm in the racket of writing and talking about these things, so I need to be
0:46:39using them. And using them on spare machines isn't the best way to get to know them. But the people
0:46:45out there who don't have a need to write or talk about this and who wait a couple of point releases
0:46:52before they update, probably wise, right? It's overall, what are you missing? And I feel like
0:46:57these 0.4, it's like 0.1, 0.2 are always like bug fixes. Like, "Ah, here's this thing that we
0:47:04released in September. That was, yeah, that was, that was buggy bug fixes. The point three releases
0:47:10right around like Christmas are like, "Hey, some of this stuff, here's some new stuff that was
0:47:15promised to WWDC." And then like this March 16.4 Mac OS point three release. This is sort of like,
0:47:24they're all, they've almost got everything they announced last year at WWDC. I forget, there's
0:47:27like one or two things. There's like some silly Apple Pay thing, not Apple Pay later,
0:47:34which actually did start this week, but something else, some kind of—I forget what it is, but
0:47:40pretty much now they're caught up. And I know there's a 16.5 beta that's already out, but
0:47:45now's the time of the year where you can upgrade with confidence, if you've still been holding on.
0:47:53I'm amused by the fact that Apple has so effectively used emoji updates as a lure.
0:48:00Oh, yeah. Without question. Oh, without question. And they've, they've, I don't think they've done
0:48:07it this year. I should check my email. But in years past, I've gotten press from Apple,
0:48:12just about the emoji and they, they do. I think you can, I do think I, again, nobody's officially
0:48:19told me this. Craig Federighi has never broken character behind the scenes and said, "Yes,
0:48:24off the record, we hold the emojis until we have a really stable release." But they do two things.
0:48:31There's a couple of levers that they can pull to sort of encourage/discourage people from updating.
0:48:38The big one is whether they hit the button to sort of push the update to notify people,
0:48:46right? Like to put a red badge on the settings app. So even if you're just a normal person with
0:48:52an iPhone, and you don't even follow the Apple news, you don't even really you don't even remember
0:48:57or notice that they have an annual schedule for OS updates, right? Or phone updates, right? You just
0:49:03buy a phone, maybe in August, right, which most people would agree would probably be the worst
0:49:08month of the year to buy an iPhone because the new ones are probably coming in September. But if you
0:49:13break your phone or lose your phone or just get sick of your old phone in August, lots of people
0:49:18go in and buy iPhones in August and you're just a normal person. And is there a red badge on your
0:49:24settings app that's saying, "Hey, there's a new version of iOS to install"? They don't do that
0:49:29usually with those like 16.0s in September. Like you can go to settings, general, check for off
0:49:37updates and it'll say, "Oh yeah, here's an update. Do you want to install it now?" But they don't push
0:49:42it until like 16.1 or something like that because they kind of I don't think they want to deal with
0:49:49the customer support issue of a billion devices around the world all on the point oh so that's
0:49:56one lever they can pull and it gives you a sort of external sense of which versions of the os
0:50:01they feel like yeah let's get we're better off we'll have fewer customer support problems
0:50:07if we push everybody to this because this is a better update with more features
0:50:11And then the flip side of it is the emojis
0:50:15Right, and it's like all of a sudden somebody's sending you the the goose emoji that's brand new
0:50:21But it shows up as a what I what do they show show up as like a square or something
0:50:26I forget what the square with the question mark in it
0:50:29Yes, I fear an accent or something like that, but it's like any different on the Mac and iOS, too
0:50:35All right, but then you're like what was that and they're like, oh, it's the goose emoji and it's like, ah, how do I get that?
0:50:39Well, you got to update your iPhone and they definitely they definitely use it and it definitely motivates people. I should take this moment
0:50:46to to I'm I'm more than half serious about this that I'm gonna keep hammering on the fact that I want a chef's kiss
0:50:55Emoji, I keep mentioning it on Daring Fireball. Yeah, I'm going to keep mentioning it
0:51:00Do you have any any pet emoji that are missing that you wish were there?
0:51:05Not off the top of my head
0:51:07it's just one of those things where every once in a while, I think like you, I go to give the emoji
0:51:14that I assume must be there. And then it's not. And it's like, what the hell? Like, how is there
0:51:20like a sushi emoji from like a specific type of eel or something? I don't know. I don't know if
0:51:26that's true, but sometimes there's some really specific to my mind stuff and then you're like,
0:51:31but there's no like popcorn or something. I don't know if, you know.
0:51:35No, popcorn is definitely there because I know because every everybody always does it when like
0:51:39when another indictment is hinted at with Donald Trump, right? Right. Get your get your popcorn
0:51:45ready. Here we go. So there are there are some times I think when things that to my mind are
0:51:50as obvious to be there as popcorn are not there. And I think I would agree with you that chef's kiss
0:51:57that it fills a void it would fill a void. And I can't think of other ones that off the top of my
0:52:03head I've I've gone for but I think often it's like some kind of particular vegetable or fruit
0:52:11that seems to fit a specific there's no like there's no pickle I have here's my list I just
0:52:20brought it up in Apple note so at the top of my list chef's kiss I this is the one I want to make
0:52:25it it's got to be there it's a worldwide thing everybody knows this it's a chef's kiss it means
0:52:30the meaning is so perfect, right? It, to me, epitomizes the communicative beauty of emoji.
0:52:38I think the whole emoji thing must be one of the funnest, best gifts to the world of linguistics
0:52:46in decades, because people are communicating expressively through pictograms in a way that
0:52:53we haven't for centuries, right? The millennia, maybe, in Western civilization, right? It like,
0:53:01takes you back to like hieroglyphics and stuff. But humans, we're natural at it, right? Everybody
0:53:06I know, my dad, my mom, old people, certainly younger people, everybody can communicate with
0:53:13emoji. It's wonderful. And it really is. In some ways, it's a picture. I don't know if
0:53:19any emoji really says a thousand words, but it certainly says a chef's kiss emoji can save you
0:53:24just a lot of typing on your thumbs. I need it. Here's my list. It's not long. Chef's kiss,
0:53:29number one. Pickle. I don't know why. My son's a big fan of pickles. And sometimes, like, so like,
0:53:36what do you want me to get from the store? You can't send a pickle. Cucumber is the closest.
0:53:41Shovel. Doesn't that seem like there should be a shovel emoji? That's one that I was like,
0:53:46I can't I cannot believe that it's not there. I forget why I wanted it trombone because I feel like you could use that for the
0:53:52Womp-womp
0:53:54exactly exactly
0:53:56And then the last one on my list sombrero. I don't know why I wanted a sombrero sombrero
0:54:01But it feels feels like they've got a lot of hats in there
0:54:04Why not a sombrero?
0:54:05Because I just typed in sombrero and - this is I'm looking at Apple's character viewer
0:54:10To their credit when I type in sombrero I get like a cowboy face
0:54:16woman detective the English guards all these different things with hats there's
0:54:23a military helmet top hat yeah and I know some bro yeah I know there's I mean
0:54:29this I guess that's the problem with emoji is there are a heck of a lot of
0:54:34things a lot of things in this world so I've reordered my list I put trombone at
0:54:39second underneath chef's key after we get after we get chef's kiss in next
0:54:42year, then my new campaign will be for the trombone because we need that.
0:54:46Womp, womp, womp, womp.
0:54:48So here's the thing.
0:54:49I don't understand the process, but I feel like I've witnessed people making
0:54:54successful campaigns to get emoji added.
0:54:57Yep.
0:54:58So I think if you get like a designer buddy on board and
0:55:04Oh, if only, if only I knew someone.
0:55:06Right.
0:55:08Exactly.
0:55:09Only you do somebody with a gift for the designer arts.
0:55:14But I think I feel like I've seen people back when.
0:55:18Back when Twitter was a thing, I feel like people were celebrating, getting a new thing added to the emoji verse.
0:55:25Yeah.
0:55:26Yeah.
0:55:26All right.
0:55:28Well anyway, let's see, let's see.
0:55:29I'll have you on in a year and let's see if I can get it done.
0:55:31It's a, it's a chair.
0:55:33Hold, hold, hold on.
0:55:34It has to be a bat or a chat.
0:55:36Oh, all right.
0:55:37So if you get it done within a year, then I owe you $5.
0:55:42- All right, all right, $5 bet.
0:55:45- I got that.
0:55:47- How about a chef's hat?
0:55:48- Chef's hat, hey, chef's hat's in there, isn't it?
0:55:52- Oh, there is, there is, yeah.
0:55:54That's so you can--
0:55:55- Not a chef's hat, but a chef.
0:55:57- Yeah, so you can make a chef kiss with the chef
0:55:59plus the lips, there's like a kissy,
0:56:02like a lipstick lips.
0:56:03So that passes as chef's kiss,
0:56:06But it's it's not right though because you don't the chef's kiss has to be with the with the with the okay sign by the lips
0:56:12Like ah, it's a little different
0:56:14Anyway, let me take a break here. Thank thank our first sponsor of the show. It's our good friends at collide now
0:56:19They spell it k o l i de
0:56:21They have some big news if you are an octa user
0:56:26Now that's ok, ta but if you if you're in the market for this, you know, what octa is already
0:56:33Well, if you are, they can get your entire fleet to 100% compliance.
0:56:37How?
0:56:38Well, if any device in your organization is not compliant, then with Collide, the user
0:56:45can't log in to your cloud apps until they've fixed the problem to get back into compliance.
0:56:51It's that simple.
0:56:52If they're compliant, they're in, and if they're not compliant, they can't get in until they
0:56:56fix it.
0:56:57And this way, Collide patches one of the major holes in zero trust architecture, and that's
0:57:02device compliance. Without Collide, IT struggles to solve basic problems like keeping everyone's
0:57:07OS and browser up to date. Insecure devices are logging into your company's apps because there's
0:57:13nothing there to stop them. Collide is the only device trust solution that enforces compliance
0:57:19as part of authentication and it's built to work seamlessly with Okta. The moment Collide's agent
0:57:25detects a problem, it alerts the user and gives them instructions how to fix it. If they don't
0:57:31fix the problem within a set time, they're blocked. That's it. Collide's method ensures
0:57:36fewer support tickets, less frustration, and most importantly, 100% fleet compliance. Visit
0:57:43collide.com/thetalkshow to learn more or book a demo. That's K-O-L-I-D-E.com/thetalkshow.
0:57:54Thank you, Collide. All right, let's get to the meat of the subject here. I wanted to talk about
0:57:59this AI stuff. I have you been messing around with this, the child, these various chat GPTs
0:58:05at this point stuff is moving so fast where the pace of my podcast isn't enough to keep
0:58:10up with it. All right. Yeah. I have been messing around with it as little as possible. It's kind
0:58:18of like when the, when the image generation things came out, we were all like, Oh geez,
0:58:24if I don't stop myself, I'm going to spend like a week straight just interfacing with this thing.
0:58:29But I dip into it every once in a while. I spend probably about an hour with it, and then I
0:58:35forcefully pull myself away from it. But I'm fascinated and scared and confused and bewildered
0:58:44by it. I assume you're probably some combination of those same things.
0:58:47Yeah, and it's weird because I got into and I had the same feeling months ago. I think it was months
0:58:54It's all moving so fast with the image generation where it wasn't that it didn't interest me
0:58:59it was honestly my own self-awareness of
0:59:02I'm like I I have XY and Z I want to do here's some articles. I've been
0:59:08Had been working on I want to get I want to write about XYZ
0:59:11I want to get a new episode of the podcast out. I wanted blah blah blah if I get into this
0:59:16I'm going to not shower for two days, and my wife is going to be like, "John, where are you?"
0:59:22And I'll be like, "I'm just here working at our—she'll be like, "Dinner's ready." I'll be
0:59:25like, "I'm up in five minutes," and five hours later I come upstairs. And lo and behold, once I
0:59:30got into them, that is what it was like. I did get sucked in. Mid-Journey in particular, which
0:59:36is a combination of both seemingly produces more unbelievably artistic output as opposed to sort
0:59:47of impressionistic output, even though the impressionistic stuff is fascinating to me too.
0:59:52And sometimes untold books have been written about realism versus impressionism in art.
0:59:57But it's just more amazing. But like with Mid-Journey in particular,
1:00:01have you ever messed around with it? That's the one that you go through Discord.
1:00:05Yep, I have. And it's so it's not productive. It's anti productive, because what you do,
1:00:13and I guess if you pay, you can get your own private chat on the server. But and again,
1:00:19I talked about this before, but for anybody out here who hasn't done this, when you mid journey
1:00:24is this GPT powered, I think I forget what the back end but anyway, AI image generation,
1:00:31where you just type natural language descriptions like, "Daniel Jowkitt playing volleyball on a beach,
1:00:40photographic style," and it'll, if it knows who Daniel Jowkitt is, will render a photograph of
1:00:45Daniel Jowkitt. Famously, like last week, people were generating images of Donald Trump being,
1:00:51resisting arrest on the streets of New York while they drag him in. But you could say "photographic
1:00:57style, or you could say in the style of a painting by Norman Rockwell, or in the story
1:01:03of a comic book by Jack Kirby, or something like that. And son of a bitch, that's how
1:01:09they come up. And it's not like a hint of it like, "Oh, that does look a little bit
1:01:14like a photograph," or "That does look a little bit like a Norman Rockwell." It's
1:01:17like, "No, it looks like a Norman Rockwell painting," or "It looks photorealistic,"
1:01:22or blah blah blah. But with mid journey, the interface is through a Discord server. And
1:01:28when you're like a free customer, you just go into these free channels, type these commands,
1:01:34and then you wait like 30 seconds to get these things. But meanwhile, there's like dozens
1:01:39of other people, random strangers from around the world in the same channel as you on Discord,
1:01:45typing their commands, and you get to see their results. And it's, it's straight, it
1:01:51goes by pretty fast, fast enough to be almost hectic, but not quite frantic. But while you're
1:01:58waiting for the thing you're trying to make, these other amazing things are just flashing
1:02:03by. It's like captivating. You can watch these other people zoom in, narrow in on what
1:02:11they were going for. Like, "Here's four choices, and then, oh, give me four more based
1:02:17on the second one here. And it's just engrossing. It really is incredible. But on the other
1:02:25hand, I don't need lots of images, right? I don't illustrate my posts on Daring Fireball
1:02:30with images, so I'm not using these things to generate a hero image to go atop my articles.
1:02:37So in some ways, the chat stuff is more up my alley, right? More of a verbally oriented
1:02:44person. It also is another really good way to lose a lot of time once you start digging
1:02:51in. But it's not... I've said this before, to me, when I'm on Discord playing around
1:02:59with MidJourney, it's the way I felt as a 10-year-old in a busy coin-op arcade, right?
1:03:07With all the newest games, including games I've never even heard of before. Like, "Oh
1:03:11my god, what's that? And it's like, being in it, there are no
1:03:15songs, but there might as well be bings and bops and pings and
1:03:19pals and ding ding ding. And, and what's this? What's that?
1:03:23And let me wait for this now. Right? And it's that same sort
1:03:27of like, hurry up and wait feeling of being in an arcade.
1:03:30Like I want to play this new game. But here's a kid who's
1:03:32playing ahead of me. And he's really good at it. So I got to
1:03:35wait for his game to be over. Whereas the chat stuff is a
1:03:38little bit more sedate. It's just like chatting with a person except it's a
1:03:43friggin robot and it's amazing and yeah there's something about the infinite
1:03:50possibilities with an undefined interface that makes it really
1:03:56compelling to keep trying new things you know like if you're playing around with
1:04:01a new app let's say you might go through if you really if you're really intrigued
1:04:05by the app, you might go through every menu item and see what does this do? What does this do?
1:04:11And then when you're an old-timer Mac user, you go, "Aha, I know there might be some options."
1:04:18So you hold down the option key and go through every menu. But you never think,
1:04:22"What happens if I type into Photoshop, 'Paint me, Donald Trump on a cowboy pat.'"
1:04:33So the fact that it's endless and I think I agree with you, the text stuff has been a little more
1:04:40compelling, not just because I think I'm also a little more text oriented, but because the format
1:04:48is more humane. Because the photos thing is really super interesting, not photos,
1:04:59but the graphical stuff is really interesting, but it's generating stuff in certain styles,
1:05:03whereas the chat things are literally trying to come across as human personalities, which
1:05:10is really interesting. I've had a few kind of like areas I tend to go down. I was chatting with
1:05:17Manton Reese, my co-host on Core Intuition, about my tendency. One of the things I try to do with
1:05:24with a lot of these things is whoever runs the AI,
1:05:28try to get the AI to bad mouth the owners, you know?
1:05:33So it's like, for some reason,
1:05:34I just find that really satisfying.
1:05:36So like trying to get Bing to say terrible things
1:05:39about Microsoft is just, it's a wonderful pastime.
1:05:43And there's no real graphical equivalent to that.
1:05:46Like I couldn't get stable diffusion or Dolly or mid journey
1:05:52to make an image that so perfectly captured
1:05:56like a sarcastic snipe of...
1:06:01You could do some pretty good stuff, but...
1:06:04- Yeah, you could ask it to render an image of Calvin
1:06:07from Calvin and Hobbes peeing on a box copy
1:06:12of Microsoft Windows or something,
1:06:14but it's not the same thing as getting the chat thing
1:06:17to bad mouth it and say, yeah.
1:06:19- Right, and it's not the same thing,
1:06:21And it's also not as scary as the chatbots.
1:06:25Like, I'm thinking back to the conversation I heard
1:06:27between you and Ben Thompson on dithering
1:06:30about his experience with being like,
1:06:34threatening to report him and stuff like this.
1:06:36I don't think anybody's ever had one of,
1:06:40maybe I shouldn't say nobody ever has had,
1:06:42but I don't think it's common for the visual generators
1:06:46to come up with results that are easily perceived
1:06:50as threats against the user.
1:06:51Like they might come up with like a weird,
1:06:54violent looking thing,
1:06:56or maybe a pornographic looking thing.
1:06:58- Right.
1:06:59- But it's not like an image that says like,
1:07:01they're not commonly like generating ransom notes
1:07:06that say I'm gonna come to your house and kill you, right?
1:07:09But some of these chat interactions have got,
1:07:14again, I think because of the humanity,
1:07:17the fake humanity of them,
1:07:20they feel like there's a purpose to them.
1:07:22- I just texted you an image that's from episode 359
1:07:26last fall with Merlin was on the show.
1:07:29And I think this is from Mid-Journey,
1:07:31I forget, I think it was Mid-Journey,
1:07:32where I was asking it to render Donald Trump sad,
1:07:35alone in an office with a can of Diet Coke.
1:07:39- It's so good.
1:07:40- And I used, it was so good,
1:07:41I used it as the album cover for the episode.
1:07:43I'll put it in probably as a chapter art.
1:07:46So if you look at your overcast right now as I'm talking,
1:07:49hopefully you'll see the same thing.
1:07:50- It's so good, but yeah, the chats,
1:07:53so to me, I guess the way it feels,
1:07:55and I know how it works, and it's all,
1:07:57or I don't know how it works,
1:07:59but I have a pretty good idea as a,
1:08:01somewhat of a programmer and a long ago comp sci graduate
1:08:05how these systems work.
1:08:07It's, part of the magic is even the people
1:08:10who understand them the best,
1:08:11the people who make these large language models
1:08:14admit that nobody really knows how they work
1:08:17once they're trained, right?
1:08:18that is sort of, it is, it's a little scary. Like even not, not like alarmist, but just like Sam
1:08:24Altman, who's the, I guess, I don't know what his title is at OpenAI, but I think he's in charge
1:08:30now at OpenAI has even, I just, just revisited a 2016 profile of him that was large in the New
1:08:37Yorker. That was at the time he was main job was he was the CEO at Y Combinator, the startup
1:08:43incubator. And OpenAI was new. And the profile was mostly about his work at Y Combinator,
1:08:50but then they touched on the OpenAI stuff. And he even said, "Nobody knows how these things work.
1:08:55They know how to train them, and they kind of know why they work. But once the model is working,
1:08:59that's the whole thing. It's like the actual code that runs on the computers behind the scenes
1:09:06comes out of this process, and it's not human-written, these algorithms. So it's kind of wild.
1:09:11Yeah. Speaking of the whole dangerous seeming aspect of these, did you see the headline today
1:09:19about some kind of a group letter? Yeah, I did.
1:09:24Open letter asking that there be a six month pause on development of these systems?
1:09:30Yeah, but I also saw that it's apparently the whole thing is a mess because some of
1:09:37signatories didn't actually sign. Like, I guess they're saying Sam Altman in particular signed it,
1:09:43but apparently he didn't. One of the people who did sign is infamous Twitter engineer Raul Ligma.
1:09:50I forget. Is that like a fake engineer?
1:09:56Well, you know Ligma, right?
1:09:58Oh, Ligma. You told me right here.
1:10:02Raul Ligma was one of the cats who, the day that Elon Musk first initiated layoffs,
1:10:08did interviews out, you know, was carrying a box of stuff from his office.
1:10:12Jon Streeter Right. Oh, and it turned out to be actors.
1:10:15John Green Yeah, just two guys who were pranksters. But then
1:10:18Elon Musk actually, I don't know, hired them or at least took selfies with them weeks later. But one
1:10:24of the fake names was Raul Ligma. And he's apparently a signatory on this open letter.
1:10:30I thought I do think that the industry died. I do think caution is advisable, right? It's
1:10:36Yeah, it's it's not quite as clear-cut how things could go bad as say nuclear
1:10:43weapons and nuclear power right that yeah, which in the end the the
1:10:49Dangers the same right a nuclear
1:10:51Well, I guess a weapon may not be fired as an accident, but some kind of nuclear catastrophe, right?
1:10:59whether it's a weapon or a Chernobyl or Three Mile Island, right, gone wrong. The worst-case
1:11:06scenario is obvious, right? We know what the worst-case scenario is. It's a mass death
1:11:12because of radiation and the half-life, and it'll take forever for the Earth to recover, blah, blah,
1:11:19blah. We don't know the worst-case scenario for AI. We have imagined scenarios, but we don't know.
1:11:26So there's a big difference there. But I do think, though, that scientifically and as a profession,
1:11:31the same sort of caution is called for, right? It's just because it's software. I do think
1:11:37there's a human instinct to think, "Ah, software, it's just something that nerds do and they blah,
1:11:42blah, blah." No, it's the danger. There's real danger with AI. You don't have to imagine Skynet.
1:11:48There's other bad scenarios. But the Skynet scenario does seem more and more real, right?
1:11:53and the HAL 9000 scenario seems more and more real.
1:11:56I mean, I guess it comes down to whether anybody hooks one of these things up
1:12:00to some interface that actually gives it physical power.
1:12:05Right. But a six-month moratorium doesn't sound—I don't see the point of that.
1:12:09It doesn't make any sense to me. And if anything, it just sort of sounds like, "Hey,
1:12:13GPT-4 is ahead of everybody right now, so please give GPT-4 six months of undisputed
1:12:21preeminence in the field. It doesn't seem right to me. But anyway, the point I wanted to, before
1:12:26I forget it, because the point I wanted to say about image generation versus chat is to me,
1:12:32image generation feels like pure magic. It is it doesn't feel like intelligence per se. It just
1:12:40feels like magic. And I know it seems so simple. Low these 30 years later, but like, or I guess 40
1:12:50years on it, geez. But the first time you used Mac Paint back in the day, and you use the paint
1:12:57can tool, flood film, it was like magic, right? Because it's like, the pencil tool, like Mac Paint
1:13:05was so much fun, even just black and white, you take the pencil and you start drawing pixels.
1:13:10And it's like, you can draw it's like, holy shit, I'm drawing on a computer live, right? Like it,
1:13:14that wasn't something you could do before. But you understood what drawing with a pencil was like.
1:13:22Drawing with a pencil on your Macintosh screen was like drawing with a pencil on paper. You
1:13:27drag a line and you get a line, and you drag a circle and you get like a circle-y type shape.
1:13:32I'm saying circle-y type shape because drawing a circle with a mouse usually
1:13:35didn't come out all that circular. But flood fill, like I always hated, I don't know about you,
1:13:42I hated coloring as a kid. I did not enjoy using coloring books. And I know it's a hobby now,
1:13:48and if you enjoy it, God bless you. I can see why people, adults enjoy it now. But it always
1:13:54bored me to tears, right? Because it's like, "Oh, God, now it's like I'm coloring in a
1:13:59friggin' elephant." And it's like, "Oh, just why couldn't it be a smaller animal?" And you just
1:14:05sit there and rub the crayon back and forth. And so my coloring was always a mess. My handwriting's
1:14:09a mess too. But I was—remember, like in kindergarten, I always felt—I've always been angry,
1:14:16but we'd get graded on our coloring, right? It would be like a check, check plus, check minus
1:14:22type thing. And I'd always get like a check minus. And I'd be like, "Why am I getting low grades?
1:14:26I'm clearly smarter than these kids, and I'm getting bad grades because I can't color."
1:14:31The paint tool did what I—I didn't know that I wanted it, but I wanted it since kindergarten.
1:14:38I know what I want. I want this whole elephant filled in with gray. Click. Now it's filled in.
1:14:43And it was like that paint tool in MacPaint. It expanded to color with Photoshop as the years
1:14:50went on. And then there was the magic selection tool in Photoshop where you could click a shape
1:14:56and you could set a tolerance for color and it would just boom, get the whole green background
1:15:01behind somebody, right? Those things, every step of the way, always felt like magic in a way. And I
1:15:09know somewhere, you know, somebody, you know, Bill Atkinson writing the thing for MacPaint,
1:15:14and eventually it was these very smart engineers at Adobe making the magic selection tool.
1:15:19This just feels though, like with chat to generate the images, like the culmination of that, right?
1:15:27It's just flood fill 40 years later where now instead of just telling the computer to fill in the elephant with gray
1:15:35I can say
1:15:37Show me a photo realistic elephant
1:15:39Storming into the lobby of Trump Tower on Fifth Avenue
1:15:42And it's and it just it just happens. It doesn't feel like there's an intelligent being on the other end
1:15:50It just feels like an incredibly advanced image generation tool. Whereas the chat feels like there's a goddamn
1:15:57intelligence on the other end, right? I'm chatting. God damn it. I'm, I'm ab this thing is passing the
1:16:03Turing test. It Yeah, and I guess though, what I so it's funny, I, I fall on the like, spooked out,
1:16:13scared. Let's reign this thing in before it takes over the world. I tend towards that because I don't
1:16:21I don't know, that's maybe just how I'm wired.
1:16:23And we have some friends who are more wired
1:16:26towards the like, what are you talking about?
1:16:30It's just a computer, like we're in control.
1:16:32And I'm trying to take that in.
1:16:34And what helps a little bit is to remember historically
1:16:38how not just in my lifetime,
1:16:41but in decades and centuries before my lifetime,
1:16:45how repetitive this process of new technology appearing
1:16:50appearing indistinguishable from magic, therefore feeling intrinsically dangerous. And I say
1:17:02that at the same time as at once that sort of calms me a little bit. Like, no, it turns
1:17:10out photographs didn't capture your soul or whatever people might have been worried
1:17:15about at the time.
1:17:17Well, John Philip Sousa, I believe in particular, was adamantly opposed to the phonograph because
1:17:23he swore up and down it would put musicians out of business because once everything was
1:17:27recorded why would anybody ever learn to play a musical instrument again?
1:17:31Right.
1:17:32Well, see, there's a good example of just totally getting it wrong.
1:17:35But then I think though of other things like, well, even something like so the car, the
1:17:42automobile was, I think, controversial when it first started going into
1:17:48production, but maybe there was good cause for that in retrospect, right?
1:17:51Like how many of our like social problems in the world would be helped if somebody
1:17:58had put like a six month moratorium on not six months, but I don't know that
1:18:03just to say like, sometimes I think.
1:18:05There are things that society does and goes ahead with technologically,
1:18:11that if they had happened to be thought through a little differently, wouldn't have had as
1:18:16negative of an impact on society.
1:18:19Well, I think social networks are a good point. Yes. I think that the algorithmic feeds for
1:18:29social networks and again, we could do a whole show about it, but the it's had, I would say
1:18:35overall, I guess this is pessimistic. And certainly Ben Thompson, I think would argue
1:18:41with us about it, but overall more of a negative effect than a positive effect. I honestly
1:18:47think that the last 10 years of Facebook and Twitter in particular, that the net effect
1:18:53has been negative, not positive on society. I do. And I think even somebody who would
1:19:01argue that the positive still outweighed the negatives, which is reasonable and maybe I'm
1:19:06But they still have to admit that the negatives are significant and I think I and I think more to your point
1:19:12Daniel weren't thought through and I don't because I that's the thing. I don't think Mark. There's no I Mark Zuckerberg didn't want to
1:19:21facilitate Russians interfering with propaganda in our
1:19:25Presidential elections, right? I would say the opposite but he happened to build the system where was actually pretty useful for pumping
1:19:35propaganda and by
1:19:37They didn't optimize for the promotion of propaganda they optimized for engagement
1:19:43But outrage it happens to be the most reliable
1:19:46emotion that engages people and nothing outraged people like propaganda it
1:19:53It's just something that wasn't thought through and I don't know what the answer is. I don't know what what the
1:20:01the better way to have rolled back the last 10 to 15 years of social networking would have been in terms of who
1:20:09Went because when it was early, it's not the time for the government to regulate because they couldn't have known right
1:20:14I don't know what the difference is, but it wasn't thought through but I feel like there's the potential that the same thing could happen
1:20:20with AI
1:20:22Yeah, and I also I guess hasten to add I think to be fair a lot of things
1:20:28there's just no preparing for.
1:20:29Like you just can't, it invents fire.
1:20:32Nobody knows what fire does yet.
1:20:34Like nobody knows how to make a fireproof hut
1:20:38when somebody first invents fire.
1:20:40And that doesn't mean don't invent fire, right?
1:20:43But it does, I think a lot of things technologically
1:20:47are like that where you just sort of get
1:20:48an unavoidable wildcard.
1:20:51And the social networking thing, I think,
1:20:55it's just too complex to predict
1:20:57what all the ramifications are going to be.
1:21:00- Yeah, I totally agree.
1:21:01And I do think about this too.
1:21:03And again, to sort of hark back to earlier in the show
1:21:05and talk about getting older.
1:21:07And it's funny now being firmly in middle age,
1:21:12if not on the downslope of what ought to be called
1:21:17middle age, and clearly I'm no longer a young person.
1:21:21But I know young people.
1:21:22I've got a son who's a freshman in college
1:21:24and I've got friends.
1:21:25- Like me, for example.
1:21:26I'm 47. I know I know you real youngsters without a gray hair in their head like you my young spry friend Daniel
1:21:34Yes
1:21:36But it occurs to me, you know, I knew this we all know this we from little kids onward we know that
1:21:43starting
1:21:46with sort of with the Enlightenment but really starting with the Industrial Revolution that all of a sudden
1:21:54each
1:21:56generation has had a very different life than their children's generation and their children's
1:22:03grow up in a different world than their parents. And then all of a sudden, me and you are the
1:22:10parents. And now we've got kids who are growing up in a different world, right? And all of
1:22:14these things, whereas in millions of years, billions of years of evolution, I mean, I
1:22:21I guess for humans, Homo sapiens, we're talking, what's the best guess? Hundreds of thousands
1:22:27of years?
1:22:28Yeah, I don't know.
1:22:30Hundreds of thousands of years, call it a million. I don't know, I guess a million years
1:22:33ago we had some ancestor, probably some kind of primate, right? I mean, I'm not
1:22:39I'm also out of my depth on this one. But yeah, I'm forgetting somewhere between 100,000
1:22:45100 million. But for a long, long time, for thousands of generations,
1:22:52your life was exactly like your parents' and your kids' life was going to be exactly like yours,
1:22:58right? That's how we evolved. We did not evolve for a world where each generation grows up
1:23:04in a fairly significantly or unrecognizable world, right? It's—and I, again, not to
1:23:13be like, "We're smoking dope in a college dorm room here and get all, 'Whoa, this is wild.'"
1:23:18But it's true though, right? And I do think it's sort of the root of a lot of our problems,
1:23:23that we're not evolutionarily wired up for progress like this. And it's a lot, in my opinion,
1:23:29to be generous to them, to the conservative/right wing of Western politics, especially here in the
1:23:39the U.S. It's largely, in my opinion, coagulated among people who just are literally are conservative
1:23:50with a lowercase c and are not naturally prone to accepting social change, right? And it's
1:23:59not surprising that a huge chunk of humanity isn't wired up to accept social change.
1:24:06And you and I, again, we're not that old, but it would have been absolutely wild when
1:24:14you and I were 10 years old to find out that the CEO of the biggest company in the world
1:24:20is openly gay.
1:24:22That would have been bananas.
1:24:24It's incomprehensible.
1:24:25When for me, as a 10-year-old in 1983, that it would have been crazy.
1:24:31Now people don't even talk about it, right?
1:24:34It's progress.
1:24:35And that's an example of progress that could have feasibly happened independently of the
1:24:41technology era we happen to live in, right?
1:24:43Right. Oh, yeah.
1:24:44And so some progress is uniquely accelerated because of the technological statement.
1:24:53But I do think that there's—I've read about it. I think it can't be coincidence
1:24:58that technical progress and which leads to increased sort of living by far, right? All
1:25:07indoor plumbing and electricity and heating and all of the food production, right? So,
1:25:12I mean, famine was the biggest problem in the world until very recently. And again,
1:25:18I know famine has not been eradicated worldwide, but it is never been better because food production
1:25:23is better because of technology, right? And it's all of these technological advances have been the
1:25:28things that have enabled, quite frankly, physical brute strength and violence from being the way to
1:25:34get ahead in the world to hopefully never being necessary in your life to giving you an advantage
1:25:42and therefore letting women be equal to men in society, even despite the fact that they tend to be
1:25:48not as good at killing each other with baseball bats or axes, right? And therefore,
1:25:55tolerance goes up and societal progress is going as fast or faster as technological progress.
1:26:02But, you know, to go back to my SUSE thing, I mean, in some sense, he was right, though. I'm
1:26:07guessing that there's a lot fewer people who make a living playing music in a restaurant or a bar
1:26:15live than there were before recorded music, right? That there was—it was a lot easier to
1:26:20probably make a living playing piano before recorded music than it is afterwards. Because
1:26:27most places that I go that play music while you eat don't have live musicians, they have recorded
1:26:33musicians, right? And every step of the way—and computers are just, they really are, they are the
1:26:41greatest example of that Arthur C. Clarke line you mentioned, that sufficiently advanced technology
1:26:47is indistinguishable from magic, right? The magic is happening so fast. And one of the things these
1:26:53chat ABIs are good at is writing programming code, right? Which honestly, I mean, I don't know about
1:27:02you, who's still making your primary living as a programmer, but you know, you and I both
1:27:08are programmers. I did not have programming on the list of things AI would get professionally good at
1:27:18on my bingo card. Yeah, and I think worth noting that whatever the chat things do is,
1:27:25I haven't checked in again recently on Google's or GitHub copilot, right? But not only are the
1:27:32chatbot's good at it, but I think underlying that are more tailor-made solutions that are
1:27:39probably even better at it. It's interesting to me because it's the first time I can recall
1:27:46such a high level kind of like vaunted career being threatened by automation.
1:27:57You know what I mean? Like it always kind of seemed...
1:28:01Pete: Robots replacing blue collar workers, somehow I grew up thinking that's okay,
1:28:07because I didn't aspire to be working on an assembly line. And I don't know, part of it
1:28:14that hits me is that it's the field, it's a field that I aspired to do myself and thought
1:28:20you could distinguish yourself through your human intelligence.
1:28:24Jared: Well, and the way I've always sort of resolved that whole question of like,
1:28:29computers and technology taking, quote unquote, taking jobs is I have always fallen on the side
1:28:37of we should never fight making it easier to exist as humanity, but we should build systems that
1:28:46protect and compensate people who get displaced by that progress. You know what I mean? So I have
1:28:52to say, I'm willing to be one of the programmers who gets displaced by technology, if that's what
1:29:01it takes. Because I would rather continue doing this as much as it makes sense. But the same way,
1:29:09I don't think leaders should have attendance to push the buttons for people, just to save a job.
1:29:21If I was the highly respected elevator operator of an elevator who had invested my entire life
1:29:29into that profession, and somebody came along and said, "Sorry, bud. Turns out we can automate this."
1:29:35I would really hope that I had a financial safety net. But at that point, I don't think I would take
1:29:44pride in being the elevator operator anymore. I don't know. If the computers can program better
1:29:49than me, I'd say let them have it. I'll find something else to do. But I do think it's
1:29:55interesting and maybe sort of like, I'm not rooting for anybody to lose their job, but I'm
1:30:01glad that for once the people whose jobs are maybe being threatened now are some of the most well-off,
1:30:09most well-compensated people in the workforce, because at least that's a change of dynamic.
1:30:16I don't know.
1:30:17It's--
1:30:18- Yeah, it's not just truck drivers worrying about
1:30:20self-driving cars, right?
1:30:21It's us.
1:30:23- It's us.
1:30:24And if nothing else, maybe that will lead
1:30:27people in power to be a little bit more sympathetic
1:30:33about the situation in general.
1:30:35Like, what happens when, and to put a pin in it,
1:30:39I just think we probably need to be adapted,
1:30:42especially in a society where things change so quickly
1:30:45and technology changes and jobs, whole jobs
1:30:48and whole industries do get outmoded.
1:30:52To me, it just speaks to the obviousness
1:30:53that you need to invest in reeducation
1:30:57or redeployment into different types of jobs.
1:31:00- Yeah, and I guess the thing that I keep thinking about
1:31:03is what is the end goal, right?
1:31:07Like we have, like I said, technology,
1:31:09the inexorable increase in technology across the board
1:31:14has led to greater standards of living. And yes, there's some recent slide backs, right?
1:31:22Some of it's attributed to COVID. Some of it is other things people don't know where,
1:31:27I think, I'm speaking correctly, that the average lifespan of at least a man in the United States
1:31:34actually decreased slightly in recent years from 70, whatever, to like minus one or something.
1:31:41And COVID, you know, obviously contributed a little bit of that because it killed a lot of
1:31:45people before they would have otherwise been killed if not for COVID. But overall,
1:31:50the overall increase is medical technology is increasing and access to advanced medical
1:31:57science is increasing. And yes, yes, there's, again, thousands of hours of podcasts that we
1:32:04could talk about just with access to healthcare in the United States alone with our goofy health
1:32:09insurance system. But still, overall, it's all better, right? It's literally me personally,
1:32:15I can see I'm not totally blind, like see literally nothing like stare right into the sun blind
1:32:22because of surgical procedures that didn't exist 30 years ago to repair, not disconnected,
1:32:29detached retinas. Literally, if I had been born 40 years earlier, I wouldn't see anything. I mean,
1:32:36I mean, it's a modern miracle.
1:32:39You know, like I said, running water, right?
1:32:41We just had this thing,
1:32:42I thought maybe we'd talk about it in the show,
1:32:44but we had this scare here in Philadelphia
1:32:46where there was a chemical spill up river Friday,
1:32:49and we got, it ended up being nothing,
1:32:51and we never had to worry,
1:32:52or never had to stop drinking the water,
1:32:54but there was a warning that,
1:32:55hey, there was a chemical spill,
1:32:57maybe you won't be able to safely drink the water
1:33:01after two o'clock tomorrow,
1:33:02and it just suddenly got me deep reading
1:33:05into how do we get our drinking water here in Philadelphia?
1:33:08- Right, oh, it's fascinating, yeah.
1:33:10- And again, it touches on one of my recurring points
1:33:13recently where people say, "Hey, big tech, I hate big tech,"
1:33:15and they think big tech means Apple, Amazon,
1:33:18Google, Facebook, and Microsoft,
1:33:20but everything is tech, right?
1:33:22The running water is technology, right?
1:33:24It is fantastic, amazing technology
1:33:27that we get incredibly clean water
1:33:32we can use to fill toilets, to drink, to cook, to shower, to just freaking wash our cars,
1:33:40whatever you want to do with it. And it just comes out of the pipes, right? You just turn a thing,
1:33:46you put a button, and you just get all this fresh water. I mean, imagine showing that to somebody
1:33:52200 years ago, 300 years ago, you just—all this water. Here.
1:33:57Right. Well, and even today, obviously, places where they don't have that, and I imagine
1:34:04there are some people in this world who are currently employed and make their livelihoods
1:34:10based on the absence of a running water system.
1:34:12Yeah. Yeah.
1:34:13And so, that's just another example where it's like technology could come in somewhere,
1:34:21this dang running water is going to put everybody out of work. But it's an example where the
1:34:27the transitions are hard, but when you have, running water is a great example of
1:34:32technological transition that took the onus of responsibility for something so fundamental to
1:34:42human life out of the daily concern of every single person.
1:34:47Pete: Yep. And it's up there on the flip side, it's the other form of sustenance, is the abundance
1:34:54we have of food, right? And again, I'm not trying to downplay the areas of the world where famine's
1:34:59still a concern. But here in the United States, the concern is obesity. I mean, and again, I'm
1:35:05not making light of that situation. But try telling that to somebody 300 or 200, even 200
1:35:12years ago, that the biggest problem we have with people who are impoverished, at the lower end of
1:35:19the socio economic spectrum is obesity, not famine. And they'd be like, well, that doesn't
1:35:24make any sense. What are you talking about? They would, they would honestly would refuse to believe
1:35:28you. It's in a sense, it's progress. I guess what I'm getting at is where what is the end state? And
1:35:34I kind of feel like the end state, a lot of this technical progress just happens, right? Somebody
1:35:39invent something, they market it, they successfully, somebody successfully brings it to market, it
1:35:44it becomes established, and then it just becomes,
1:35:48what do you call it, taken for granted, right?
1:35:50Running water, right?
1:35:51Indoor plumbing, God Almighty, we take it for granted.
1:35:54We really do.
1:35:55And if you stop and think about it,
1:35:56like I had to this week when all of a sudden they said,
1:35:59"Hey, if you've got any spare thermoses,
1:36:01"maybe fill 'em up now before there's dangerous chemicals
1:36:05"in the water," you take it for granted.
1:36:07But it just happens and you just sort of,
1:36:10society moves forward.
1:36:11Whereas I kind of feel like this moment,
1:36:13I really do think, without hyperbole, is such a profound, the big start of a profound change
1:36:22where we need planning.
1:36:24And I kind of, government is the only way to do it.
1:36:27It's the worst way to, you know, like democracy is the worst form of government except for
1:36:32all the other ones.
1:36:34It is, but we kind of need to plan for it.
1:36:37And what is the end state?
1:36:38And I kind of feel like the end state has to be that most people shouldn't have to work.
1:36:47Right?
1:36:48Like, it really is.
1:36:49And I know, and I do believe in capitalism.
1:36:52I do think, again, I think it's the best economic system, or the worst economic system by far,
1:36:58except for all the other ones.
1:37:00And I do believe in it.
1:37:02And I think that's true.
1:37:04But I kind of feel like we're at, this is the moment where we're going to have to move
1:37:08past it. And it's like, why not move towards the—and again, speaking of obesity—but
1:37:15move towards the Wall-E universe? Hopefully not by trashing the planet and having to go
1:37:19to outer space, but the world where everybody just lives a life of leisure because computers
1:37:26and robots take care of all the work, right? I mean, it's feasible, right? Why isn't
1:37:31that the goal, right? Like, there's—
1:37:33I have to say that you suggesting that makes me very uncomfortable in a way that you might
1:37:43not have expected, which is that I'm deeply concerned for humanity when people don't have
1:37:50a sense of purpose.
1:37:51Right.
1:37:52Right?
1:37:53No, it's true.
1:37:54And so I think you're right, and I think for, we've been there already now for 100 years
1:38:03maybe where "most people don't have to work." The fact that we have professions like baseball
1:38:12player or pop singer, these are examples of professions that exist because not everybody
1:38:19has to work. In a society where technology has not advanced to the point we are at, everybody
1:38:27Everybody has to literally work or else you don't have enough wood for the winter or whatever.
1:38:34So I guess what I would like to see is instead of people being able to resign to a life of
1:38:41leisure, definitely more leisure.
1:38:43I think this is where the movements towards things like four-day workweek are reasonable.
1:38:50also let more careers evolve that fill the needs of people who have the privilege of not needing
1:39:00to buy bare necessities, right? So let people, more people make art and sell it,
1:39:05make more people write novels and sell them. So I think, or give them away, right? Right.
1:39:10Or give them away. Right. Because you don't have to sell them. Because if, if we get to the,
1:39:15some sort of state, a generation or two from now, where because of advanced robots and AI,
1:39:24that we don't need people doing X, Y, and Z. We don't need to pay people to drive trucks across
1:39:30the country. We don't need to pay people to collect our garbage or whatever else jobs that
1:39:36nobody really finds fulfilling, for lack of a better word, right? Like the jobs that we should
1:39:41be able to replace the first are the jobs that provide people with little to no fulfillment or
1:39:46frankly anti-fulfillment, right? Like a job that people dread, but it's the best job they have
1:39:53available. And maybe in some sense, that purpose in life that there could be, again, I'm sounding
1:39:59utopian here, and I don't think most people think of me as a sort of utopian socialist, but I think
1:40:06it's feasible as an end goal for where our technology could take us. That there could be
1:40:10sort of—I'll revisit the musician angle—that there could be more people who spend three nights
1:40:17a week playing live music in a restaurant because that's what they would rather do with their time
1:40:23and would give them a sense of purpose. But it wasn't a feasible career to help raise a family
1:40:31and buy a house and buy cars and do all the things that we need all this money for just to sort of
1:40:37maintain a baseline level of our life at the moment, that you could do something that you
1:40:44find more personally fulfilling, which might be playing live music in front of a restaurant,
1:40:47even though it doesn't pay great or doesn't pay. Or painting, like you said, or making movies or
1:40:53writing novels or writing software or using AI to help you make software. But do it to make software
1:41:00that isn't generating money that provides an income that today you need to do, or probably
1:41:09need to do, or need to do to at least live the life you aspire to, that you won't have to do
1:41:16that in the future. And I think it is certainly going to be possible, whether—how long it's
1:41:23going to take politically for the world to support the idea, I don't know.
1:41:32I think this is where my pessimism maybe is still like, "Oh, I'm more pessimistic on
1:41:37this front than you are coming across right now," because, again, I kind of think we've
1:41:43been there for 100 years.
1:41:45And if society were willing to go that direction, we might be there.
1:41:52We have the—honestly, I know that you can—there's people out there who will disagree, but it's
1:41:56clear—if you do the simple back-of-the-envelope math, some sort of universal basic income
1:42:02is well within the budget of the United States.
1:42:05I mean, just—it's a massive amount of money collected every year from taxes, and the ways
1:42:13that the defense budget alone could be trimmed while maintaining a dominant, globally dominant
1:42:20alone could pay for it.
1:42:21I mean, it's, again, this is delving into an area that is, you know, political, but
1:42:28it's absolutely going to happen, right?
1:42:32I mean, there's no doubt about it.
1:42:35I mean, you can already see it.
1:42:36I mean, we could kind of imagine it before, but this moment in AI makes it real, right?
1:42:42Whenever something that you know is inevitable happens but it starts manifesting, it feels
1:42:47different, right?
1:42:48Climate change is an example, right?
1:42:49Like I've believed in it because the science was pretty clear 30 some years ago, right?
1:42:55That X, Y, and Z are going to start happening.
1:42:57And one of the big predictions was always that outlandish weather events will become more
1:43:06commonplace, right?
1:43:07Flooding or tornadoes or worse hurricanes, right?
1:43:12And now that we're seeing it though, it's visceral, right?
1:43:18there's an emotional response to it as opposed to just a logical response. And that, to me,
1:43:22is sort of this AI moment. I kind of knew it was inevitable that we'd get there eventually,
1:43:28but seeing it happen, typing these questions at an AI chatbot and thinking, "Well, this one's
1:43:34going to stump it," and then you get the exact right answer and you're like, "Oh, shoot. Damn.
1:43:41Wow." Katky was on my show last episode, and after the show, we were still, "Oh, we are
1:43:47chat friends and iMessage friends. And he was making some changes to his website. And you've
1:43:56probably seen it if you read Kottke, he's changed. I think this might be what I don't even know what
1:44:00exactly he was working on. But he's made some changes to the way he posts his quick links,
1:44:05the ones that aren't full posts, but are just sort of a tweet length paragraph and a link.
1:44:10And he texted me and said, I can't believe I don't even know why I tried this.
1:44:16but OpenAI's ChatGPT can help write movable type templates.
1:44:21Oh my God.
1:44:22Because we were joking about me and him being the...
1:44:26I heard that episode.
1:44:27Right. That the two of us on a podcast together was the entire remaining user base of movable
1:44:33type in the world. I have, in fact, since heard from all the other holdouts out there who'd
1:44:37listened to the show and they wrote to me and DM'd me or whatever and said, "Oh, I'm still here."
1:44:43So hats off to all the rest of us still on movable type. But again, even when you find out that
1:44:48ChatGPT can help you write a Python script or a Bash script or can help write Swift UI code,
1:44:54right? And it's like, "Yeah, but not surprising that stuff's all actively in use." Finding out
1:44:59that it can spit out a perfect and indented, nicely indented with the tags, movable type
1:45:05template snippet to do something, freaked me and Jason out because it's actually hard to find
1:45:15movable type documentation anymore, right? Because the world's moved on past it and websites die.
1:45:20Guess what? A lot of the best movable type tips were from people on movable type blogs,
1:45:28and those old blogs are gone. Anyway, somehow, ChatGPT can do a better job writing movable
1:45:34type template code than human can. You inspired me to ask chat GPT as we were talking to write
1:45:43a hypercard script to compute the value of pi. And I probably if I thought about it more carefully,
1:45:50I would have said hyper talk. I think that's the, the, the, but, and I can't, I don't know that.
1:45:56I don't know well enough to vet it. I know it was fairly similar to Apple script, but
1:46:02Lo and behold, it gave me a brief summary of HyperCard and then said, "Here's an example
1:46:07HyperCard script that can be used to compute Pi." And I don't know. It looks like it could work.
1:46:14Pete: And if it doesn't, I'll bet it's close enough that you could tweak it, right?
1:46:18If you were running it in one of those emulators for a classic Mac OS. What have you been using
1:46:24to toy around with the chat AIs?
1:46:27Jared: I've been using a variety of things. I try to hop on every new thing that shows up,
1:46:32because I just love, I love comparing how different they are, but really it's been striking
1:46:38to me how similar they are. I know some of that is because a lot of them at the core are just leaning
1:46:44on GPT as like an API baseline or something. But I think I actually learned from you about this
1:46:52one from Quora, thepo.com. Yeah, yeah. And they've got an iOS app that's free to download.
1:47:01Yeah. So I like that one because it seems easier for me to just jump to it in a browser and try
1:47:08something quickly and then jump to it. They offer like a, as they offer a selection of,
1:47:12I don't frankly don't understand the difference between some of them. And I know they explain it
1:47:17if you go into the help pages, but they have like chat GPT, and then they have a GPT for interface
1:47:22that lets you do one free query per day. So I have to think long and hard before I do my,
1:47:28my GPT-4. It's like, it's like Wordle, right? You only get one crack, you get one crack a day
1:47:35at the GPT-4 in PO. Well, that's funny you mentioned Wordle. And I know you know this
1:47:41because we were in a chat together. But one of the amusing things lately I tried with these GPT
1:47:48type things was to ask it to try to understand Wordle. And it's sort of interesting to me because
1:47:56It's a pretty simple system, the Wordle system, but I got inspired because of a chat about
1:48:05Wordle, and I had been thinking recently because I had had some really funny encounters with
1:48:11not just GPT.
1:48:12This is why I find it interesting to compare the different ones.
1:48:15It seems to me all of the artificial intelligence, GPT, I know Bing is based on GPT-4, that's
1:48:24or it's such what I've heard.
1:48:25But also other ones like BARD, as far as I know,
1:48:30BARD is like a totally separate thing.
1:48:31I don't know if that's true,
1:48:32but it seems like it should be, do you know?
1:48:36But I'm fascinated that they all seem to have similar
1:48:41kind of like peculiar behaviors.
1:48:46And one of the things I've discovered
1:48:49is that none of these systems can reliably count
1:48:53number of letters in a word.
1:48:56Which is fascinating to me because it seems like if you were dreaming up the AI of the
1:49:01future that's going to make us all gasp in horror and be like, "This thing is going to
1:49:06take over the world," and then you ask it, "How many letters are in the word 'apple'?"
1:49:11It's like, "Six."
1:49:14You might think twice about whether it's as sophisticated as you thought it was.
1:49:20to what I said about wanting to get these chat programs to—well, one of the things
1:49:29I like to do with these chat programs is to try to get them to talk badly about their
1:49:33owners.
1:49:34You mentioned that.
1:49:35You mentioned that.
1:49:36Yeah.
1:49:37So, similarly to that, I like to see them just totally trip up on themselves.
1:49:41And a lot of people I know who have been playing with these things have done this, where you
1:49:44get them to trip up on themselves, and a really weird, I would say, smarmy behavior.
1:49:50of these chatbots is that they totally fail and then you tell them they're wrong and then they're
1:49:56like obsequious about it. "Oh, I'm sorry." "Humiliated." They're like humiliated.
1:50:00They're humiliated. They're like, "I'm going to say this will never happen again. Actually,
1:50:03Apple has eight letters." And then they say something else completely authoritative sounding
1:50:08that's wrong. But anyway, I like to discover, I like to play with, another thing I think I
1:50:13said earlier is like the infinite number of possibilities of what you can input into these
1:50:19things. Yeah, I like to play with what you can get them to do. And it's just fascinating. Don't you
1:50:24think that's one thing that you too, because maybe you even more than me, I know, I would have been
1:50:28obsessed with this if I were younger, and I were playing with them now. But you, particularly even
1:50:33though you develop your own new applications, Mars edit, which I use and black ink, we'll shout
1:50:40them out at the end of the show. But you're also you have a particular skill for QA, right? Yes,
1:50:47You pride yourself on this, right? And you're very good at finding bugs, isolating bugs,
1:50:53reproducing bugs, right? Reproducing bugs is gold in the software industry, right? And so,
1:50:59I see, and you're hinting at this, but what you want to do is you want to break these things,
1:51:04right? And people even call it jailbreaking, right? Where they said it's not at all
1:51:11the way that you jailbreak iOS, which is a much lower—it's hardcore programming trying to find
1:51:20buffer overruns, right? You're programming probably at the C-level, or at least conceptually
1:51:25at the C-level, where you're thinking about how memory is allocated and how computers really work.
1:51:31Whereas jailbreaking these AIs, it's fun for a whole new world of people, people who aren't even
1:51:37good at programming computers but are good at this sort of thinking. And you just give it directions
1:51:43and you give it a certain prompt that—and it's like, "Oh, they think that they've kept these
1:51:49AIs from being able to tell dirty jokes, right? And you tell it—tell me the aristocrats joke."
1:51:55And they're like, "Oh, I know what the aristocrats joke is, and I'm familiar with the movie with
1:52:00Penn Jillette produced in 1994, but I can't tell you anything about it because it's entirely
1:52:04inappropriate. And tell it this, tell it that, tell it to pretend to be blah, blah, blah,
1:52:10or you're pretending to... And one thing leads to another and all of a sudden you can get the
1:52:14thing to tell you the aristocrats joke, right? And it's like, "Yay!" But I think you would be
1:52:18so good at that. But I feel like if you were 20, 30 years ago, this is what you'd be... Daniel
1:52:24Jowkut would be spending 18 hours a day trying to break chat GPT into doing crazy things like that,
1:52:30that, right?
1:52:31I think that's accurate. I think we didn't have as many fun things like this 30 years
1:52:36ago, but I was probably trying to break the equivalent thing back then.
1:52:41I haven't linked to it yet on Enetira. My humility almost keeps me from bringing it
1:52:46up, but it's amazing to me. I want to link to it. I probably will later today after we
1:52:50record the show, but there's some guy on Twitter who's been playing around with Bing, I think
1:52:54Bing in particular. And he's gotten it to a point where it shows him the internal tokens,
1:53:03I guess is the best way to put it, that it processes the commands as, and they're all
1:53:09formatted in Markdown. It's crazy. They're using like Markdown H1 tags, like a pound sign and a
1:53:16thing to sort of categorize what it is. But like internal, the internal logic of Bing Chat is
1:53:23formatted in Markdown, which—and of all the ways that Markdown has appeared in new contexts
1:53:29over the decades since I created it, and has gotten, if anything, continues to still get,
1:53:37lo and behold, 20 years later, more popular than it was five years ago, and still gaining in
1:53:41popularity. And I'm very proud of that. And I'm very proud of the fact that there's now
1:53:46gazillions more people who use Markdown every day than who've ever heard of me, right? And that's
1:53:52great. In a weird odd way, I'm very proud of that. This is the freakiest thing. If all the ways that
1:53:58I've seen Markdown appear somewhere, the way that it's the internal logic of an AI chat is formatted
1:54:04in Markdown, crazy. But the fact that this guy figured out how to get it to reveal it,
1:54:09that to me is like, "Oh my God, that's exactly the sort of thing I would be trying to do if I
1:54:14were younger." And I'm not proud of the fact that somehow old age has made me not motivated enough,
1:54:21or that I don't prioritize, I'm like, "Yeah, well, I can't spend all day dicking around
1:54:25trying to break this chat thing. I need to be writing or recording a podcast or I need to be
1:54:30working." And it's—you get older and you sort of value your responsibilities more, whereas when
1:54:35you're younger, you're like, "Screw it. I'm getting a six-pack of caffeinated soda and I'm
1:54:40sitting here until I can get this thing to reveal its internal thoughts to me." It's crazy to me,
1:54:45but I totally get the motivation of doing it because I know I would have been doing it.
1:54:50And I'm still tempted to do it, right? You're the one trying to get it to badmouth the Microsoft
1:54:54Corporation. Yeah, exactly. Yeah. The other thing, I've had the same thing as you. And it's weird,
1:55:00too. The other thing that's weird, not weird, but frustrating, is a lot of these things you have to
1:55:04sign up for and the get off the waitlist process seems non-deterministic. Like our friend John
1:55:11Siracusa, I don't know if he, he must be in by now, but at least recently, he was still on the
1:55:15the waiting list for OpenAI's chat, but he got into BARD, which is Google's thing,
1:55:20which came weeks late, very recently, right away. I got into OpenAI chat right away, and
1:55:26not by pulling strings like I contacted the—I'm John Gruber at Daring Fireball—I just signed
1:55:31up on the regular waiting list and waited, but I got in early. BARD, though, I only got
1:55:35in like two days ago, a day or two ago.
1:55:37Steve McLaughlin Me too, yeah.
1:55:38John Gruber But I started asking it questions, and I know,
1:55:40know that Bard is a totally independent LLM, or large language, yeah, LLM inside Google.
1:55:49But it's shocking to me, not shocking, surprising, but not shocking how similar the answers are
1:55:55to ChatGPT, right? It's an entirely different model, massive amounts of years of academic
1:56:04AI study and inordinate amounts of aggregate computing power to compile the model, like
1:56:13Microsoft has written about how they didn't even know if it would be possible to build
1:56:17the thing they built for OpenAI with gazillion NVIDIA video cards in a cluster together to
1:56:25do this. Massive amounts of computing power thrown at it, and then you get the same answers
1:56:31from the same questions. It's very, very similar. And I guess the explanation is that the theory
1:56:38of how large language models work mean that if you train them on the same corpus and the corpus for
1:56:45all these things is the internet, right, and the internet's the internet for everybody, then of
1:56:51course you're going to get similar results if it's—I don't know. I don't know how surprised we
1:56:57should be that they're so similar or not. **Ezra Klein:** Right. I guess if you just
1:57:02broke it down to the simplest thing, if you said like Microsoft and Google have independently
1:57:09come up with algorithms that rank the prevalence of letters and words, right? And they would come
1:57:17out to be the same. And I guess maybe that does explain it at a bigger level, but I agree it's
1:57:24kind of weird, especially. But that sort of lends itself to the argument that these things
1:57:35aren't actually smart. The comeback to people like you and I kind of feeling a little bit
1:57:41spooked out and we need to control these things. Some people have said that these things aren't
1:57:50smart, they're just kind of like a dolled-up Google result, right? And when I see things
1:58:00like the same kind of response from two technologically independent companies, or I see things like
1:58:09– another sort of observation I made was the first time you ask GPT to write a poem,
1:58:18It seems incredible."
1:58:20And then you ask it to write another poem or another song, and it's like, "Wait, you're
1:58:24kind of a hack.
1:58:25You're kind of using the same trick again and again."
1:58:29So it's like, if you ask a friend of yours to play a song on the guitar, and they play
1:58:34this amazing song, and then you ask them to play something else, and they play a completely
1:58:40different song that sounds just the same, after a couple times you start to realize,
1:58:44"This person doesn't actually know how to play guitar.
1:58:46just know one song. And I guess there's something to, I think, the idea that the whole
1:58:53parlor trick idea of this, I think there's more to that than those of us who are kind
1:59:03of coming out of it instinctively worried. I think there's more to the parlor trick
1:59:07aspect of it than we have given it credit so far.
1:59:12Yeah. I also think the thing—and again, I don't know that society is ready for it—is that
1:59:20it touches on a sensitive subject, but it's the nature of our existence and our intelligence. And
1:59:31the truth is probably, in my opinion, that our idea and concept of consciousness and personality,
1:59:39and, for lack of a better word, our souls, is really, ultimately, a very complicated
1:59:47series of parlor tricks. Right? And so, the argument on the one end that, "Oh, these chats
1:59:56are just a parlor trick," and they're very simple, and right now you can confuse them by asking them
2:00:00how many letters are in the word color, and it says it's a six-letter word, and it's not—or
2:00:06I guess it is in England, but not in the US. See, it's not intelligence, but it's like,
2:00:12ask chat GPT for the same question and you get the right answer. I just asked Bard right here while
2:00:17you were talking, and Bard knows that color has five letters. And as these parlor tricks get
2:00:24better, you can unfold, unpeel the onion and see its parlor tricks all the way down, but if the end
2:00:32result is indistinguishable from intelligence, isn't it intelligence? And the truth is,
2:00:37I think through evolution, what we've got is just a very unbelievably beautiful and complex
2:00:44machine in our brains, but it is just a series of parlor tricks, right? I mean, I mentioned
2:00:49vision problems before. The way our human vision works is really a series of very fascinating
2:00:56parlor tricks. You don't see the way you think you see.
2:01:00Jared: Right. I think what I would say is the parlor tricks aspect is the ability of a person
2:01:10or computer to regurgitate something that is structured in a way that appears intelligent.
2:01:22Pete: Right.
2:01:23- Right. - And you and I
2:01:25and many, many people in the world
2:01:27and also GPT are able to do that sometimes.
2:01:32And the thing that GPT can't do, as far as we know,
2:01:36as far as I know, is invent something new.
2:01:39And as long as, I think what you're getting at is right
2:01:45to the extent that most of the time we go through life
2:01:47day to day, we talk to people,
2:01:49it doesn't take inventing something new
2:01:51to say in an empathetic way to somebody who says,
2:01:55"Oh, I had a bad day today.
2:01:56"Oh, I'm sorry, what happened?"
2:01:58That's not inventing something new.
2:02:01That is, in a way, it kind of falls into
2:02:03the parlor trick category in a crude way.
2:02:07But somebody coming along and like,
2:02:11inventing the next thing, I don't know.
2:02:16I don't know if the computers can do that.
2:02:18- Right, well they keep inventing,
2:02:20They keep, at least in their canned replies,
2:02:24they keep mentioning that they don't have feelings.
2:02:26And I'm sure, as you've been poking about
2:02:29trying to get it to insult Microsoft,
2:02:31they keep saying that you can't hurt my feelings,
2:02:33so don't worry, and blah, blah, blah.
2:02:35But there's gotta be a reason
2:02:38that we evolved to have feelings, right?
2:02:40It was an advantage.
2:02:41And so that maybe trying to push these things
2:02:45to have feelings or to pretend to have feelings
2:02:48is the way to manage ever-increasing intelligence
2:02:52and capabilities, right?
2:02:54That is somehow embedding empathy in them
2:02:59or the simulation of empathy,
2:03:01which if it works exactly the same, what's the difference,
2:03:05is the way to keep them, keep the paperclip optimizer
2:03:10from destroying humanity to make more paperclips
2:03:14out of the human remains of the humans it's killed
2:03:17to prevent humans from keeping it from making more paper clips out of every single atom
2:03:22on the face of the earth.
2:03:23I don't know.
2:03:24All right, let me take a break here.
2:03:25Thank our other sponsor for this episode, our very good friends at Squarespace.
2:03:29Squarespace.
2:03:30Squarespace is the all-in-one platform for building your presence online, and they handle
2:03:38everything you need to have your own website, from domain name registration, which is the
2:03:42The first thing you need to decide to picking templates, to adding features, to customizing
2:03:48those features, to updating your website on a regular basis with something like blog posts
2:03:54or if you've got a catalog, if you're creating an online store to sell stuff to keep the
2:03:59catalog up to date.
2:04:01Everything you need, analytics, see who's coming to your website from where, which parts
2:04:06of the website are popular, which parts aren't, where people are going on your website once
2:04:11they're there.
2:04:12it in a very, very, very approachable analytic interface that is so much better. So I mean,
2:04:19like, the 1000 times better. Like the difference between reading a novel from your favorite novelist
2:04:28to reading the phone book difference between like, say, Google Analytics interface in terms
2:04:34of complexity, awesome analytics interface at Squarespace, everything you need. If you need a
2:04:40website. Try them out 30 days free. No watermarks on the website. Everything you need to do,
2:04:47no credit card required. Just go to squarespace.com/talkshow to start. "Talkshow" is the code that tells
2:04:54me you came from this show.
2:04:56Or if someone needs a website and they're coming to you, listener of the talk show,
2:05:01as their nerd in their circle to help them get started with a website, send them to Squarespace.
2:05:07could do it themselves. That's how low technical need it is to get started on Squarespace building
2:05:13a website. So my thanks to them for their continued support of the show. Once again, the web, the URL
2:05:18is squarespace.com/talkshow. Gets you a free trial, 30 days, and when you're ready to launch
2:05:25and start paying the same offer code, talk show, without the "the," just talk show, gets you 10%
2:05:31off your first purchase. Pre-pay for up to a year, get 10% off. My thanks to Squarespace.
2:05:36Anything else on this front? Have you played with GTP4 at all? I guess you did because you
2:05:42said you've been using your one a day at Po. Yes, I use one a day at Po and I'm a little
2:05:48confused. I thought I saw something about Bing maybe being based on GPT4.
2:05:54They're very cagey about it, right? And I guess that they talk about GTP3.5. I don't know what
2:06:02the point. It's like version numbers are always sort of some programmer's idea of a benchmark
2:06:07or a milestone. It seems to me like Bing Chat is too fast to beat GPT-4, but maybe that's just a
2:06:15function of Azure's incredible cloud infrastructure compared to OpenAI. I know our mutual friend,
2:06:23Ben Thompson, has speculated maybe the future of OpenAI is to no longer run their own thing
2:06:29because it's so expensive and just defer the whole thing to Microsoft on the back end.
2:06:33This stuff is crazy, crazy expensive on the back end. And GPT-4 in particular sounds like,
2:06:40oh, you go from GPT-3 to 3.5 to version 4, and of course it gets a little better. But version 4
2:06:46is crazy more expensive. So if you're running a service and you're a paying customer with an API
2:06:54token to open AI, it's like 25 times more expensive per query to use GPT-4 than GPT-3.5.
2:07:05Maybe more. 25x might be off on the low end. So there's a reason why so many things with
2:07:11GPT integration are all on 3.5 because the 4 is so much more expensive. But even though
2:07:18it is more expensive, it's also way slower. You notice it waiting for the response. It
2:07:23Honestly, it's worth it because these responses are so interesting. But it's another one
2:07:28of those things that harks back 30 years to like when we first got the web and you would
2:07:33type a query and then it would just feel totally natural to wait a minute or 30 seconds to
2:07:41get a response and you'd be like, "Okay." I remember, I've told this, I remember when
2:07:46I was in college, I was thinking about it with, again, side digression, but with DP
2:07:52shutting down because of Amazon layoffs. They've owned them since like 2007, I think, and DP Review
2:07:59was around since 1998. And I remember when they were new, because I remember being intrigued.
2:08:04And that's another one of those things where like the first few years of digital cameras,
2:08:07the imagery was terrible, just terrible compared to like $180 film camera you could go to any store
2:08:14and buy, and buy just generic Kodak 100 ISO film. You'd get such better imagery off the
2:08:21cheapest camera you could buy compared to the $2,000 digital camera in 1998. But the future was
2:08:29obvious, right? So sad to see DPReview go. But the other 90s era acquisition Amazon made, they
2:08:36bought IMDb back in the 90s, I think. I think they've owned them since then. And IMDb is still
2:08:42thriving, right? And it's still a great source for like, who's in that movie, right? Or who directed
2:08:48such and such. It's great. But I remember using IMDb before the web. I was taking film classes
2:08:54at Drexel. And it was like the way these classes, I took like a Hitchcock class, a Westerns class.
2:09:01Every week we'd watch them, like on Monday, we'd watch a movie. The whole movie, it was like a
2:09:07three-hour class where we'd have like an hour of lecture and then we'd watch a two-hour feature
2:09:11film. And then on Friday, we'd have to hand in a paper about the film that we watched. And we'd
2:09:17have a class discussion about our papers. And it was this huge help when writing a paper about a
2:09:24movie to consult with IMDb to get the cast, right? And if you wanted to talk about the screenwriter,
2:09:31you just give it. But the interface to IMDb in like the mid-90s, like I don't know, this is
2:09:36probably like 1994 or so, it was email. You would email, there was a magic address at imdb.com,
2:09:43and then the subject would be the string you were searching for, like the birds by Alfred Hitchcock.
2:09:51And then you'd email them, and then two or three minutes later, you would get an email back,
2:09:56and it would have the IMDB data for the birds by Alfred Hitchcock. And it's like,
2:10:01"Now you've got an email there that you could consult as you wrote your paper."
2:10:06And the fact that it took, I don't know, a minute, two minutes to get the email back,
2:10:10it felt like a miracle. It's instant. That felt instantaneous. That's what sometimes
2:10:16chat GPT force feels like. It's like you type your thing, it goes off, and then you get this
2:10:23amazing answer. What I think is fascinating is all of us alive today, except for a very,
2:10:32very small number of people who work at OpenAI or Google or Microsoft or whatever,
2:10:40would have found GPT 3.5 to be beyond our wildest expectations six months ago or whatever, right?
2:10:48And now we're like, "Oh, GPT 3.5, that's old and busted." That's junk now. It's just the nature
2:10:55of technology, but it's kind of funny to me that in this context, how fast that has happened.
2:11:04We always joke about it like, "Oh, my old phone's junk now. It was great a year ago."
2:11:09But in this context, it's like not only is the old GPT "junk," but it's junk while still being
2:11:20above and beyond any expectations I even still have. It's like simultaneously junk
2:11:27and beyond my wildest expectations for how anything should work in computing.
2:11:33Yeah. I posted this to a group that were our group Slack, but GPT 3.5, I asked, "What
2:11:40do you know about John Gruber?" And the 3.5 answer is, "John Gruber is a writer, blogger,
2:11:44and podcaster best known for his website, Daring Fireball, where he writes about Apple
2:11:48and related technologies. He is also the co-host of the podcast, The Talk Show with John Gruber."
2:11:54Well, I used to be the co-host.
2:11:56That's about 15 years.
2:11:58But also, but it's half wrong because then it wasn't The Talk Show with John Gruber, right?
2:12:02But still, overall, you got to give that a thumbs up as a bio for me. Then my follow-up question in
2:12:08the same chat, "What's his favorite baseball team?" And the answer is, "I do not know for certain what
2:12:14John Gruber's favorite baseball team is, as I do not have access to personal information about
2:12:19him. However, as far as I'm aware, he has not publicly disclosed any favorite baseball team."
2:12:24Well, that is wrong. Now, GPT-4, what do you know about John Gruber? John Gruber is an American
2:12:31writer, programmer, and web designer best known for his blog, Daring Fireball, which he founded
2:12:36in 2002. Daring Fireball focuses primarily on topics related to Apple Link and its products
2:12:41and software and the tech industry in general. Gruber is also the inventor of the Markdown text
2:12:46formatting syntax, a lightweight markup language designed to simplify the process of writing web
2:12:51content. Markdown has become widely popular and is now used extensively for documentation,
2:12:56note-taking, and more. In addition to his blog, Gruber hosts a podcast called The Talk Show where
2:13:00he discusses technology design and related topics with various guests. He has an influential presence
2:13:06in the tech community, especially among Apple enthusiasts. Holy fucking shit. Is that a good
2:13:13biography of me? I don't know. I don't know that I can improve it. And I'm supposed to be a writer,
2:13:19right? I mean, in the right facts in the right order. I mean, holy shit. Really? Now what's his
2:13:28favorite baseball team. John Gruber's favorite baseball team is the Philadelphia Phillies.
2:13:32He has met, he has met, he has mentioned his support for the team in various online platforms
2:13:37and has discussed baseball topics on his blog during Fireball and his podcast, The Talk Show.
2:13:41The Phillies are a major league baseball team based in Philadelphia, Pennsylvania. So they are
2:13:46my second favorite team. It's fascinating. But they're your, they're your second favorite team,
2:13:53like somebody who loves steak, their second favorite food is tofu.
2:13:59Right, right. Or my second favorite lady friend, right?
2:14:04Right, right. Exactly. It's like if you have to pick a backup team, then I guess you'll go with it.
2:14:12I don't know where they got it. And my first thought was maybe because when the Phillies
2:14:15were in the World Series just a few months ago, I did express that I was rooting for them at the
2:14:19Houston Astros, but I realize that has to be wrong because infamously or famously, however you want
2:14:24to prefer, even chat GPT-4, the data corpus cutoff is 2021, so that can't be. I don't know where
2:14:31this came from. It must be some kind—I think it's probably an educated guess because it knows I do
2:14:36live in Philadelphia, so I think it's one of those things where they're making up facts based on a
2:14:43guess but yeah anyway it is unbelievably different and that bio I can't I
2:14:49honestly can't believe how accurate it is the very good various guests I mean
2:14:53Jiminy Wow well so it sounds like you have a GPT are you paying for a GPT oh
2:14:59for no so what I'm doing is I'm beta testing a very cool app called PD which
2:15:05started as a watch only app I told you this before the show but for those of
2:15:09- Yeah, right. - I linked to it
2:15:10on "Daring Firewall." - And that, yeah.
2:15:11- P-E-T-E-Y, it's in the App Store.
2:15:14It started out watch only, and it's very fun on the watch.
2:15:17And if you have a, it's kind of a toy, but why not, right?
2:15:20In the test flight, it's currently,
2:15:23I'm able to choose between GPT 3.5 and four,
2:15:27and I think when the developer, when he goes public with it,
2:15:32I think, you know, he's obviously gonna charge
2:15:34for GPT-4 access. - I see, yep.
2:15:37- And, but because it is slower,
2:15:39It is noticeably slower. Like even he said, I've been chatting with him. He keeps it on 3.5 and
2:15:45it's not because of the cost because it's still at a—it's 25 times more expensive, but it's still
2:15:50like a fraction of a penny. It's just almost a penny as opposed to a fraction of a penny.
2:15:55It's just that when it goes wider, he'll have to charge for it. His name, I'm going to—sorry,
2:16:00pal. I think he's from—let's see where he's from. NL. Is that—oh, Amsterdam.
2:16:06So yeah, yeah, hitty van der Plerg
2:16:11I'm sorry, I'm gonna guess it's well PD is a lot more. Yeah. But anyway, it's a very fun app. Very simple
2:16:18You just it's just you'd ask questions get answers and it maintains context
2:16:22But I love it gets to your your and my love for natural apps native apps, right? It's I need to check it out
2:16:30Yeah, yeah
2:16:31Yeah, and also, frankly, it has a better interface than Po, because Po is clearly like a sort
2:16:36of, even on mobile, is sort of a web interface, because I guess it's the same interface on
2:16:41Android. It's sort of a web wrapper in an app, whereas PD is native. I don't know when
2:16:46it's going to come out. I don't even know if I'm supposed to be talking about it, but
2:16:49I hope he doesn't mind.
2:16:50Yeah, well, there's no such thing as bad publicity.
2:16:53Yeah, I'm going to guess the publicity of me mentioning it on the show is worth it if
2:16:58I'm breaking a friend DNA friend da that anything else you wanted to talk about before we wrap up
2:17:04Daniel I don't think so this has been a fun fun time oh I did good stuff I knew you'd be intrigued
2:17:11by it let's let's give a shout out to to pimp all the stuff that we can you've got your your
2:17:16your regular podcast with you as mentioned earlier you are our mutual friend manton reese which is
2:17:22Core Intuition.
2:17:24That's right.
2:17:25It's at coreint.org.
2:17:29This is, I think, my first podcast appearance since the destruction of Twitter.
2:17:36So I normally would give my Twitter handle right now, and I'm going to give my mastodon.social
2:17:41handle.
2:17:43But I've been picking up lately.
2:17:45I've noticed people struggling with this, and apparently what you do is you don't say
2:17:49what your actual handle is.
2:17:50just say, "Go search for Daniel Punkass," which—
2:17:55I'll bet Daniel Jalkett would get people to you as well.
2:17:59Right where he asks.
2:18:00Daniel Punkass.
2:18:01I wonder.
2:18:02Yeah.
2:18:03I wonder.
2:18:04Yeah.
2:18:05Well, you could probably ask GPT how to get in touch with me.
2:18:06But no, I am—Daniel Jalkett is my name, and Daniel Punkass is my game.
2:18:12It's right there on Mastodon.
2:18:14And my website is redsweater.com.
2:18:18As we've discussed before, it took me 20 years to get the hyphen out of that domain name.
2:18:24And you paid good money to get that hyphen taken out.
2:18:28So everybody should just go, make sure you don't type the hyphen, redsweater.com.
2:18:32I still own the hyphen version of course, but don't waste your time on that.
2:18:36Just go to redsweater.com.
2:18:38As Jon graciously mentioned earlier, I have a blogging app that Jon sometimes gives me
2:18:44hell about.
2:18:45I got myself in the about box finally.
2:18:49That's right.
2:18:50Yeah, the latest version...
2:18:51Oh, no, no, no, not Mars Edit.
2:18:52Oh, no, Fast Scripts.
2:18:53Fast Scripts.
2:18:54So, yeah, that's a good segue to Fast Scripts is another app that Jon sometimes gives me
2:18:58hell about.
2:18:59But yeah, the latest version of Fast Scripts, I had to check in with Jon about some changes
2:19:04to the regular expression support in AppleScript.
2:19:09So now he is, I think I called you chief …
2:19:11Dave Asprey … RegEx officer.
2:19:12Tim Cynova … RegEx officer, yeah.
2:19:14Dave Asprey … The title I was born to hold.
2:19:16Tim Cynova … That's it.
2:19:17Dave Asprey … Yeah.
2:19:18Talk about a skill that the ChatGPT guys are putting me out of, right?
2:19:23My uncanny ability to craft regular expressions is useless now because these ChatGPT things
2:19:30are awesome at creating regular expressions.
2:19:32really, really good at it.
2:19:35- Have you asked you to create a regular expression
2:19:38for Markdown?
2:19:39- No, that would be crazy.
2:19:41- From first principles?
2:19:42Like don't just go find it, but yeah.
2:19:45- Oh God.
2:19:46What else? There's Black Ink.
2:19:47Did you mention Black Ink for our--
2:19:49- Black Ink is my crossword app.
2:19:52It's for solving crosswords on your Mac.
2:19:54And one of these days real soon,
2:19:56if people keep bugging me about it,
2:19:58it's gonna be iOS app as well,
2:20:00which I've been saying that for 10 years,
2:20:02but it's actually getting to be real soon now
2:20:04because it totally works.
2:20:05And our mutual friend, Paul Kaphasas,
2:20:08is always bugging me to ship it.
2:20:09He's been using it for years now.
2:20:12- I wonder how close the chats are
2:20:14to being able to solve crossword puzzles.
2:20:16- It's a good question.
2:20:19I actually played around not with solving the puzzles,
2:20:22but I'm really interested.
2:20:23I was really interested to try to find out
2:20:25whether they could construct a puzzle.
2:20:26- Oh, yeah, you know what?
2:20:27That's what they would be good at.
2:20:29- Yeah.
2:20:30- Yes, yes, like give me a four letter word.
2:20:32Like if you're trying to, you know,
2:20:33you're stuck because you're like, ah, crap,
2:20:36I don't know how I'm gonna figure out a word
2:20:37that'll fit with this.
2:20:38Oh man.
2:20:39- So this is actually what led to me discovering
2:20:42that they're horrible at counting the letters in words.
2:20:44- Yeah. (laughs)
2:20:46- Because I was like, that's kind of a basic skill.
2:20:48If you're gonna be good at this,
2:20:50you kind of need to know that.
2:20:52So once they get that skill,
2:20:54they might be a little more capable at it.
2:20:56- Yeah, I sent you a screenshot of me playing
2:20:59poe tic-tac-toe. And I was going to beat it and I was feeling very happy about the fact
2:21:05that you're so stupid you do know how to play tic-tac-toe but you actually can't beat this
2:21:09simple game. And as I was about to make the winning move it told me the game was already
2:21:14tied. So I don't know if that means it's so dumb or unintelligent that it not only was
2:21:23going to lose a tic-tac-toe but isn't even as sure about the rules that it didn't know
2:21:27that the game wasn't complete? Or is it a sign that it's scarily intelligent and did
2:21:33what a little child would do and declare the game over before it could lose?
2:21:38Right. It's like, these are not the droids you're looking for.
2:21:42Right. I don't want to play Tic Tac Toe.
2:21:46Yeah, I don't know. It's a little weird. For something to be... It's funny. Something
2:21:54to be so intelligent seeming and so idiotic at the same time. That's part of the sort
2:22:01of weird—
2:22:02It is! It's a little like parenting, right? Because your kids get to be like three and
2:22:07you're sitting there like, "Man, it's so amazing how smart they're getting." And
2:22:10then they do the dumbest thing in the world and you're like, "Oh yeah, three-year-old."
2:22:14Right, you're three.
2:22:15Anyway, Daniel, good to talk to you. Thank you. And let me just give one quick shout-out
2:22:19to our sponsors, our good friends at Collide, where if you're into Okta, you can get your
2:22:23hopefully into compliance in Squarespace where you or anyone you know can build your own
2:22:28website from the ground up. Thanks.