The Talk Show

00:00:00 ◼   ► So we're recording here on Friday, June 29, and I've got Matthew Pansarino from TechCrunch

00:00:06 ◼   ► here who just dropped a huge exclusive this morning on the new maps coming to Apple Maps.

00:00:15 ◼   ► And reading your story, I don't know if this was a problem for you or it would have been

00:00:19 ◼   ► for me, is writing the whole story is knowing when to use capital M for the actual product

00:00:26 ◼   ► maps and when to use lowercase m map to talk about the actual map within the app maps.

00:00:33 ◼   ► That would have driven me nuts while I was writing it. And I think you got them all right,

00:00:44 ◼   ► but it's tricky.

00:00:46 ◼   ► It is. And there are some points at which I used lowercase maps on purpose. There's

00:00:52 ◼   ► 101 different maps in the piece. So the 101 different times I say maps. So getting them

00:01:03 ◼   ► all right was a little hairy from time to time. But yeah, some sentences are literally

00:01:09 ◼   ► like maps and then the other maps is three words away, you know? So you have to be really

00:01:16 ◼   ► cautious about it.

00:01:17 ◼   ► - I don't know what else.

00:01:19 ◼   ► Somebody on Twitter, I think it was Neven Mergin

00:01:21 ◼   ► the other day pointed out that on iOS,

00:01:23 ◼   ► the apps that Apple ships with iOS,

00:01:25 ◼   ► the only ones that have interesting names

00:01:30 ◼   ► or original names are Safari, iTunes.

00:01:33 ◼   ► - Oh, right, mm-hmm.

00:01:35 ◼   ► - I forget what else.

00:01:36 ◼   ► But like he said, if they launched Safari today,

00:01:39 ◼   ► they would just call it web.

00:01:41 ◼   ► And I don't know if that's true or not.

00:01:42 ◼   ► I don't know if that's true,

00:01:43 ◼   ► but it's certainly true for all the other ones

00:01:46 ◼   ► where it's just mail, notes, maps.

00:01:49 ◼   ► And I'm not saying that maps should have a cutesy name

00:01:52 ◼   ► like Safari, but it certainly makes writing

00:01:54 ◼   ► about the difference between the actual map data

00:01:57 ◼   ► and the app difficult.

00:01:59 ◼   ► - Yeah, I don't know where their naming

00:02:05 ◼   ► kind of took a left turn into bland land.

00:02:10 ◼   ► I think it was sort of indicative of this fact

00:02:12 ◼   ► that they wanted everything to be universal

00:02:15 ◼   ► everything to be to feel like, you know, if they're building out these native apps and giving them the

00:02:20 ◼   ► the remit to own that space on the phone, that they should be named that way. And I think Safari,

00:02:27 ◼   ► of course, was just like, hey, if we're going to put a web browser on this phone, which is

00:02:31 ◼   ► essentially Safari that we've just, you know, slightly reframed to work on this phone, we'll

00:02:37 ◼   ► just call it Safari.

00:02:38 ◼   ► All right. So the big story is the lowercase M maps in Apple Maps are, Apple has been at work on

00:02:48 ◼   ► their own home, completely homegrown next generation level of maps for four years,

00:02:55 ◼   ► and they're going to start rolling out next week in for San Francisco, the city of San Francisco

00:03:02 ◼   ► for iOS 12 beta users. And then by fall, quote unquote, Northern California for everyone.

00:03:08 ◼   ► Trenton Larkin Yeah, technically,

00:03:10 ◼   ► Bay Area for many users. So you'll get San Francisco, you'll get Cupertino, San Jose,

00:03:17 ◼   ► you know, some of the East Bay, that sort of thing. But it's all up all up there. Home base,

00:03:23 ◼   ► obviously, for Apple. All right.

00:03:30 ◼   ► And it's I, well, how would you describe so the idea, part of the story here is that this is entirely Apple's homegrown mapping solution. So what what have we been seeing for the last seven years in Apple, right?

00:03:44 ◼   ► Yes. So I mean, I think there's two, two big things. One, you've got Yes, what is the difference

00:03:49 ◼   ► between the two? And then to, you know, exactly what is Apple building itself and not even

00:03:55 ◼   ► in the new maps. And so the the maps that we've been seeing so far, especially the ones

00:04:01 ◼   ► that we saw at launch, were really a, an Apple front end or face, you know, whatever you

00:04:08 ◼   ► you want to call it the UI of maps, but powered by data from partners. A large amount of that data

00:04:16 ◼   ► came from companies like TomTom and OpenStreetMaps. OpenStreetMaps being a sort of UGC

00:04:24 ◼   ► collaborative mapping effort. And then TomTom being of course, one of the largest commercial

00:04:28 ◼   ► providers of mapping data in the world. They're, I don't get, I don't know, they're, I think

00:04:36 ◼   ► they're a German company? That's probably gonna be wrong. Somebody's gonna correct me. But anyhow,

00:04:40 ◼   ► they're not a US company. They've been building maps for a long time. They've been using a lot of

00:04:46 ◼   ► things most famously, I think, for those of us who are old, in their own consumer GPS devices. You

00:04:53 ◼   ► know, the TomTom devices were some of the first, you know, sort of add on GPS devices you could

00:04:58 ◼   ► put in your car before Navi systems were, you know, came with every Prius or whatever.

00:05:05 ◼   ► So that data formed what they called a base map. And so in mapping, you have a structure

00:05:13 ◼   ► that is largely based on layers. And obviously, most famously, you've got like a 2D layer

00:05:20 ◼   ► and then like a satellite layer. And then Apple has added a 3D layer with Flyover. And

00:05:27 ◼   ► Google has done much the same over the years. But those layers of maps that comes from traditional

00:05:32 ◼   ► map making as well, where they'd have overlays that kind of showed different types of information.

00:05:37 ◼   ► But the base, the base map is sort of the most important because it's what you build the layers

00:05:44 ◼   ► on top of to represent or to to modify. And so the base maps were always external data,

00:05:51 ◼   ► third party data not gathered or collated directly by Apple, but instead provided by

00:05:56 ◼   ► partners, updated by partners, maintained by partners, and then, of course, licensed

00:06:02 ◼   ► by Apple. So there were a bunch of limitations there. One big major limitation is if you

00:06:08 ◼   ► need to make an update or correction to a map that has to go through a loop where a

00:06:13 ◼   ► correction gets noted, either internally at Apple or of course from, you know, Apple's

00:06:18 ◼   ► eventually launched external reporting tool. You can go in there and hit report, "Hey,

00:06:25 ◼   ► this location is wrong," or whatever in your copy of maps. All of those reports had to

00:06:31 ◼   ► then go out to the third party, be corrected by the third party, and then updated in the

00:06:36 ◼   ► data that it delivered to Apple. And then some of that data I'm sure is delivered live

00:06:41 ◼   ► and some of it is sort of static, like in other words hosted by Apple and they just

00:06:46 ◼   ► update it however often the third party provider provides them with a batch update. So you have a

00:06:51 ◼   ► long loop between the time that something is known to be wrong and the time it is in fact fixed or

00:06:58 ◼   ► corrected in maps for the consumer. There's a fairly long loop there that could last could be

00:07:04 ◼   ► months or even longer sometimes. So one of the that's one of the major problems that Apple needed

00:07:10 ◼   ► solve, to fix, so to speak, by building its own base maps and owning those base layer

00:07:15 ◼   ► maps is that the loop was shortened by essentially an entire company, right? So now it's within

00:07:22 ◼   ► Apple, those changes can be made aggressively, more frequently, at a more rapid pace. And

00:07:33 ◼   ► And then the second part of the equation is you have external licensees, so you have limitations

00:07:40 ◼   ► as to what you can do with that data.

00:07:42 ◼   ► So if you have, for instance, Apple Maps doesn't have local caching, and some of that has to

00:07:47 ◼   ► do with licensing agreements.

00:07:48 ◼   ► You know, the maps data can't be stored locally, it has to be streamed, etc.

00:07:53 ◼   ► So anyhow, there were very solid limitations that caused Apple to come to this realization

00:08:01 ◼   ► that if maps, which is a cornerstone of a lot of different functions of the phone, not just maps,

00:08:08 ◼   ► if they're going to own that, then they should own that. And that's why they decided to start

00:08:13 ◼   ► building their own base maps, and then adding their own or their own remixes of layers on top of

00:08:21 ◼   ► that.

00:08:25 ◼   ► The licensing question gets into the entire origin story of Apple Maps because, you know,

00:08:34 ◼   ► famously the backstory is before there was Apple Maps, there was, you know, the iOS shipped with

00:08:43 ◼   ► a maps app that used Google Maps as the backend. And I loved your lead. The lead of your article

00:08:52 ◼   ► is I'm not sure if you're aware, but the launch of Apple Maps went poorly.

00:08:58 ◼   ► That's such a great lead, man. I mean, it's like, man, that's a sentence.

00:09:06 ◼   ► Part of it wasn't just that it was a bad launch and, you know, and had some really

00:09:19 ◼   ► glaringly bad errors, you know, bridges that that look like they were, you know, like from

00:09:24 ◼   ► satellite view look like they'd melted. You know, driving directions that would send you

00:09:30 ◼   ► on impossible routes, or into the ocean into the ocean, right? Yeah. You know, I mean,

00:09:38 ◼   ► there's just all sorts of comical thing into the ocean, who knows what else.

00:09:48 ◼   ► But it was that it was on the heels of an app that until that rolled out was using Google Maps and

00:09:56 ◼   ► didn't have those problems. And so you know, the reaction was, why would Apple do this? Why didn't

00:10:02 ◼   ► they just stick with Google Maps? And, right, I thought that that was the one question I had

00:10:07 ◼   ► with your story where you got to talk to Eddie Q, who's in charge of the maps team. And he makes it

00:10:14 ◼   ► sound in a quote, I forget where it is here in the story about whether or not they wanted

00:10:18 ◼   ► to get do maps in the first place. What here's here's the we decided to I forget where the

00:10:29 ◼   ► quote is. But I don't think I think that's a little disingenuous. I felt like they had

00:10:33 ◼   ► to like the backstory I know of the situation is that their hand was forced to do their

00:10:39 ◼   ► own maps if they wanted to remain competitive because Google was asking for more. You do

00:10:48 ◼   ► touch on the privacy issue repeatedly. You talked to a bunch of people at Apple. You

00:10:54 ◼   ► spent a lot of time on this story. Everybody you spoke to emphasized that this whole new

00:10:59 ◼   ► thing that they've built was designed with privacy from the ground up. We'll get into

00:11:03 ◼   ► why that's important because they're using individual iPhones to collect data. But that

00:11:09 ◼   ► privacy issue is really what drove them away from Google back in 2010 or whenever this

00:11:17 ◼   ► was because the maps that they had from Google originally in iOS weren't vector-based. They

00:11:23 ◼   ► were bitmaps. And for those of you who don't know, it's a huge difference. It's the difference

00:11:29 ◼   ► between a scalable graphic format, you know, like SVG versus a just pixel by

00:11:38 ◼   ► pixel that's a bitmap, just a bunch of pixels. And so the old Apple Maps were

00:11:43 ◼   ► bitmap based and so at different resolutions it would be a different just

00:11:48 ◼   ► bitmap graphic and you'd zoom in and didn't really you'd have to wait for new

00:11:54 ◼   ► new tiles to come down. Whereas if you have a vector map it's just a huge

00:11:59 ◼   ► difference in terms of as you pinch and zoom it scales smoothly. They didn't have the licensing

00:12:06 ◼   ► rights to do turn-by-turn directions and that is, you know, in hindsight it's crazy, right? That's

00:12:11 ◼   ► one of those things like when you think about the iPhone originally in 2007 didn't even have video,

00:12:16 ◼   ► didn't shoot video on the camera. It's like, well that seems nuts. Like not having turn-by-turn

00:12:22 ◼   ► directions on your phone built in seems nuts now and they didn't have it then and they couldn't

00:12:28 ◼   ► at it without acquiescing to licensing terms from Google that they were unwilling to swallow.

00:12:34 ◼   ► They were between a rock and a hard place, effectively. They couldn't say yes to Google

00:12:38 ◼   ► without agreeing to things they didn't want to agree to, and their own technology wasn't

00:12:43 ◼   ► ready to go. That's basically the story of why Apple launched Maps, even though it was

00:12:50 ◼   ► in pretty shoddy shape.

00:12:52 ◼   ► Yeah, I mean, I think basically that's, I think you're exactly right.

00:12:56 ◼   ► Obviously, it's first strategically from a business angle, you know, that's why they

00:13:01 ◼   ► had to do it.

00:13:04 ◼   ► I didn't get too much into that because I didn't really, you know, they wouldn't have

00:13:07 ◼   ► talked about it.

00:13:08 ◼   ► Not, not so probably not, but at the same time, there have been endless pieces written

00:13:14 ◼   ► about that era of Apple maps, you know, of, of the goings on.

00:13:18 ◼   ► But I mean, I've written some over the years, you know, of the goings on behind the scenes

00:13:21 ◼   ► and you have and plenty of other people have analyzed it and plenty of good reporters have

00:13:26 ◼   ► like gotten stories about sort of what went wrong. And I think everybody kind of understands there

00:13:30 ◼   ► was, you know, dysfunction within the organization that led to, you know, the maps kind of being

00:13:36 ◼   ► shipped in the state that they were, but they were also not helped by any means by the timetable,

00:13:42 ◼   ► and the business pressures and everything else. Right? So you had you had like a perfect storm of

00:13:47 ◼   ► situation where you had to launch something, and yet that you knew that thing wasn't going to be as

00:13:52 ◼   ► good as you wanted it to be. And I think they acknowledge that. I think Eddie acknowledges it

00:13:57 ◼   ► pretty directly. And the maps folks that I talked to at Apple, it wasn't, you know, there was no

00:14:02 ◼   ► denial there, right? Nobody, nobody there once tried to convince me that the original version

00:14:09 ◼   ► of maps was just great. You know, I mean, obviously, very famously, Tim apologized publicly.

00:14:14 ◼   ► But these are also people that have worked on maps for a long time and been at Apple for many, many years.

00:14:19 ◼   ► They have very senior people working on this project now.

00:14:21 ◼   ► And so, you know, many of them might have had the instinct to kind of to defend themselves in that regard.

00:14:28 ◼   ► And I think basically it was, you know, we knew we had to do it.

00:14:33 ◼   ► We did it with what we had available to us at the time, which is partner data.

00:14:36 ◼   ► And, you know, you saw what happened.

00:14:39 ◼   ► So since then, I think that many people on the Maps team

00:14:43 ◼   ► and people at Apple do feel that they've improved

00:14:46 ◼   ► Maps significantly, and I think that they have.

00:14:48 ◼   ► And I don't mean to say that it's perfect,

00:14:51 ◼   ► or I don't mean to say that it's in a state

00:14:54 ◼   ► where I think it's all fine,

00:14:56 ◼   ► and why are they even doing this new thing?

00:14:58 ◼   ► It certainly isn't in that arena at all,

00:15:01 ◼   ► but it's a damn sight better than it was when it launched,

00:15:05 ◼   ► in terms of directions and usability,

00:15:07 ◼   ► and of course it's got like transport in now,

00:15:09 ◼   ► and a bunch of other nice improvements.

00:15:12 ◼   ► So they've sort of been gussying up the house,

00:15:16 ◼   ► as they've gone, and it's certainly in much better shape now

00:15:20 ◼   ► but they were gussying up a house

00:15:23 ◼   ► that they built in a hurry.

00:15:25 ◼   ► - Right.

00:15:26 ◼   ► - And so if the timbers are crooked,

00:15:27 ◼   ► you know, and the foundation is leaning to one side,

00:15:31 ◼   ► you're never going to be able to build as high as you want.

00:15:33 ◼   ► And that's, I think, where they ended up.

00:15:35 ◼   ► And Eddie's answer to me when I asked him about like,

00:15:38 ◼   ► why now and how long did it take and all of that stuff.

00:15:41 ◼   ► He said it took four years,

00:15:42 ◼   ► which would mean that it was started,

00:15:45 ◼   ► let's call it in the 2014, 13, 14 kind of range,

00:15:50 ◼   ► somewhere there.

00:15:51 ◼   ► And they felt that they were starting to get critical mass

00:15:56 ◼   ► on the number of devices that were out there

00:16:00 ◼   ► so that they could start collecting data from them

00:16:02 ◼   ► in a meaningful but yet privacy conscious way

00:16:06 ◼   ► allowed them to augment the information that they'd have from maps. And they knew at that

00:16:11 ◼   ► point, they came to this realization, hey, we launched maps because we had to for a variety

00:16:16 ◼   ► of reasons as you outlined. But we also realized that the future of all of these devices is

00:16:22 ◼   ► location based. Whether you know, I don't care whether it's AR or, or, or GPS direction,

00:16:30 ◼   ► GPS based directions or APIs that are provided to developers. There are just dozens of threads

00:16:37 ◼   ► that touch maps. You know, maps is just everywhere in the phone. You know, it's an integral part of

00:16:45 ◼   ► a majority of features of the phone. Can't find a single marquee feature of the phone that doesn't

00:16:51 ◼   ► tap it somehow. I mean, you say, "Oh, the camera's the reason people buy phones." Yeah, sure,

00:16:57 ◼   ► That's great, but you know exactly where every photo was taken because of maps, you

00:17:02 ◼   ► know?

00:17:03 ◼   ► And then like there's, it touches everything.

00:17:04 ◼   ► And so they figured if we're gonna, we, we already own it, we already took the responsibility

00:17:08 ◼   ► to do it.

00:17:09 ◼   ► You know, it's already our bag.

00:17:10 ◼   ► You know, it's already, we're already holding the bag.

00:17:13 ◼   ► So if we're gonna hold that bag, we should figure out how to build it from the ground

00:17:18 ◼   ► up and stop relying on partners.

00:17:20 ◼   ► And this is a constant refrain as you've written about us.

00:17:23 ◼   ► many people have noticed over the past few years

00:17:27 ◼   ► is that Apple is making itself into something

00:17:30 ◼   ► that owns all of its key core features,

00:17:33 ◼   ► whether that's processor design or UI or maps.

00:17:38 ◼   ► So they figured, hey, we've got a couple of

00:17:41 ◼   ► colliding circumstances that allow us to do this now.

00:17:46 ◼   ► One, we've got resources up the yang,

00:17:48 ◼   ► so we can field these vans, we can start building them,

00:17:53 ◼   ► can learn how to make them. I mean, people started seeing these vans on the on the street in like

00:17:57 ◼   ► 2015 or whatever. They were obviously building them in 2014 to try and get them on the street in

00:18:03 ◼   ► 2015. And then you look at now and we're, they're just starting to use some of the data that they've

00:18:09 ◼   ► collected with these vans. So it wasn't like they started collecting data back then. And now all

00:18:13 ◼   ► the data is going in, they had to figure out how to do it, right? Like they had to build a van, had

00:18:17 ◼   ► to hire the people build the vans, do all that stuff. Well, there's a there's a there's a couple

00:18:21 ◼   ► mentions in your story about the building custom tools right so a big

00:18:27 ◼   ► part of that is is the software that drives these bands and that uses that

00:18:32 ◼   ► data you also mentioned that they've built custom tools so that Apple's own

00:18:38 ◼   ► human I think they're called editors map editors but you know human beings

00:18:43 ◼   ► that's the way they were referred to me as well anyway right human beings can

00:18:47 ◼   ► look at, you know, there's been reports that, you know, there's something's wrong with this,

00:18:53 ◼   ► you know, the this intersection or the entrance to this building or something like that. And

00:18:58 ◼   ► that they've got tools so that they can just go in and edit it. And then that the fix goes

00:19:03 ◼   ► live, you know, taking out that loop that you were talking about, with the licensed

00:19:08 ◼   ► stuff from TomTom, where then they could, you know, the old days, it was like file,

00:19:12 ◼   ► they'd file a report with TomTom and wait for TomTom to update their thing, hopefully,

00:19:17 ◼   ► and then it comes back and could roll out. But now they've built these tools. And you

00:19:20 ◼   ► know what? It takes time to build good tools.

00:19:23 ◼   ► It does. And I'll tell you, I saw, I don't know, maybe, I don't know, eight to 10 different

00:19:31 ◼   ► tools or different applications of tools that they had built. And the scenarios that are,

00:19:38 ◼   ► I mentioned some of them in the piece, but I didn't really get

00:19:41 ◼   ► into it as much as I could, because you got to sort of stop

00:19:43 ◼   ► writing at some point. But the the tools are very interesting.

00:19:47 ◼   ► They are, they are meant for humans to use. So they're,

00:19:50 ◼   ► they're meant for human editors to be able to rapidly clip

00:19:53 ◼   ► through, you know, issues or flagged issues and decide what

00:19:58 ◼   ► to do about them, right, and to make a human call of what to do.

00:20:02 ◼   ► So in the case of an intersection, you know, a new

00:20:04 ◼   ► intersection goes up and and something somebody flags it that Oh, this told me

00:20:09 ◼   ► to turn left or whatever. And they go to look at that incident report, they tap

00:20:13 ◼   ► through to the next flag for that section of the map. And the flag says,

00:20:19 ◼   ► Hey, something's up with the lanes here. And they have tools that they can just

00:20:22 ◼   ► basically click on a lane and say, Oh, okay, you're supposed to be attached to

00:20:25 ◼   ► this lane. And they click on another one and bloop, right, it attaches it. And

00:20:29 ◼   ► from that point on, once they commit that change, they then commit it, it goes

00:20:32 ◼   ► into review, it gets reviewed, and they get it's approved, right? But all within Apple's

00:20:37 ◼   ► organization now, it appears in the map. Boom, done. Right now that lane connects properly,

00:20:42 ◼   ► and it doesn't tell you to make a u turn instead of a left turn, right? Or doesn't tell you

00:20:46 ◼   ► to go from the inside lane to the outside lane because one of the new things is better

00:20:49 ◼   ► lane guidance with like more accurate lanes, including bike lanes and everything else.

00:20:56 ◼   ► They're like, I'll just say this as an aside, their lane guidance has gotten significantly

00:21:00 ◼   ► better.

00:21:01 ◼   ► - Yeah, and lane guidance is one of those things that, we talked about this through

00:21:08 ◼   ► various aspects of maps, but it's one of those things that just affects cognitive load a

00:21:14 ◼   ► massive amount with maps because you're going 80, or well, you're going 55.

00:21:20 ◼   ► (laughs)

00:21:21 ◼   ► Whatever, you're going, insert legal speed limit here, and you're trying to figure out,

00:21:28 ◼   ► I take this exit? Do I need to be in the left hand side of the exit or the right hand side?

00:21:32 ◼   ► All of these decisions that you're making in a rapid clip with a two-ton vehicle, and

00:21:37 ◼   ► any time that you can get very, very precise like, "Hey, no, you need to be right here.

00:21:41 ◼   ► You're in the right place. It's all good. We got this handled." It improves safety.

00:21:46 ◼   ► It improves, it lessens stress. There's a lot of benefits there. So yeah, lane guidance

00:21:52 ◼   ► seems like one of those conveniences, but in effect, I think it makes driving safer.

00:21:56 ◼   ► You know, so you don't have to jet over five lanes.

00:21:58 ◼   ► It's also the sort of thing that can change very rapidly when there's construction.

00:22:05 ◼   ► What used to be maybe there were two lanes that you could take to get off here because

00:22:09 ◼   ► it's busy, but now it's just one lane.

00:22:11 ◼   ► You know, the lane guidance has to be different temporarily.

00:22:14 ◼   ► Absolutely.

00:22:15 ◼   ► Yeah.

00:22:16 ◼   ► And that's the kind of stuff that they're hoping will get flagged and corrected very

00:22:20 ◼   ► quickly.

00:22:21 ◼   ► the flagging will happen automatically through machine learning tools or, you know, kind of

00:22:27 ◼   ► algorithmic segmentation. And then some of it will happen just via humans reporting it. And some of

00:22:33 ◼   ► it they use, they're using city data. You know, the city says, "Hey, we got a new subdivision,

00:22:39 ◼   ► and here's all the addresses for the subdivision." You know, they can say, "Okay, cool, let's task a

00:22:44 ◼   ► van. The van goes and runs the subdivision. So it has now it has exact tracking data, imagery, and

00:22:52 ◼   ► GPS location and LIDAR data. So a 3D point cloud of the neighborhood. And then you can go in and

00:23:00 ◼   ► correct all the addresses to make sure they're attached to the right houses within a few

00:23:04 ◼   ► minutes. I mean, I saw the the operations I mean, one person could clear a subdivision in just a

00:23:10 ◼   ► couple of minutes to make sure that all the addresses are pinned to the right houses and

00:23:13 ◼   ► that the directions are correct and then that road system can get added very very quickly.

00:23:18 ◼   ► All right, I have a lot more to say about maps or to ask you about maps but in the

00:23:22 ◼   ► in the meantime let's take a break here and thank our first sponsor brand new sponsor on the show.

00:23:26 ◼   ► They have sponsored my website before but they're brand new on the show Kolide. K-O-L-I-D-E. They're

00:23:33 ◼   ► a new startup and they're working to solve security challenges for tech companies that

00:23:38 ◼   ► that run large Mac fleets,

00:23:40 ◼   ► and that's a lot of tech companies out there

00:23:42 ◼   ► that are largely Mac-based.

00:23:44 ◼   ► They believe that these organizations, Macs,

00:23:46 ◼   ► which are often driven by software engineers and designers

00:23:50 ◼   ► who have broad access to valuable intellectual property

00:23:53 ◼   ► and customer data, are low-hanging fruit for compromise,

00:23:56 ◼   ► not because macOS is somehow insecure, it's not,

00:23:59 ◼   ► but just that the daily grind of software development

00:24:01 ◼   ► causes employees to disable important security features

00:24:04 ◼   ► or maybe make risky decisions

00:24:06 ◼   ► with where they put data temporarily.

00:24:08 ◼   ► Examples, what good is full disk encryption

00:24:10 ◼   ► if your users are making unencrypted time machine backups?

00:24:14 ◼   ► Or what good is using SSH keys

00:24:16 ◼   ► to access a production server

00:24:18 ◼   ► if that key isn't encrypted and resides in a share folder?

00:24:21 ◼   ► There's all sorts of issues like that

00:24:24 ◼   ► and Colide is tackling them specifically

00:24:27 ◼   ► for tech companies with large Mac fleets.

00:24:29 ◼   ► And they've just launched their first product.

00:24:31 ◼   ► It's called Colide Cloud.

00:24:33 ◼   ► And it's the first product in the industry

00:24:35 ◼   ► that takes these type of user-driven security holes

00:24:37 ◼   ► seriously and helps you understand the cause and effect

00:24:40 ◼   ► relationships between the many seemingly innocuous

00:24:44 ◼   ► or disconnected actions that users can take

00:24:46 ◼   ► and the potentially catastrophic consequences

00:24:49 ◼   ► if something like valuable, like private customer data

00:24:52 ◼   ► or intellectual property were somehow leaked.

00:24:55 ◼   ► Colide not only allows you to detect dozens

00:24:57 ◼   ► of these situations, but it also includes

00:24:59 ◼   ► a non-obtrusive menu bar app, just a simple little thing

00:25:02 ◼   ► that sits in the menu.

00:25:03 ◼   ► And then you can deploy this to the users

00:25:06 ◼   ► effectively communicate your organization's security policy.

00:25:11 ◼   ► It doesn't just help employees who are unaware of the policy, but also highly technical employees

00:25:16 ◼   ► who get it, who understand these things, but who sometimes turn off security features temporarily

00:25:21 ◼   ► for convenience and then forget to turn them back on.

00:25:24 ◼   ► All of this is powered by OSQuery, an incredibly performant and open source agent that was

00:25:30 ◼   ► created by Colides founders during their time at Facebook.

00:25:34 ◼   ► OSQuery is 100% open source, doesn't degrade the performance of your Macs, and the project

00:25:41 ◼   ► is committed to providing important information to security analysts without violating the

00:25:47 ◼   ► privacy of end users.

00:25:49 ◼   ► You can try it free for the first 10 devices.

00:25:53 ◼   ► Colide Cloud is a free product for up to 10 devices, and you can sign up on their website

00:25:58 ◼   ► today at colide.com.

00:26:01 ◼   ► That's all you need to know.

00:26:03 ◼   ► dot com, 10 devices for free. It's a great service. It really seems like a bunch of interesting

00:26:11 ◼   ► people working there in my interaction with them. So go check them out. Really appreciate

00:26:15 ◼   ► their sponsoring the show. So here's the thing to me, and I think it applies to more than

00:26:22 ◼   ► maps is that there's this basic argument out there that in the areas where Apple is behind,

00:26:31 ◼   ► say maps and Siri are as two examples and you know Siri is a broad range of

00:26:37 ◼   ► products underneath one umbrella term but let's just say just to keep to keep

00:26:41 ◼   ► the conversation moving and competitors like Amazon and Google are less privacy

00:26:50 ◼   ► focused than Apple and the there's this argument that because Apple isn't

00:26:56 ◼   ► collecting more data and then aggregating it in the cloud and doing all this work up there.

00:27:04 ◼   ► That's why they're behind. And I've asked Apple executives about it, you know, like on my show.

00:27:10 ◼   ► And they're adamant that their strategy of distributing this and having the private stuff

00:27:17 ◼   ► local, that your phone is the only device that knows that you, Matthew Panzorino, are making

00:27:24 ◼   ► this trip from A to B and that any what do they call them slices of that trip

00:27:32 ◼   ► that while you segments segments of the trip that are transmitted to the cloud

00:27:38 ◼   ► so that Apple can help improve maps are done in a completely anonymous way with

00:27:43 ◼   ► these rotating numeric IDs that aren't associated with you such that not only

00:27:48 ◼   ► does Apple not know and can't know that there's no way it could put the pieces

00:27:54 ◼   ► it has back together to know that Matthew Panzarino went from point A to point B, that

00:27:59 ◼   ► they can't even tell that someone went from point A to point B because the individual

00:28:04 ◼   ► segments each have a different rotated ID.

00:28:07 ◼   ► Right.

00:28:08 ◼   ► Basically, basically what I think Apple's strategy is that they are doing massively

00:28:16 ◼   ► parallel computation to do these things like improve Siri and improve maps, but that the

00:28:25 ◼   ► parallelization is these 100 billion or however many, not 100 billion, 100 million, however

00:28:33 ◼   ► many active iPhones and iPads and Macs there are.

00:28:36 ◼   ► - 1.3 billion or something like that, yeah.

00:28:39 ◼   ► - You know, that you multiply the computing power of these devices by the number of devices

00:28:43 ◼   ► and there's an incredible amount of parallel computing power out there.

00:28:48 ◼   ► And I think that what they're saying with maps is that's sort of one of the

00:28:53 ◼   ► vectors they're taking to, to improve this.

00:28:56 ◼   ► Yeah. I mean,

00:28:58 ◼   ► what they're doing has a lot to do with the concept of edge computing.

00:29:03 ◼   ► And, you know, for those listeners who may not know, you know,

00:29:07 ◼   ► edge computing is basically, it's a type of cloud computing,

00:29:11 ◼   ► but it involves more distributed nodes that are sort of on the quote unquote edge of the

00:29:19 ◼   ► internet, which basically means where the internet comes in contact with the physical

00:29:22 ◼   ► world. So for instance, an edge computing device may exist on the furthest node of a

00:29:28 ◼   ► network to support local devices, let's say at a facility, and it can serve up files and

00:29:35 ◼   ► do all these kinds of things that normally a cloud server, let's say, hosted on AWS would

00:29:41 ◼   ► would do, but it could do it faster because it's closer and it could do it with redundancy and you

00:29:47 ◼   ► could do it with security because it's the only node that could communicate with the remote nodes,

00:29:51 ◼   ► you know, and it itself communicates back with the server, etc, etc. Right. There are a bunch

00:29:57 ◼   ► of different benefits to edge computing, and I'm a dummy, so I don't really like I'm no edge

00:30:01 ◼   ► computing expert, right. But edge computing or fog computing is basically this idea that you're going

00:30:06 ◼   ► going to have a lot of these powerful individual nodes that make up the cloud.

00:30:12 ◼   ► And each one of those nodes will have its own responsibilities. And to me, I

00:30:17 ◼   ► think a lot of people do not talk about it in this context. And maybe somebody

00:30:20 ◼   ► who listens to this will be like, "Oh, well, here's why it's fundamentally

00:30:23 ◼   ► wrong and different and all that." But to me, it shares a lot of

00:30:27 ◼   ► characteristics with edge computing because it allows the local device to

00:30:31 ◼   ► take care of its local business. And that by local device, I mean your

00:30:34 ◼   ► iPhone to say, "Hey, I'm going to take care of the personalization aspects. I'm going

00:30:39 ◼   ► to run AI and ML on your photos. I'm going to handle personal requests and then handle

00:30:45 ◼   ► translating those into generic requests and then passing those on and getting that back

00:30:49 ◼   ► and translating them back into personal requests." Right? You know, all of that stuff. And then

00:30:56 ◼   ► times 1.3 billion of those. And then you have the central computing stuff that Apple is

00:31:03 ◼   ► doing on the generic parts of that data, the parts of that data that will benefit the whole

00:31:08 ◼   ► by providing a base layer of compute power or machine learning models.

00:31:16 ◼   ► Like one of the announcements at WWDC that ties in closely with this is that Apple has

00:31:23 ◼   ► given developers the ability to create their own training models very, very easily using

00:31:29 ◼   ► Xcode to pop it open, you know, train a library on a set of images and get like a nice personalized,

00:31:37 ◼   ► you know, training bolus for their particular app rather than having to buy a generic one

00:31:42 ◼   ► and then retrain it with unique images. And one of Apple's big selling points on it is

00:31:48 ◼   ► that it's incredibly small. It's like 70% smaller and a hell of a lot faster than a

00:31:55 ◼   ► normal training model, because all of the base training is built in. So the Apple provides all of

00:32:03 ◼   ► that automatically. And then you only have to add in your specialized training on top of it. So it's

00:32:08 ◼   ► this idea that they are providing like a bunch of generic stuff that will get you 70, 80% of the way

00:32:14 ◼   ► there. And then you bring the last 20%. And that is your personal stuff, either from an app

00:32:20 ◼   ► developers perspective, obviously, or you as an individual being able to take advantage of all

00:32:26 ◼   ► the vast compute power that Apple has at its disposal. And then guess what, if I want to

00:32:30 ◼   ► process my 1500 photos that show up on vacation or allow them to be processed, that's all happening

00:32:36 ◼   ► locally and securely and remotely on my device. One of the things that this new maps project

00:32:46 ◼   ► is going to bring or says supposedly going to bring.

00:32:49 ◼   ► And the examples that they've shown

00:32:51 ◼   ► in the screenshots they provided you certainly do

00:32:53 ◼   ► is filling in the areas between roads.

00:32:59 ◼   ► So if it's a city filling in buildings

00:33:02 ◼   ► that actually have the outline of the building,

00:33:07 ◼   ► the actual building based on some combination

00:33:10 ◼   ► of satellite views and the street views that they might have

00:33:15 ◼   ► But that's so for example,

00:33:17 ◼   ► not when you're in the satellite view,

00:33:19 ◼   ► but when you're actually on just the map view,

00:33:21 ◼   ► you could see something like your example

00:33:25 ◼   ► is like the Salesforce Tower,

00:33:26 ◼   ► which has a very distinctive look.

00:33:29 ◼   ► Every city has buildings like that, right?

00:33:32 ◼   ► There's maybe a bunch of generic buildings,

00:33:33 ◼   ► but you can, like here in Philly,

00:33:36 ◼   ► City Hall would be one.

00:33:38 ◼   ► It's just a very distinctive building.

00:33:40 ◼   ► And it has been a tremendous advantage for Google Maps.

00:33:45 ◼   ► And for me as a very visual person,

00:33:48 ◼   ► I use Apple Maps most of the time,

00:33:50 ◼   ► but I have to admit that being able to navigate

00:33:53 ◼   ► and see just the sort of,

00:33:55 ◼   ► and it's a tricky illustration problem, right?

00:33:58 ◼   ► That you're looking at a 2D map from the top down,

00:34:02 ◼   ► but you wanna get a sort of 3D representation

00:34:05 ◼   ► of the top of the building, which might be a skyscraper,

00:34:08 ◼   ► it might just be a two or three story building,

00:34:10 ◼   ► how do you reflect that?

00:34:12 ◼   ► And it looks like they're bringing it.

00:34:14 ◼   ► And then the other thing is filling in details

00:34:16 ◼   ► like park areas and grass and things like that

00:34:19 ◼   ► where Google's had that in great detail

00:34:23 ◼   ► and Apple, it just looks like dead, it's just white space.

00:34:27 ◼   ► - Right.

00:34:28 ◼   ► Yeah, they've got some definite improvements there.

00:34:32 ◼   ► I mean, those come from, and I asked this,

00:34:34 ◼   ► like, okay, well, what's your primary source

00:34:36 ◼   ► all this data, you know, is it imaging? Is it the vans? Is it satellite? You know, what

00:34:41 ◼   ► is it? And the answer really varies depending on what it is. Like for instance, the ground

00:34:45 ◼   ► cover, the foliage, they are using significantly higher resolution satellite images now. I

00:34:52 ◼   ► didn't ask for the provider, I don't even know if they would have said, it's not really

00:34:55 ◼   ► that germane, but whatever the previous stuff that they were using, they're using better

00:34:59 ◼   ► stuff now, higher resolution stuff, which allows them to more even they run, of course,

00:35:05 ◼   ► vision algorithms on that, which tell them, "Hey, this is a tree, this is grass, this is a pathway,"

00:35:12 ◼   ► you know, that sort of thing, which allows them to really, really flesh out the outlines of foliage

00:35:18 ◼   ► or in the example, the Delta near Fort Bragg or whatever, you know, like that, there are definitely

00:35:25 ◼   ► some opportunities for it to more accurately represent the physical world in a way that is so

00:35:34 ◼   ► improved, it can actually improve your ability to use the map, right? Like you could say like,

00:35:38 ◼   ► Oh, yeah, that's the edge of the grass. Oh, that's where I'm standing. Right. You know,

00:35:41 ◼   ► as opposed to Oh, I'm standing somewhere in this blank, empty space. It's not just for looks. It's

00:35:46 ◼   ► actually very effective as a as a marker, you know, to tell you where you are in the world.

00:35:51 ◼   ► Trenton Larkin Well, and that was a big thing for them,

00:35:53 ◼   ► adding things like, in addition to the roads of a city having the here's the actual path of the foot

00:36:00 ◼   ► footpaths through the park, right?

00:36:02 ◼   ► And oh yeah, I get it.

00:36:03 ◼   ► Yeah, there's the circle in the middle with the fountain

00:36:05 ◼   ► and oh, I can go this way or that way.

00:36:07 ◼   ► I recognize, you could recognize the park

00:36:10 ◼   ► by the footpath map.

00:36:12 ◼   ► And that's the sort of thing that Apple Maps

00:36:14 ◼   ► was lacking for a while.

00:36:16 ◼   ► Another thing, here's a segment from your article.

00:36:21 ◼   ► And this to me is a big deal.

00:36:23 ◼   ► And I was just thinking about this at WWDC

00:36:25 ◼   ► a couple weeks ago.

00:36:29 ◼   ► said it also allows for access points to be set making apple maps smarter about the quote last

00:36:34 ◼   ► 50 feet of your journey so you've made it to the building but what street is the entrance actually

00:36:40 ◼   ► on that's like such a huge problem and i guess i i never really thought about it but i think eddie

00:36:49 ◼   ► q is probably right that a lot of buildings they'll say that they want to be they want their

00:36:53 ◼   ► address to be 400 market street because that's a nice address and that's the street they want to be

00:36:58 ◼   ► on. But for architectural reasons, the entrance is on the street around the corner, you know,

00:37:05 ◼   ► and so you you get directions to 400 Market Street, and there's the building, but where do

00:37:10 ◼   ► you go in? I've seen this and a friend of the show, Paul classes and I were in when we were at the

00:37:15 ◼   ► WDBDC reminder, we were walking to lunch at the in San Jose. And a fellow, you know, obviously,

00:37:25 ◼   ► English as a second language was there's like an immigration building there. And he had just gotten

00:37:30 ◼   ► out of, you know, like a cab or an Uber or something. And he had the address. And he

00:37:35 ◼   ► obviously had like, you know, some kind of important meeting for his, you know, whatever.

00:37:40 ◼   ► But I mean, he's obviously an immigrant, he's meeting at the immigration center, he's standing

00:37:45 ◼   ► right where he's got a phone. And it's telling him, he's at the right place. You're there. And

00:37:50 ◼   ► And he asked us, he was like, asked us if we knew

00:37:52 ◼   ► how to get in, and we didn't, we're not from San Jose,

00:37:55 ◼   ► but we wanted to help the fellow.

00:37:57 ◼   ► And you know what the truth was,

00:37:58 ◼   ► he was actually like 500 feet away and around a corner

00:38:02 ◼   ► from where he needed to go.

00:38:04 ◼   ► It was nowhere, and it wasn't even visible, right?

00:38:07 ◼   ► But he was at the right place, you know?

00:38:09 ◼   ► And so--

00:38:10 ◼   ► - And buildings get reconfigured over time, right?

00:38:12 ◼   ► Like the doors used to be here,

00:38:14 ◼   ► but now they're over there because of zoning restrictions

00:38:16 ◼   ► or some new owner, whatever.

00:38:19 ◼   ► Yeah, exactly. Yeah, but that last 50 feet is a huge next level in mapping just in terms

00:38:26 ◼   ► of having the street address. You know, that was that that was that's just table stakes down.

00:38:30 ◼   ► You know, the actual knowing it is a huge part of it. And I know that all the self-driving car

00:38:36 ◼   ► or not self-driving car, the ride sharing car companies like Uber and Lyft are all working on

00:38:41 ◼   ► the same problem too, because it's a huge issue with getting picked up at the right place that,

00:38:46 ◼   ► you know, knowing that if you're at a certain building, you know, the place where you walk

00:38:52 ◼   ► out the door, let's say it's raining, so you don't want to stand outside, the door that they should

00:38:57 ◼   ► pull up to might be very different than the street address of the building. Yeah, exactly. And I used

00:39:04 ◼   ► to complain constantly about that a couple of years ago with like Uber and Lyft is that, you

00:39:09 ◼   ► know, that last 50 feet, either somebody dropping you off or somebody picking you up was just a mess,

00:39:14 ◼   ► you know, like, oh, I got to run down the street, they don't know where I am. If I put in the address,

00:39:19 ◼   ► it actually puts it halfway down the block because of that whole door thing. All of that is a huge

00:39:23 ◼   ► mess. And they've gotten much better at it by measuring where people get picked up and dropped

00:39:28 ◼   ► off actually. Right. Hey, lead on nine times out of 10. The person wants to be here, not there.

00:39:34 ◼   ► Right. And it's especially bad in a city that you're unfamiliar with. I was in,

00:39:40 ◼   ► I was in Chicago recently, and I love Chicago, but I'm not there all the time, and had an Uber

00:39:48 ◼   ► driver picking me up. And it was just, you know, we had to resort to the phone call, you know,

00:39:52 ◼   ► and it's like, I'm here. And it's like, well, I'm here too. And I don't see you. And of course,

00:39:56 ◼   ► he's around the corner, you know, and, you know, we worked it out. It was all, you know,

00:40:00 ◼   ► all worked out. But, you know, it added four minutes of stress to, to what ordinarily is

00:40:08 ◼   ► supposed to be completely stress-free. It's like the whole appeal of these things.

00:40:12 ◼   ► Chris Willis Yep, exactly. And so the tools are a big part of that. Obviously, there is

00:40:18 ◼   ► an opportunity for Apple to gather data from phones to say like, "Hey, trips often end right

00:40:32 ◼   ► here." Right? So maybe this is where the front entrance is. Let's assign that to an editor.

00:40:37 ◼   ► And an algorithm may do that.

00:40:40 ◼   ► They may not be a person that has to say that, right?

00:40:43 ◼   ► It's just saying, hey, we've got the door here,

00:40:46 ◼   ► and it doesn't match where everybody walks to.

00:40:49 ◼   ► So what's the variance?

00:40:51 ◼   ► And so they assign it to a person, an editor.

00:40:54 ◼   ► The editor has the tool to do the access point check.

00:40:58 ◼   ► So they go and look at it, and they say,

00:40:59 ◼   ► oh, I see what's going on here.

00:41:01 ◼   ► And it's literally as simple as,

00:41:03 ◼   ► it looks like a vector illustration, to be honest.

00:41:07 ◼   ► You know, like, hey, I'm gonna drop a point

00:41:09 ◼   ► and I'm gonna drag to another point and drop another point.

00:41:11 ◼   ► So it's like, hey, I'm gonna create an access point here,

00:41:13 ◼   ► drag it out, connect it to the main road

00:41:16 ◼   ► so I know it's coming from the main road,

00:41:17 ◼   ► and then this is exactly the point where they need to end up.

00:41:20 ◼   ► And so that way, when you're navigating

00:41:22 ◼   ► and then attach it to that address,

00:41:24 ◼   ► assign it to that address,

00:41:25 ◼   ► and then that way, if I put in that address,

00:41:27 ◼   ► no longer am I directed to just a random spot

00:41:29 ◼   ► along the street, I'm directed right to the access point.

00:41:32 ◼   ► whether that's on the road or off the road.

00:41:34 ◼   ► - And there are landmarks to help you orient it, right?

00:41:37 ◼   ► Like you're driving the car and you see like,

00:41:38 ◼   ► oh, there's an awning, you know, there's an awning.

00:41:41 ◼   ► Oh, I see that it's this restaurant up here

00:41:43 ◼   ► with the awning, there it is.

00:41:44 ◼   ► - Right.

00:41:46 ◼   ► - All right, let me take a break again

00:41:47 ◼   ► and thank our next sponsor.

00:41:48 ◼   ► Sorry, good friends at Eero, E-E-R-O.

00:41:51 ◼   ► Eero has a new second generation router

00:41:56 ◼   ► and their new beacon.

00:41:57 ◼   ► And they allow you to easily build your own wifi system

00:42:01 ◼   ► that works way better than traditional Wi-Fi routers.

00:42:05 ◼   ► Look, traditional routers have built clunky devices,

00:42:10 ◼   ► they have terrible software,

00:42:11 ◼   ► and they try to blanket as big of an area as possible

00:42:14 ◼   ► with a single range of Wi-Fi.

00:42:19 ◼   ► Eero uses a mesh network,

00:42:22 ◼   ► and they have their own proprietary TrueMesh technology.

00:42:25 ◼   ► And the gist of it is that you have multiple small devices

00:42:28 ◼   ► spread about your home or your office,

00:42:30 ◼   ► and they communicate with each other.

00:42:32 ◼   ► And each one, from your device's standpoint,

00:42:36 ◼   ► it looks like one single WiFi network,

00:42:38 ◼   ► one network name, one password,

00:42:41 ◼   ► and your device will just magically connect

00:42:45 ◼   ► to the closest one that's giving that part of your home

00:42:48 ◼   ► or your office strong power.

00:42:50 ◼   ► Their first generation product is fantastic

00:42:53 ◼   ► and really got all sorts of awards and acclaim,

00:42:57 ◼   ► but their second generation one,

00:42:58 ◼   ► They've added a third 5 gigahertz radio,

00:43:01 ◼   ► and it's now tri-band and twice as fast as its predecessor,

00:43:05 ◼   ► which lets its customers do more simultaneously

00:43:07 ◼   ► in every room in their home.

00:43:08 ◼   ► It's a really great addition.

00:43:10 ◼   ► And just like the original one,

00:43:13 ◼   ► it looks identical to the original one,

00:43:15 ◼   ► or practically identical.

00:43:16 ◼   ► It's just a very small, apple-like puck,

00:43:20 ◼   ► and it's very small, unobtrusive.

00:43:22 ◼   ► It'll look good in any room in your house.

00:43:24 ◼   ► Sits flat on the surface, plugs in with a power adapter,

00:43:27 ◼   ► nothing clunky about it.

00:43:28 ◼   ► and you can connect the multiple ones over Ethernet.

00:43:31 ◼   ► So if your home is wired up for Ethernet,

00:43:33 ◼   ► you can hook each one in to Ethernet

00:43:36 ◼   ► and they'll communicate to each other that way,

00:43:38 ◼   ► or wirelessly.

00:43:39 ◼   ► So if you don't have an Ethernet hookup

00:43:41 ◼   ► everywhere in your house,

00:43:43 ◼   ► they'll just communicate to each other wirelessly

00:43:45 ◼   ► in any combination.

00:43:46 ◼   ► And it really couldn't be easier to set up.

00:43:48 ◼   ► And they have a great app for the phone

00:43:50 ◼   ► that is how you deal with it

00:43:52 ◼   ► and set the name of the network, see who's using it,

00:43:55 ◼   ► all sorts of stuff.

00:43:57 ◼   ► The app has a terrific user interface.

00:44:00 ◼   ► And their new product, the Beacon, is half the size,

00:44:04 ◼   ► but still more powerful than the first generation Eero.

00:44:07 ◼   ► And you simply plug it into a wall socket

00:44:09 ◼   ► like a little ambient light.

00:44:15 ◼   ► A nightlight is the word I'm looking for.

00:44:17 ◼   ► And it can just join the network.

00:44:20 ◼   ► And it's even more unobtrusive because it's just

00:44:23 ◼   ► a little tiny thing that you plug into wall sockets,

00:44:25 ◼   ► you know, maybe at the top of stairs or something like that.

00:44:29 ◼   ► Really an amazing addition to their lineup

00:44:32 ◼   ► and a great way to maybe add, you know,

00:44:34 ◼   ► just to get one more,

00:44:36 ◼   ► one more Eero device,

00:44:37 ◼   ► like on the top floor of your house or something like that.

00:44:40 ◼   ► And it even actually does have an LED nightlight

00:44:43 ◼   ► that you can turn off if you don't want,

00:44:45 ◼   ► but it uses an ambient light sensor.

00:44:47 ◼   ► So it only turns on when it thinks you need it in the dark.

00:44:51 ◼   ► So anyway, it's a great product.

00:44:53 ◼   ► I'm talking to you right now over an ero network. I've had a great experience with it. I really

00:44:58 ◼   ► had regular software updates. It's just, it's just a great, great product that I would,

00:45:02 ◼   ► I would recommend even if they weren't sponsoring the show. Um, you get free overnight shipping

00:45:09 ◼   ► to the U S or Canada by visiting ero.com and then select overnight shipping when you order

00:45:16 ◼   ► and enter the promo code, the talk show to make that overnight shipping free. So if you're

00:45:22 ◼   ► listening to this show, whenever day you're listening to it, you can have your Eero in

00:45:25 ◼   ► your hands tomorrow for free, shipping-wise at least, by using that code "THETALKSHOW"

00:45:33 ◼   ► and just going to Eero.com. That's for the US and Canada. Great product.

00:45:41 ◼   ► So why do you think they didn't announce the maps at WWDC?

00:45:44 ◼   ► I mean, the answer I got was just to give it more air. I don't know if that's only part

00:45:50 ◼   ► of the answer. Usually that's the way Apple rolls. You'll get part of the answer but

00:45:55 ◼   ► maybe not the whole answer. So some conjecture about the whole answer could be like it just

00:46:00 ◼   ► wasn't ready. It was just too crowded up on the stage. They had too many other things

00:46:04 ◼   ► to get to, whatever the case may be. I mean it doesn't make sense, I don't think, to

00:46:09 ◼   ► announce something like this during the platforms keynote, like after the fact where it would

00:46:15 ◼   ► have more room. It's just not really developer focused, at least not yet. It would have to

00:46:20 ◼   ► be on the main stage and so maybe they just ran out of time. It was right on an hour and

00:46:26 ◼   ► a half or two hours, whatever it is that they normally do. I don't know. But the answer

00:46:33 ◼   ► I got was basically just wanted to give it more air, separate it from the developer news

00:46:37 ◼   ► and give it its own sort of moment to like, "Hey, this is our new project. Here's what

00:46:41 ◼   ► we're pushing forward on. Here's what you can look forward to."

00:46:44 ◼   ► My guess is and I have no insight info on this but my guess is simply didn't make the cut

00:46:51 ◼   ► You know that they they probably considered it. It may have even been in the keynote at one point and in the interest of time

00:46:57 ◼   ► But well we can roll this out, you know later in a month as a separate thing

00:47:04 ◼   ► And like it's not developer focus yet, you know

00:47:10 ◼   ► It's just that the actual, everything developer-wise in terms of using map APIs and iOS is unchanged.

00:47:20 ◼   ► As this rolls out, just the maps that users will see will look better and have more information.

00:47:25 ◼   ► Yeah.

00:47:26 ◼   ► I mean, I think they have an opportunity right now because there's a beta next week where

00:47:34 ◼   ► the Bay Area will come online and people will be able to play with it and get an idea.

00:47:38 ◼   ► and then of course, Northern California later in the year

00:47:42 ◼   ► when iOS actually launches, iOS 12.

00:47:47 ◼   ► But I think they have an opportunity to sort of say,

00:47:50 ◼   ► to put it alone and not put it on stage

00:47:53 ◼   ► with everything else that's gonna be shipping to everybody.

00:47:57 ◼   ► You know what I mean?

00:47:58 ◼   ► Because technically while this is shipping to everybody,

00:48:01 ◼   ► everybody doesn't live at the bank.

00:48:03 ◼   ► Everybody doesn't live in Northern California, right?

00:48:05 ◼   ► Whereas everybody can use the new attention features

00:48:10 ◼   ► in iOS, the new app limit features.

00:48:15 ◼   ► Everybody could use that.

00:48:16 ◼   ► I don't care whether you live in Australia

00:48:17 ◼   ► or here or wherever, right?

00:48:18 ◼   ► Everybody could use that.

00:48:20 ◼   ► If you launch this map,

00:48:21 ◼   ► you're not really launching it to everybody.

00:48:23 ◼   ► So it's sort of like,

00:48:25 ◼   ► you and I had a little discussion about Google's thing.

00:48:30 ◼   ► What was it called?

00:48:30 ◼   ► It's like Duplex. - I/O?

00:48:31 ◼   ► Or Duplex, yeah.

00:48:32 ◼   ► - No, Duplex.

00:48:33 ◼   ► Remember it was like, hey, you announce Android P,

00:48:35 ◼   ► which like every Android user on the planet

00:48:39 ◼   ► with a compatible phone, so a very small amount actually,

00:48:43 ◼   ► can get this theoretically coming right up, right?

00:48:46 ◼   ► Like everybody will have access to it

00:48:48 ◼   ► that can have access to it.

00:48:49 ◼   ► And then here's duplex, which we don't even know

00:48:51 ◼   ► when we're gonna ship it or what it is, right?

00:48:53 ◼   ► And putting those in right next to one another

00:48:56 ◼   ► can create like a false dichotomy in the consumer's mind.

00:49:01 ◼   ► Like, oh, this is ready to ship or whatever.

00:49:03 ◼   ► So maybe that's the case too,

00:49:04 ◼   ► because it's such a limited area that it's launching in.

00:49:06 ◼   ► They don't want it right alongside iOS features,

00:49:09 ◼   ► which will ship to everyone.

00:49:10 ◼   ► - I thought this was an interesting quote from Eddie.

00:49:13 ◼   ► He said, after talking about some specifics

00:49:16 ◼   ► of the work that they're doing,

00:49:19 ◼   ► we don't think there's anybody doing this level of work

00:49:22 ◼   ► that we're doing, Ads Q,

00:49:25 ◼   ► which is a pretty bold statement for the Maps app

00:49:30 ◼   ► that has been in second place,

00:49:33 ◼   ► or arguably distant second place for a very long time to Google Maps.

00:49:37 ◼   ► Right? Like that quote makes it sound as though he thinks they're doing stuff

00:49:41 ◼   ► that Google isn't doing like that.

00:49:44 ◼   ► It's not just about getting Apple maps on par with Google maps,

00:49:47 ◼   ► but that if they're successful at what they're trying to do,

00:49:50 ◼   ► it could put them ahead, which is a really bold statement.

00:49:53 ◼   ► Yeah. And that could be referring to some of the things that they're doing with,

00:49:58 ◼   ► um, with 3d point cloud capture. Um,

00:50:02 ◼   ► and then they could also refer to, of course, future stuff, right, that he was unwilling

00:50:06 ◼   ► to yet talk about. The 3D point cloud stuff was really the most interesting. So there's

00:50:13 ◼   ► basically like three new, very, very vaguely or in simple terms, there's three types of

00:50:22 ◼   ► new data that they're putting into maps to make it better. One, they're putting in more

00:50:29 ◼   ► and more frequent probe data. Probe data has sort of been present in maps for a while,

00:50:37 ◼   ► but really not to this level. By probe data, I mean those segments of a trip that gets

00:50:43 ◼   ► sliced up and anonymized, right? So this data will give people much, hopefully much more

00:50:49 ◼   ► realistic ideas of what traffic is. The sort of implication they felt that I got from them

00:50:55 ◼   ► was that they feel that they'll do a better job than Waze or just as good. You know, there's

00:51:00 ◼   ► no real reason to even use Waze with this new traffic stuff, which is, you know, I mean,

00:51:05 ◼   ► that's one way to position it. And I think if you're going to think of another one out

00:51:09 ◼   ► there that's doing this, Waze is certainly the competitor. But then you have, so in addition

00:51:16 ◼   ► to probe data, you've also got high resolution satellite data. And then high resolution satellite

00:51:22 ◼   ► data, of course, provides them the ability to create foliage and to determine what's

00:51:27 ◼   ► a public pool and what's a baseball diamond and a tennis court and to label those. So

00:51:32 ◼   ► you're like, "Hey, meet me on the tennis court." And you can give directions right to the tennis

00:51:35 ◼   ► court, that sort of thing. And not just any tennis court, but this tennis court, the one

00:51:40 ◼   ► on the left. So they got all that. Then they have the cars, right? So the vans. So you've

00:51:47 ◼   ► get the van, this van's trundling around, it's capturing LIDAR information, which is

00:51:52 ◼   ► the same stuff that self-driving cars use to see the world around them. Basically what

00:51:56 ◼   ► they do, LIDAR does is it's a camera that whips around and takes a many, many, many

00:52:01 ◼   ► quote-unquote pictures a second and creates a sort of map of the world in 3D based on

00:52:09 ◼   ► points. So it says, "Hey, I just received a bounce back from this. I'm going to assign

00:52:14 ◼   ► of point at that distance in space from me, right? And then the same thing all the way

00:52:20 ◼   ► around many, many millions of times. And then you end up with a scenario where you got,

00:52:25 ◼   ► you know, oh, hey, that's a sign. It's yes, it's made up of, you know, 10,000 contiguous

00:52:31 ◼   ► points. But I know it's a flat sign because they're all the same distance away. And so

00:52:35 ◼   ► on and so forth for buildings and everything else.

00:52:38 ◼   ► Sort of sort of like the way do you think it's fair to say it's sort of like the way

00:52:40 ◼   ► that in visual effects like when you see like Andy Serkis performing one of his

00:52:45 ◼   ► you know roles and he's wearing a motion capture suit with dots on it so that

00:52:48 ◼   ► when they create the character the dots can be used to sort of recreate a

00:52:54 ◼   ► three-dimensional it's that but in reverse right what if the camera was

00:53:01 ◼   ► making the dots rather than the dots being painted on it right you know

00:53:03 ◼   ► because the world hasn't been conveniently painted with the

00:53:07 ◼   ► reflective dots to help the LIDAR and self-driving cars.

00:53:12 ◼   ► Ironically, they actually can tell the difference between retro-reflective surfaces and not,

00:53:17 ◼   ► which is one of the ways that they determine what's a sign, like a stop sign or a road

00:53:22 ◼   ► sign, because they're all reflective, right? And so it's one of the situations or cues,

00:53:28 ◼   ► visual cues that they use. But exactly right. You don't have the dots. You got to make them

00:53:32 ◼   ► yourself. You got to make this cloud yourself. And if you make a high-resolution enough cloud,

00:53:37 ◼   ► you know, I've seen the data, the raw data coming out of that and it looks like a gray,

00:53:42 ◼   ► gray untextured 3D model of the world, right? Like you would get early on in the process of

00:53:49 ◼   ► creating a 3D movie or animation or anything like that. Which provides rich data.

00:53:56 ◼   ► - It's like what Daredevil sees. - Yes, exactly. That's right. You're

00:54:02 ◼   ► getting this vision of the world that isn't textured with textures or color, but it's

00:54:07 ◼   ► certainly, you know, all of the distances, everything is away from you, you know the

00:54:12 ◼   ► shape of it, you know, all of that stuff. And then, well, I'll finish this, we can talk

00:54:22 ◼   ► about it. I'm interested to hear what you think of this. But the thing that they do

00:54:25 ◼   ► is they combine that three data, that point data, they combine it with a high resolution

00:54:40 ◼   ► images, take it from the car at a very near range. They're essentially Canon lenses. I

00:54:46 ◼   ► don't know what's under the hood, probably Canon bodies, but they're like fish eyes and

00:54:51 ◼   ► various focal lengths of lenses. So they get a nice overlap all the way around the car

00:54:55 ◼   ► as eight cameras. So they take these high resolution images and then they use those

00:55:02 ◼   ► to create like panos that the people can view, you know, similar to what a street view would

00:55:06 ◼   ► be, but not obviously exposed to the consumer. They did not announce street view and would

00:55:11 ◼   ► not tell me if they were going to do it. But the editors can see those panos, right? When

00:55:17 ◼   ► they go to say like, "Oh, was this a street sign say 35 or 40 or, you know, is this address

00:55:22 ◼   ► really on that building, they can look at the pano and pan around it and see exactly

00:55:25 ◼   ► what the car saw right there so they can make an edit call, right? And it's attached to

00:55:29 ◼   ► all of the data. Anonymized, of course, so, you know, license places and faces are blurred

00:55:35 ◼   ► out. But the other thing that they use this high-resolution camera data for, which I found

00:55:41 ◼   ► freaking amazing and I don't, I'm of, I feel like I should know if Google does this or

00:55:47 ◼   ► not, but I don't believe I do and I probably should look it up so that somebody doesn't

00:55:51 ◼   ► call me on it. But it's basically called an orthogonal recreation or reproduction. And

00:55:57 ◼   ► so what it does is it takes the 3D world that is created with that point cloud and then

00:56:02 ◼   ► takes the high resolution, very high resolution images because it shoots them in vertical

00:56:05 ◼   ► too, some of them. So that they're full, you know, full res all the way down the street.

00:56:12 ◼   ► And it maps those images onto the 3D objects to create texture. So when like an editor

00:56:19 ◼   ► is looking at a street from the top and trying to decide, "Now, does this street connect

00:56:26 ◼   ► underneath these tree canopy? I can't tell." Right? Like they're looking at a satellite

00:56:31 ◼   ► and they don't know whether the street stops or whether it goes all the way through to

00:56:35 ◼   ► the other side where the trees are covering everything in between. They have no idea,

00:56:38 ◼   ► no way to tell. Right? And so either, yeah, they can go into the pano and try and look

00:56:43 ◼   ► through the trees or whatever, or they can turn on the ortho reconstruction, which is

00:56:49 ◼   ► basically a full 3D scene, texture with color and picture and everything. Incredibly high res,

00:56:55 ◼   ► much higher res than satellite would be because of the distance, right? And it's mapped all onto the

00:57:01 ◼   ► 3D object and they could just look right under the trees. The trees disappear because the car could

00:57:08 ◼   ► see underneath the trees, so now the editor can. It's like having x-ray vision from street level

00:57:13 ◼   ► and powerful telescopic vision from above combined together.

00:57:18 ◼   ► It's pretty wild.

00:57:20 ◼   ► And I'm not saying they will or won't or whatever,

00:57:24 ◼   ► but you could easily see this being used

00:57:27 ◼   ► as a 3D reconstruction at street level,

00:57:30 ◼   ► much like flyover is used at the flyover level.

00:57:33 ◼   ► - Right.

00:57:34 ◼   ► So when, to me the big question, this sounds great,

00:57:40 ◼   ► the examples they showed before and after

00:57:42 ◼   ► of the Bay Area or look terrific.

00:57:45 ◼   ► The big question, and they didn't want us to give you any,

00:57:50 ◼   ► they didn't give you any timeline other than the Bay Area

00:57:53 ◼   ► in Northern California in the fall,

00:57:54 ◼   ► but how long do you think this is gonna take

00:57:56 ◼   ► to roll out everywhere?

00:57:58 ◼   ► I mean, 'cause it makes a big difference in the,

00:58:02 ◼   ► is this something to be excited about or not?

00:58:05 ◼   ► If this is-- - Right, absolutely.

00:58:07 ◼   ► Especially for people that live overseas,

00:58:09 ◼   ► they're like, yeah, that's fine, but look,

00:58:12 ◼   ► I don't even have transport yet, you know, or whatever.

00:58:14 ◼   ► But I, you know, I don't know, you know, obviously they didn't give any exact

00:58:18 ◼   ► timelines. It wouldn't commit.

00:58:19 ◼   ► The one thing that I can say is that they have teams all over the world so that

00:58:26 ◼   ► Apple Maps teams are not local only.

00:58:29 ◼   ► They are all over the world.

00:58:30 ◼   ► They do have cars running in many states already and overseas.

00:58:36 ◼   ► So they've got the, the, the, uh,

00:58:41 ◼   ► sort of recording mechanisms up and running as far as to how long it takes to

00:58:45 ◼   ► parse them and put them in. I don't know, but I can't say this,

00:58:50 ◼   ► like it didn't take them four years to put this into the map, right?

00:58:53 ◼   ► It took them four years to build the processes and tools.

00:58:56 ◼   ► So my guess is that they get Northern California up and running,

00:59:01 ◼   ► they get Southern California up and running,

00:59:02 ◼   ► and then everything else starts to iterate fairly quickly from there.

00:59:05 ◼   ► because it's basically a matter of, you know, employing more and more trained

00:59:09 ◼   ► editors with good tools to make sure that the information is accurate and

00:59:14 ◼   ► then putting more vans on the ground. And these are just vans, man, with a

00:59:17 ◼   ► Mac Pro in the back. They're not, you know, yeah, the stuff on top is pricey,

00:59:21 ◼   ► but it's not beyond Apple's reach. This is not a matter of lack of resources at

00:59:25 ◼   ► all.

00:59:25 ◼   ► Right. Yeah. Well, that's, I mean, that's the question. I think 10,000 Apple

00:59:32 ◼   ► podcasters have asked many times on various shows is, you know, if maps are important

00:59:39 ◼   ► and Apple maps are behind and Apple is literally the richest, the wealthiest company in the

00:59:45 ◼   ► world, how can they not spend, you know, can't they, isn't there a way that they can spend

00:59:49 ◼   ► their way out of this? And the gist I get from the article is they're trying and it's

00:59:55 ◼   ► It's just, it takes time, and so like,

00:59:58 ◼   ► you know, all the money that they're putting

01:00:01 ◼   ► into the vans that drive around,

01:00:04 ◼   ► we haven't seen the results of it until today.

01:00:07 ◼   ► We now have an answer of what in the world

01:00:09 ◼   ► are they doing with all these vans,

01:00:11 ◼   ► because I don't see anything in my app

01:00:14 ◼   ► that came from them. - Right, right.

01:00:16 ◼   ► - Right, and the answer is, we've been,

01:00:17 ◼   ► you know, it takes time.

01:00:19 ◼   ► Even if you have the money, there's no way to,

01:00:22 ◼   ► you know, and it sounds to me--

01:00:23 ◼   ► - Yeah, it sounds like, look--

01:00:24 ◼   ► sounds to me like they're spending as much as you as we would hope Apple would spend on this.

01:00:28 ◼   ► Right. I mean, you if you have all the resources in the world, your limitations cease to become the

01:00:35 ◼   ► resources, right? The bottlenecks are elsewhere. And I think when people have the discussion about

01:00:40 ◼   ► why can't Apple spend more money at this or that? It's very similar to people, you know,

01:00:45 ◼   ► complaining to a developer, but like, why aren't you fixing this when they're these things? You got

01:00:50 ◼   ► these things going on, you're like, well, yeah, but we have, you know, 60 developers,

01:00:55 ◼   ► and they're all working on different things all at once. Just because we're updating this

01:01:00 ◼   ► feature doesn't mean we're not working on that feature, right? And I think that's like

01:01:03 ◼   ► the constant refrain from somebody outside looking at an organization like Apple. They're

01:01:07 ◼   ► like, why are you wasting your time on X when you could be working on Y? It's like, well,

01:01:10 ◼   ► they are working on Y. Like they're, they are investing in Y. It's just you're not seeing

01:01:15 ◼   ► the results of Y for one reason or another. And yes, that reason could be dysfunction,

01:01:19 ◼   ► right? It could be like poor management or lack of direction or whatever the case may be.

01:01:25 ◼   ► You know, I've heard many tales of that over the years with Apple Maps, just how, you know,

01:01:30 ◼   ► it's gone from one person to another and, you know, someone takes responsibility, other people do.

01:01:34 ◼   ► You have the many librarians, you know, scenario where somebody comes in and decides to change it

01:01:41 ◼   ► all. And, you know, then you have to kind of reset all of that. I've heard all those tales,

01:01:44 ◼   ► right? Just like in many other large companies, you get some dysfunction sometimes.

01:01:49 ◼   ► But sometimes it's no matter how much money you throw at it,

01:01:52 ◼   ► the bottlenecks exist outside of the realm of money to solve.

01:01:56 ◼   ► And it just takes the other limited resource, which is time.

01:02:00 ◼   ► No. Um, I think that wraps up the maps discussion.

01:02:05 ◼   ► Any other points on maps you want to make? Where'd you get to sit?

01:02:09 ◼   ► You got to ride around in one of the vans. Where, where did you get to say,

01:02:12 ◼   ► uh, shotgun. Yeah. So where, where the operator would be to like tell the,

01:02:18 ◼   ► I guess one person is just supposed to focus on driving.

01:02:22 ◼   ► And the other person's like, okay, go left here,

01:02:25 ◼   ► or go right there, and you know,

01:02:27 ◼   ► oh, let's circle around this parking lot

01:02:29 ◼   ► 'cause we gotta get the inside of it,

01:02:30 ◼   ► or whatever the case, yeah.

01:02:32 ◼   ► - Very fun.

01:02:33 ◼   ► - Yeah, pretty utilitarian bands, though.

01:02:35 ◼   ► I don't know if I'd wanna do it for a living.

01:02:37 ◼   ► (laughing)

01:02:40 ◼   ► - I think it takes a certain kind of personality.

01:02:44 ◼   ► - Yeah, it's probably relaxing.

01:02:45 ◼   ► I mean, you know, you've got your job,

01:02:47 ◼   ► So here's your little parcel of land to cover and go at it.

01:02:51 ◼   ► I don't know if I'd want to do it all the time.

01:02:53 ◼   ► - For sure.

01:02:54 ◼   ► - For a living.

01:02:54 ◼   ► - Well, I'm glad that someone's doing it though.

01:02:57 ◼   ► - Yeah.

01:02:57 ◼   ► - Well, you can kind of see though,

01:02:59 ◼   ► that why some of the companies,

01:03:02 ◼   ► the companies that have these vans doing the napping work

01:03:05 ◼   ► also happen to be the companies that are either rumored

01:03:08 ◼   ► or known to be working on autonomous vehicles.

01:03:12 ◼   ► - Yeah, I mean, it makes a hell of a lot of sense, right?

01:03:15 ◼   ► You're driving all the streets,

01:03:16 ◼   ► might as well gather the data. I've seen autonomous data being gathered and it doesn't look a whole

01:03:23 ◼   ► lot different, let's put it that way. And obviously they're not talking about it and all of that, but

01:03:28 ◼   ► the autonomous data is gathered in much the same way to give those cars and those systems

01:03:34 ◼   ► a picture of the world. Well, and the other way around where those cars could collect the

01:03:41 ◼   ► data that would be continuously updating the map, right? Like, right, let's just say that

01:03:47 ◼   ► in theory that there is an Apple car that comes out five years from now or 10 years

01:03:53 ◼   ► from now, and it's an autonomous self driving vehicle, and therefore is using LIDAR and

01:03:58 ◼   ► other technologies to see the world around it. That data could be used in the same way

01:04:03 ◼   ► that they're using iPhones now to collect data about the world. Obviously, the data

01:04:07 ◼   ► collected by those cars would help make the maps better and it needs the cars

01:04:11 ◼   ► need the maps to do what they're supposed to do you know it's like a

01:04:14 ◼   ► virtuous circle exactly and and that's one of the reasons you know that it's

01:04:19 ◼   ► one of the things Tesla has stated about its cars is that it's been good they've

01:04:23 ◼   ► been gathering data on the road and they've got that network effect going on

01:04:26 ◼   ► right sure it's helpful for a company that wants to make autonomous cars to

01:04:29 ◼   ► have maps and it's helpful for a company that has maps to have their own

01:04:33 ◼   ► autonomous cars.

01:04:35 ◼   ► Yep, exactly, exactly.

01:04:38 ◼   ► And then the one other thing, like from the maps thing, just to touch on the last thing,

01:04:43 ◼   ► is that, just so everybody knows, I mean I mentioned it in the article, but just so everybody

01:04:48 ◼   ► knows, I did ask them about the whole AR directions thing, you know, and they were just not forthcoming

01:04:53 ◼   ► at all, which is totally understandable if they don't want to talk about it yet, but

01:04:56 ◼   ► like, all of the data that they're gathering, all of the stuff that they're doing, absolutely,

01:05:01 ◼   ► hundred percent completely possible that they could position an arrow in midair,

01:05:06 ◼   ► you know, for somebody to look at. But also in my opinion,

01:05:10 ◼   ► completely a hundred percent impractical to do for a phone.

01:05:14 ◼   ► Like who gives a crap? Like, you know,

01:05:16 ◼   ► I don't want to hold my phone up in midair for more than a second,

01:05:19 ◼   ► much less to follow an arrow.

01:05:20 ◼   ► Well, I wouldn't want to do it for a long time,

01:05:23 ◼   ► but think about like walking around an unfamiliar city and,

01:05:27 ◼   ► and you get maps now and you've long had the ability to hit the one,

01:05:30 ◼   ► There's a button in the maps app while you're getting walking directions that

01:05:33 ◼   ► sort of changes from always putting north at the top to,

01:05:37 ◼   ► to orienting the map in the direction you're going, right? And that is,

01:05:42 ◼   ► that is a very early days version of AR, you know,

01:05:47 ◼   ► the fact that the map can automatically rotate POV, right.

01:05:51 ◼   ► To put your current walking direction at the top of the map,

01:05:56 ◼   ► whether it's north, south, east, or west is, you know,

01:06:00 ◼   ► poor man's AR, whereas having that button use the camera

01:06:05 ◼   ► and just quickly tell me, like, am I going the right way?

01:06:08 ◼   ► You know, like sometimes you're in a city

01:06:09 ◼   ► and you'd say such and such street

01:06:11 ◼   ► and you can't tell which way the numbers go

01:06:13 ◼   ► and so you just take a guess and start going left

01:06:16 ◼   ► and it takes, you know, takes a while to figure out,

01:06:18 ◼   ► like, oh, nope, you should have gone right,

01:06:19 ◼   ► you gotta backtrack.

01:06:20 ◼   ► If you had a little camera you could point around

01:06:22 ◼   ► at the world and it would just quick put an arrow,

01:06:24 ◼   ► like, here, this way, it would be great.

01:06:27 ◼   ► I mean, and you wouldn't, you know, again,

01:06:29 ◼   ► You wouldn't walk around New York City for 15 minutes

01:06:32 ◼   ► with your phone in front of your hand,

01:06:33 ◼   ► but you could maybe just quick drop into that view

01:06:36 ◼   ► just to make sure you're going the right way on Broadway.

01:06:41 ◼   ► - Yeah, I'd get it, absolutely.

01:06:44 ◼   ► And if you get there, it's like an arrow pointing

01:06:46 ◼   ► at the door, there's a lot of cool opportunities,

01:06:49 ◼   ► for sure, on the phone.

01:06:51 ◼   ► But I think that the major opportunities won't come

01:06:55 ◼   ► until we have goggles, a heads-up view of some sort.

01:06:58 ◼   ► some sort of heads up view, which could be the dashboard of, you know,

01:07:01 ◼   ► the windshield car. Oh, a hundred percent. Yeah. Built in HUD in a car.

01:07:05 ◼   ► You got it. Right. Um, all right.

01:07:09 ◼   ► I have a couple more things I want to talk about, but, uh,

01:07:12 ◼   ► why don't I take a break and thank our third and final sponsor of the show.

01:07:17 ◼   ► Uh, hello, pillow H U L L O. Uh,

01:07:22 ◼   ► have you ever tried a buckwheat pillow?

01:07:24 ◼   ► They are totally different than the fluffy soft pillows that most of us are used to.

01:07:29 ◼   ► It's similar to a bean bag, very heavy too compared to a regular pillow.

01:07:33 ◼   ► And I really mean it because we've had we've had them for years here in the during fireball

01:07:38 ◼   ► world headquarters.

01:07:41 ◼   ► Very similar to a bean bag.

01:07:43 ◼   ► And it allows the pillow to adjust its shape and thickness.

01:07:47 ◼   ► It supports your head neck how you want it to unlike a traditional squishy soft pillow.

01:07:53 ◼   ► If you're the sort of person who always uses two pillows

01:07:56 ◼   ► because one pillow is not enough,

01:07:57 ◼   ► you are probably gonna be very surprised

01:08:00 ◼   ► if you try a hello pillow

01:08:01 ◼   ► how you don't need two pillows anymore

01:08:03 ◼   ► because it does the thing

01:08:04 ◼   ► that you were trying to use two pillows to do,

01:08:07 ◼   ► one hello pillow does by itself

01:08:08 ◼   ► by having this sort of firmer structure.

01:08:13 ◼   ► And it also stays cool and dry compared to pillows

01:08:16 ◼   ► that are filled with feathers or foam or whatever else.

01:08:20 ◼   ► Most pillows absorb and retain body heat and moisture,

01:08:24 ◼   ► making your pillow feel warm and humid.

01:08:25 ◼   ► Buckwheat tends to breathe better,

01:08:27 ◼   ► 'cause there's air between the individual buckwheats

01:08:30 ◼   ► in the pillow.

01:08:31 ◼   ► No more flipping.

01:08:33 ◼   ► You don't have to flip to get to the cool side of the pillow

01:08:35 ◼   ► in the middle of the night.

01:08:36 ◼   ► The pillow stays cool on one side.

01:08:39 ◼   ► We've had them for years.

01:08:41 ◼   ► Everybody in the house, all three of us,

01:08:45 ◼   ► use and like them.

01:08:46 ◼   ► My wife and son in particular,

01:08:49 ◼   ► truly love them and actually get irritated when we travel and they have to use regular

01:08:54 ◼   ► pillows. They're not the sort of thing that you could take with you. They are sort of

01:09:00 ◼   ► heavy compared to a regular pillow. But if you could, they would. That's how much they

01:09:04 ◼   ► like them more than regular traditional pillows. And I know a lot of people who have been sponsoring

01:09:10 ◼   ► the show for years and I know that there are a lot of readers who have had them. And I'll

01:09:14 ◼   ► tell you, you take it out at first and it does seem so different than a traditional

01:09:18 ◼   ► you're like, "I don't know about this." But I'll tell you, it's really different.

01:09:22 ◼   ► And this is the thing, it's not like they invented this. People have been sleeping on

01:09:28 ◼   ► buckwheat pillows for centuries. They've been used in Japan extensively and remain popular

01:09:34 ◼   ► to this day there. So it's really sort of Eastern/Western culture difference that we've

01:09:39 ◼   ► had these fluffy, feather-filled type pillows and haven't had these. But it really is a

01:09:45 ◼   ► a more natural way to sleep. And if you're at all dissatisfied with your current pillow

01:09:49 ◼   ► situation, I really recommend Hello, give you know, give them a try. And they're made

01:09:54 ◼   ► right here in the USA with quality construction and materials and certified organic cotton

01:10:00 ◼   ► case that is cut and sewn for durability. And the buckwheat is grown and milled in the

01:10:05 ◼   ► United States. Like I said, I don't know how many years it's been since we've had them.

01:10:08 ◼   ► But like my wife is still on the first one they sent us. And it seemed I don't think

01:10:12 ◼   ► it seems any different than the day that we started with it.

01:10:15 ◼   ► So here's the deal that you can get as a listener of this show. Sleep on this pillow for 60

01:10:20 ◼   ► nights. 60 nights! And if hello is not for you, just send it back and they will give

01:10:25 ◼   ► you a full refund. So you cannot lose. Go to hellopillow.com/talkshow. Hellopillow.com/talkshow.

01:10:35 ◼   ► And if you try more than one pillow, you get a discount of up to 20 bucks per pillow depending

01:10:41 ◼   ► on this size. Fast free shipping on every order and 1% of all profits are donated to

01:10:49 ◼   ► the Nature Conservancy. My thanks to HelloPillow. HelloPillow.com/talkshow.

01:10:58 ◼   ► Two things I want to talk about. I want to talk about Duplex and I want to talk about

01:11:00 ◼   ► the MacBook keyboard repair program. Duplex, the news this week is that Google did exactly

01:11:07 ◼   ► what I had been saying. I wish and don't understand why they didn't do in early May when they announced

01:11:15 ◼   ► Duplex, which is actually allow journalists to not just listen to live calls, but actually

01:11:23 ◼   ► participate in them. So what they did is they set up, they actually rented out, bought out for the

01:11:28 ◼   ► day, some restaurants in San Francisco and New York, different groups of journalists,

01:11:34 ◼   ► and had them, you know, had the journalist play receptionist to answer the phone and talk to Google

01:11:41 ◼   ► duplex. And to a T, I don't know if I read every one of them, but I read a lot of them

01:11:49 ◼   ► to a T. People said that it sounds, you know, the the ums and ahs and those those things that

01:11:57 ◼   ► that the duplex voice was doing in the recording Google played at IO, that it sounds just as eerily

01:12:05 ◼   ► human-like in real life. Although Google did not release any of the actual recordings of

01:12:10 ◼   ► the journalist's phone calls. Your thoughts? Yeah, I mean, it is what it is. You know what I mean?

01:12:21 ◼   ► It's like, it always cracks me up with these companies. I love how much money they have,

01:12:25 ◼   ► you know, to like create these publicity scenarios. Like, "Hey, we're just here in this casual Thai

01:12:31 ◼   ► restaurant." You know, I think it's fine to kind of go the next step and say, "Look, it does work,

01:12:38 ◼   ► and here's how it is," and put you on the phone. They should have done that from the beginning,

01:12:41 ◼   ► but, you know, maybe they weren't ready and all that, fine. You know, or maybe they didn't

01:12:44 ◼   ► anticipate the fervor, you know, that people were under. Even the impressions that Brian Heater

01:12:51 ◼   ► went to ours, one of our writers, he went to the one in New York. He's been covering

01:12:57 ◼   ► duplex in this whole issue since the beginning. And so he was the ideal person to kind of

01:13:03 ◼   ► judge like, "Oh, okay, does this feel the same?" and all of that. And he said it definitely

01:13:07 ◼   ► feels like it works. You know, the disambiguations that they put in definitely make it feel more

01:13:13 ◼   ► human. But he pointed out two things rightfully, which is one, you can't say that you're not

01:13:18 ◼   ► trying to trick people and that you, you know, that you also want them to believe it's more

01:13:24 ◼   ► human. Like those two things are incompatible. You know, I thought the same thing. And I

01:13:30 ◼   ► think that it gets to where so many people, when I raised some questions about the way

01:13:38 ◼   ► that this was demoed at IO and said that, I guess I shouldn't have phrased it the way

01:13:45 ◼   ► that I did. And I said, what they've done just by playing a record, all only playing

01:13:49 ◼   ► a recording of two ostensible, quote unquote, actual calls. That was, that was the words

01:13:55 ◼   ► that they use by only playing recordings and not letting anybody see it live. Um, it is

01:14:03 ◼   ► indistinguishable from, you know, like a fake and that what some people jumped on that and

01:14:09 ◼   ► took it as me saying that I think this is fake or completely fake, you know, that there

01:14:15 ◼   ► were actors that they weren't, they weren't even, it wasn't even a computer generated

01:14:18 ◼   ► voice or something like that. And that wasn't what I was saying at all. You know, and I

01:14:22 ◼   ► thought I emphasized that by saying that, look, if anybody in the world, any company

01:14:26 ◼   ► or institution, you know, in today's world, it's companies like Google and, and Amazon

01:14:34 ◼   ► that do things like this. Maybe, you know, a generation ago it was research labs, you

01:14:37 ◼   ► know, like at MIT or Bell Labs or something like that. But if there's any institution

01:14:42 ◼   ► that today could could make a voice that sounds like that, it would be Google. I think that

01:14:48 ◼   ► would be the, you know, they would have the best odds in Vegas. But it's just such a curious

01:14:54 ◼   ► and bizarre way to roll it out. Because it was it. There was no nobody was shown any

01:14:59 ◼   ► evidence of it. And I think in hindsight, I think it's clear. It's not that it was.

01:15:04 ◼   ► clearly wasn't a fraud. Clearly, they have this vocal technology, right? And it is amazing. It is

01:15:09 ◼   ► a leap forward to be able to have a computer generated voice that can fool people into

01:15:14 ◼   ► thinking it's real, even if they're not just like, hey, you, you're just the, the, the person who

01:15:23 ◼   ► answers the phone at the restaurant. And, you know, you answer the, you answer, you know, a

01:15:27 ◼   ► 20, 30, 40, maybe 100 calls every day, because, you know, that's how many reservations you get.

01:15:32 ◼   ► So even if you're not thinking, hey, you know, this might be, you know, an AI, and you just

01:15:40 ◼   ► happen to not even notice and you have the phone call go come and go and without even giving it a

01:15:45 ◼   ► second thought that you might have been talking to an AI bot. Even if you're thinking about it,

01:15:51 ◼   ► like the journalists who are invited to this demo, like Lauren Good from wired, successfully tripped

01:15:57 ◼   ► tripped it up by asking about, I thought her anecdote was terrific. And it comes from the

01:16:03 ◼   ► fact that she actually in college had a job answering the phone at like a bar or restaurant

01:16:08 ◼   ► or something. And she just asked if anybody needed, if there are any kids and do they

01:16:11 ◼   ► need high chairs and does anybody have any food allergies or something like that? And

01:16:16 ◼   ► those things tripped it up and a human operator had to jump in and she thought that she could

01:16:22 ◼   ► tell a second voice came on, but she thought maybe they switched to a different, you know,

01:16:27 ◼   ► she wasn't sure that it's she didn't think it is switched to a human. She just thought

01:16:30 ◼   ► it was another device, you know, and I thought that was really telling about the compelling

01:16:34 ◼   ► this of the voice. But you know, Brian heaters exactly right. You can't say that you've built

01:16:39 ◼   ► this thing that is indistinguishable from a human and then say that we're not trying to

01:16:43 ◼   ► and and they even said that we took it in this direction because as we tested it, we

01:16:50 ◼   ► got fewer hang-ups this way. Right? Yeah, no kidding. If they think it's a human,

01:16:56 ◼   ► they won't hang up as much. No. And I'm not saying this in the sense of, oh, those

01:17:01 ◼   ► bastards at Google are tricking people. I'm just saying that they haven't, when they

01:17:09 ◼   ► announced this in early May, they clearly had not thought this through as a product.

01:17:15 ◼   ► they had was an amazing technology. And I think they wanted to demo this amazing technology

01:17:21 ◼   ► because that's what that's something Google really likes to do is really show off technology.

01:17:27 ◼   ► And they just sort of thought yada yada yada will, you know, we'll, we'll, we'll turn it

01:17:33 ◼   ► into a product and Google Assistant or something like that without having really thought it

01:17:36 ◼   ► through. And even at this point with the demo that these people had, it still isn't actually

01:17:41 ◼   ► hooked up to Google Assistant. Like, so the way that the call starts, even in these, the

01:17:47 ◼   ► canned demos was not by somebody talking into their phone and saying, Oh, okay, Google,

01:17:55 ◼   ► make me a reservation at this restaurant at seven or eight o'clock on Saturday for a table

01:18:01 ◼   ► for four and then, and then letting go of the phone. And at some, you know, five, six

01:18:08 ◼   ► seconds later have the phone ring it at the place it didn't work like that at all and

01:18:13 ◼   ► then at the end of the call they didn't show them like an appointment showing up in their

01:18:18 ◼   ► Google calendar with the reservation right they only have the part of the thing where

01:18:25 ◼   ► there's like a guy at a computer who types things a Google of you know engineer who initiates

01:18:30 ◼   ► the call. Like it's not a full system yet. Right. And at this point, they say that four

01:18:38 ◼   ► out of five calls that they do to these restaurants, success, you know, successfully get the right

01:18:43 ◼   ► reservation. And that one out of five requires a human operator to jump in. And like, there's

01:18:52 ◼   ► like a quick transcript. So like the human operator apparently can see a transcript of

01:18:56 ◼   ► what's gone on so far, what the what the request was, and then take over. That that's amazing.

01:19:03 ◼   ► Four out of five is absolutely amazing. But it's absolutely they've provided the dashboard

01:19:08 ◼   ► for the in person, right? That means they're productizing it. But it's absolutely untenable

01:19:13 ◼   ► at anywhere near that rate, for Google to actually roll this out to everybody. It's

01:19:19 ◼   ► absolutely untenable, they would need to have an army and absolute army of humans, even

01:19:24 ◼   ► at a 90% success rate to pop in.

01:19:27 ◼   ► Like this has to be--

01:19:28 ◼   ► - Now is it the Google operator that takes over though

01:19:30 ◼   ► or is it the person at the restaurant?

01:19:32 ◼   ► - No, it's a Google operator.

01:19:33 ◼   ► In other words, like if you--

01:19:35 ◼   ► - Got it.

01:19:36 ◼   ► - It's a Google employee.

01:19:37 ◼   ► Google says that they have, you know,

01:19:39 ◼   ► but nothing Google does ever requires an army of humans.

01:19:44 ◼   ► Maybe other-- - No.

01:19:45 ◼   ► - You know, it just doesn't scale.

01:19:46 ◼   ► - They would consider that a failure.

01:19:48 ◼   ► - Right.

01:19:48 ◼   ► Like you can do it if it's a paid service.

01:19:53 ◼   ► Um, virtue member of virtue, the, the, the phone company that sold five or $6,000 phones,

01:20:00 ◼   ► they'd gold blackberries or whatever. Right. And they'd have like leather backing and,

01:20:05 ◼   ► and they were very bling at the Nokia. They, they originally ran on Nokia's stuff and Nokia

01:20:11 ◼   ► was an investor in the company. And at some point while they tried to hang on in relevance,

01:20:16 ◼   ► know, they switched to Android. But they had a button on the phone. And you paid five or

01:20:24 ◼   ► six thousand dollars for a phone and it was just running. It was technically no better

01:20:29 ◼   ► than like a $25 Nokia phone. But part of what you got was this bling on the, you know, the

01:20:37 ◼   ► actual phone itself was very blingy and you know, had some premium materials like gold

01:20:42 ◼   ► and leather and stuff like that. But one of the things you also got was access to a concierge

01:20:46 ◼   ► service where you just hit a button and it would just immediately connect you to a virtue concierge

01:20:52 ◼   ► and you would say, make me a reservation at, you know, whatever restaurant and, you know,

01:20:59 ◼   ► on Saturday at seven, a table for five. And they'd say, okay, you know, Mr. Panzorino. And,

01:21:06 ◼   ► you know, like when the reservation comes through, they would, I don't know if they sent you a text

01:21:09 ◼   ► or what but there'd be some kind of thing that would let it go but it was a human on the other end and

01:21:13 ◼   ► They obviously they could afford to do that because they were selling

01:21:17 ◼   ► $5,000 phones and they were very it costs like $30 right plus of gold and

01:21:22 ◼   ► You know, it was not a you know, there weren't millions of virtue users

01:21:27 ◼   ► There were I don't know I guess thousands, you know, Google can't afford to have something like that

01:21:31 ◼   ► For even 10% of these calls because it's a free service. Google Assistant is completely free. I

01:21:38 ◼   ► I think it's look, I think the whole thing is cool. The concept is cool as a person who likes technology and you know, is

01:21:46 ◼   ► Whose job is ostensibly to absorb bleeding-edge technology and kind of like figure out what it means and all that

01:21:53 ◼   ► I think it's super cool and a great technical achievement all of that certainly not scalable at that error rate

01:21:58 ◼   ► But I think they could get that error rate down for sure which means it's just not launchable yet

01:22:03 ◼   ► But it probably will be at some point. I'm not sure

01:22:07 ◼   ► that? I'm not sure. I would I would I mean, I'll, I'll take that bad, I guess, you know what I mean?

01:22:13 ◼   ► But I get why you say maybe not, you know, maybe they'll never be able to get it there,

01:22:17 ◼   ► because there's just too many edge cases or whatever. But I just have, I just know a lot

01:22:22 ◼   ► about or little about a lot of machine learning stuff. And I think they could get it there,

01:22:28 ◼   ► because they could train it on, they could probably buy banks of recorded conversations from,

01:22:34 ◼   ► like call centers that say you're being you're now being recorded right and they could train on that

01:22:38 ◼   ► i think they could do it but regardless of that you have this sort of secondary thing about it

01:22:43 ◼   ► which is should we you know like should we be talking to robots all the time and you know is

01:22:49 ◼   ► that really going to help um i don't know i just think that this really falls square into that pool

01:22:56 ◼   ► of um you know products or or startups or features or whatever which is just really about making it

01:23:04 ◼   ► so that man boy engineers don't have to talk to people.

01:23:08 ◼   ► Like it's really, you know,

01:23:14 ◼   ► there's a whole category of startups

01:23:17 ◼   ► that fall into this arena.

01:23:19 ◼   ► And some of them have broken out of that

01:23:22 ◼   ► and have genuinely become useful to many, many people

01:23:24 ◼   ► for a variety of reasons.

01:23:25 ◼   ► You know, we just ordered Uber Eats

01:23:27 ◼   ► because it's like the baby's asleep

01:23:29 ◼   ► and I'm the only parent home, right?

01:23:32 ◼   ► So, okay, we'll order Uber Eats.

01:23:34 ◼   ► Totally useful.

01:23:35 ◼   ► Originated with people going like, you know,

01:23:38 ◼   ► food delivery services, some engineer going,

01:23:40 ◼   ► why do I have to get up and go to the thing

01:23:42 ◼   ► and stop coding?

01:23:44 ◼   ► My mom stopped bringing me food now,

01:23:45 ◼   ► so I should invent this company

01:23:47 ◼   ► that has somebody else bring me the food, right?

01:23:50 ◼   ► And I think there are plenty of opportunities

01:23:54 ◼   ► for those things to break out and become something useful,

01:23:57 ◼   ► but there's also plenty of them that feel like cop-outs.

01:24:00 ◼   ► They feel like we're sort of checking out of the societal arrangement because we can.

01:24:07 ◼   ► We have the privilege to do so.

01:24:09 ◼   ► We could pay for it or we have the technology.

01:24:11 ◼   ► We have the expensive phone that has the service or whatever, right?

01:24:16 ◼   ► So I think there's that aspect of it that I'm a little bit leery on.

01:24:19 ◼   ► Dave Asprey The other thing that makes me wonder if this

01:24:21 ◼   ► is ever going to be a product is that Google is saying that it's going to be opt-in for

01:24:24 ◼   ► the businesses.

01:24:26 ◼   ► So if you and I open up a restaurant and we do take reservations but we don't take open

01:24:31 ◼   ► table, we would have to sign up to be a duplex-ready restaurant because of these privacy implications

01:24:41 ◼   ► that people raised.

01:24:42 ◼   ► I'm not so sure that's necessary.

01:24:44 ◼   ► I wrote on Daring Fireball, I'm not convinced that it's a problem whether this thing identifies

01:24:50 ◼   ► itself as a robot or not.

01:24:54 ◼   ► my analogy was like, is that there's no loss, right? It these phone calls, because it's

01:24:59 ◼   ► a robot don't cost the restaurant any more time. They don't bother they go well enough

01:25:04 ◼   ► when they go well, that it's not the person answering the phone hasn't been hasn't lost

01:25:10 ◼   ► anything compared to if I personally just called them and made the same reservation.

01:25:15 ◼   ► So what's the argument against having a robot do it, whether the person on the other end

01:25:19 ◼   ► knows it's a robot or not. Compare and contrast to say, poker playing bots that go online.

01:25:26 ◼   ► And right, you know, you think you're playing against a bunch of other humans, but in fact,

01:25:30 ◼   ► you know, one of the opponents is a computer and is taking your money. Like, that's, that's

01:25:37 ◼   ► where you you thinking that it's a human but it's actually a bot is deceptive and wrong

01:25:43 ◼   ► morally and ethically wrong. I'm not so sure that this is but if they really are going

01:25:48 ◼   ► I mean, keep it opt-in.

01:25:49 ◼   ► I don't know how that's ever gonna take off.

01:25:51 ◼   ► I mean, I guess they could just

01:25:53 ◼   ► make a, launch a big publicity campaign

01:25:56 ◼   ► to get restaurants to do it, but I don't know.

01:25:59 ◼   ► And I'm also not convinced--

01:26:00 ◼   ► - Yeah, I mean, they're pretty adept, though,

01:26:02 ◼   ► at selling people into this channel.

01:26:04 ◼   ► I mean, they do have sales teams

01:26:07 ◼   ► that work on the programmatic side and that sort of thing.

01:26:09 ◼   ► So I think they can get there.

01:26:11 ◼   ► - I just think that if you're gonna have a computer,

01:26:13 ◼   ► talk to a restaurant and make a reservation.

01:26:17 ◼   ► doing it as an AI that talks on a phone call is such a sort of Rube Goldbergian difference

01:26:25 ◼   ► compared to like the way that they even say that if the restaurant takes OpenTable, they'll

01:26:31 ◼   ► just use OpenTable APIs and do it that way. Right? Like it doesn't make sense to build

01:26:36 ◼   ► computers that talk to each other by voice.

01:26:38 ◼   ► Jay Haynes Yeah, exactly.

01:26:40 ◼   ► Jon Moffitt Right?

01:26:41 ◼   ► Jay Haynes Yeah, we just see an API for everything and

01:26:44 ◼   ► And then, you know, everybody has their personal API interacts with all of the global API's

01:26:49 ◼   ► that we never have to talk to anybody.

01:26:50 ◼   ► Right.

01:26:51 ◼   ► Like for cinematic regions reasons are to D2 and C3PO, you know, talked to beep to each

01:26:56 ◼   ► other, but it doesn't make any sense that they wouldn't communicate completely like

01:27:01 ◼   ► by whatever the Star Wars equivalent of wifi.

01:27:03 ◼   ► Yeah, yeah, they would.

01:27:05 ◼   ► Exactly.

01:27:06 ◼   ► Or even if they use auto audible signals, it would be some sort of like rapid fire screech

01:27:10 ◼   ► of ones and zeros that like happened in a blink of an eye, you know, while I mentioned that,

01:27:15 ◼   ► let me just add this. I've been meaning to mention this on the show for a while.

01:27:18 ◼   ► I think one of the most prescient things that George Lucas, little details that George Lucas

01:27:23 ◼   ► came up with, uh, was the fact that when in the first star Wars movie, when they go into the

01:27:29 ◼   ► cantina and the bartender says to Luke, we don't serve their kind here very angrily. You know,

01:27:34 ◼   ► in other words, droids aren't welcome. They're sort of like a anti-droid, uh, bigotry in the

01:27:40 ◼   ► Star Wars universe. And I think that was so prescient because as the viewer of the movie,

01:27:44 ◼   ► all of us are thinking a human or adult children alike are thinking, Oh my God, these robots seem

01:27:50 ◼   ► so realistic, right? C3PO. It was like amazing. It was like he walks like a robot looks like a

01:27:56 ◼   ► robot. Wouldn't it be cool to have a robot like that, that can just talk to you like this.

01:28:00 ◼   ► We're thinking, man, that's cool. But in their universe, they're sort of resented by at least

01:28:06 ◼   ► some people, you know, and like the sort of degenerates who would go to the cantina. And

01:28:10 ◼   ► I think we're seeing…

01:28:11 ◼   ► Yeah, as a kid you're going, "Oh, that's so cool." And as an adult you're like,

01:28:14 ◼   ► "Oh, robots, what assholes."

01:28:15 ◼   ► Yeah. And that's what we're sort of seeing that with people's reaction to things like

01:28:19 ◼   ► duplex. All right, enough on that. Last thing, I don't have a lot to say about this, but

01:28:24 ◼   ► last but not least, Apple has announced finally a MacBook keyboard repair program. It launched

01:28:32 ◼   ► around five o'clock last Friday.

01:28:34 ◼   ► (laughing)

01:28:36 ◼   ► Imagine that.

01:28:36 ◼   ► It must have just been when it was ready, John.

01:28:38 ◼   ► - Yeah, I'm sure.

01:28:39 ◼   ► - It just was, that just must have been

01:28:41 ◼   ► when they were ready to announce it.

01:28:42 ◼   ► - I'm sure that's exactly why.

01:28:44 ◼   ► - The moment they were ready, they launched the news.

01:28:47 ◼   ► The moment, very moment.

01:28:48 ◼   ► - I think it was a little bit before five.

01:28:51 ◼   ► I think it was around like three Eastern

01:28:54 ◼   ► or something like that on a Friday in late June.

01:28:57 ◼   ► But more or less, and it is funny,

01:29:01 ◼   ► 'cause it's like, all right, here's the affected devices.

01:29:05 ◼   ► And it's effectively, it's every single,

01:29:07 ◼   ► it is, they're not even effectively,

01:29:08 ◼   ► it's just literally every single MacBook or MacBook Pro

01:29:11 ◼   ► that's come out since the new keyboard design came out.

01:29:14 ◼   ► It's like if you've got a MacBook

01:29:16 ◼   ► with the butterfly key switches,

01:29:18 ◼   ► it's eligible for this repair program.

01:29:21 ◼   ► And if you get a stuck key or any of these other problems

01:29:26 ◼   ► people are running into with this,

01:29:27 ◼   ► it is eligible for a free repair.

01:29:29 ◼   ► they'll either repair the,

01:29:30 ◼   ► to replace the whole keyboard or replace a single key,

01:29:33 ◼   ► if that's what they can do.

01:29:35 ◼   ► And if you've already paid for such a repair

01:29:38 ◼   ► and think it should have been covered,

01:29:39 ◼   ► there's some kind of thing you can go through

01:29:43 ◼   ► to get your money back.

01:29:44 ◼   ► Meanwhile, every single MacBook you can continue to buy

01:29:51 ◼   ► is either an older model,

01:29:53 ◼   ► or if it's a top of the line new model,

01:29:57 ◼   ► it still has this exact same keyboard.

01:29:59 ◼   ► Mm-hmm, which is I think the first time I can ever recall the few times that Apple is admitted to

01:30:05 ◼   ► Hardware problems with a device that it that they're still they don't have a replacement ready yet, which is kind of awkward

01:30:12 ◼   ► Yeah, it's super awkward I mean if they're you know, they shouldn't have been flat-footed on that front simply because they

01:30:21 ◼   ► They've known about the issues for a long time, you know

01:30:27 ◼   ► Even though Apple maintains that it's a very small amount of devices, it's enough.

01:30:33 ◼   ► It's enough for sure.

01:30:35 ◼   ► So one of two things going on, either they're getting ready to announce new MacBooks, which

01:30:41 ◼   ► could totally happen, and it'll happen really quickly here, and then you will have options,

01:30:46 ◼   ► and they'll just quietly deprecate all those current ones, or I don't know what.

01:30:52 ◼   ► Or they're going to be stuck with these for a while.

01:30:55 ◼   ► So either or on that front.

01:30:59 ◼   ► This is completely and utterly unsubstantiated, but I'll pass it along.

01:31:04 ◼   ► That's what podcasts are for.

01:31:05 ◼   ► This is the sort of thing I would never write on during Fireball, but I'll say on the show

01:31:07 ◼   ► is I got an email from somebody after this was announced who I don't know and who admitted

01:31:11 ◼   ► that he doesn't know this firsthand.

01:31:16 ◼   ► He knows a friend who knows a guy at Apple.

01:31:21 ◼   ► My uncle works at Nintendo.

01:31:25 ◼   ► what he said. And again, he even said, I don't know, you can't buy, you know, he knows me well,

01:31:30 ◼   ► you know, he's a reader of the site. But he said, what, what, what he heard from somebody was that

01:31:34 ◼   ► Apple got serious, you know, at some point, you know, last, you know, at some point in the last

01:31:40 ◼   ► year or so, got serious about thoroughly investigating this, like, okay, something is

01:31:44 ◼   ► up, what is going wrong, and that they looked into it, and that there's a certain part of the

01:31:50 ◼   ► butterfly mechanism that's made of a metal alloy and that this metal alloy is,

01:31:58 ◼   ► as they were producing them, was not up to their specific up to the proper specifications for

01:32:05 ◼   ► resistance to bending. And that was prone to getting slightly bent, just slightly,

01:32:10 ◼   ► because there's obviously no room with the low key travel for a large band. But that once bent,

01:32:16 ◼   ► even slightly made that particular key more likely to get stuck if like a piece of dust or something

01:32:25 ◼   ► went in there. That it was in that they looked into it, rectified it, and have since switched

01:32:32 ◼   ► to a different metal alloy that may or may not even look different in terms of what it would

01:32:38 ◼   ► look like if you pop the key off, but that it's much more rigid and less prone to bending. And

01:32:42 ◼   ► and that the more recent your MacBook or MacBook Pro is,

01:32:47 ◼   ► the more likely it is that it has this different alloy

01:32:50 ◼   ► in the butterfly switches

01:32:51 ◼   ► and will be less likely to suffer this.

01:32:55 ◼   ► I can't prove that.

01:32:58 ◼   ► There's no way Apple is ever going to say

01:33:00 ◼   ► whether that's true or not.

01:33:01 ◼   ► I think the only way we could ever find out

01:33:05 ◼   ► or even suspect whether something like that is true or not

01:33:09 ◼   ► is whether we start to see a decrease

01:33:10 ◼   ► in the number of people with,

01:33:12 ◼   ► know, if anecdotally, we observe that the more recent your keyboard is, the less likely it is to

01:33:17 ◼   ► to break down. Yeah. I mean, look, we started seeing reports of it, I think, like, you know,

01:33:24 ◼   ► very soon after they were announced, or very soon after they were launched, like a couple of people.

01:33:29 ◼   ► And then it just started started cresting, like early last year. And then, you know, there have

01:33:35 ◼   ► been plenty of like, you know, talk about it, and, you know, medium posts from independent parties

01:33:40 ◼   ► about it. And, you know, certainly folks like Marco were talking about it. Other people

01:33:45 ◼   ► were talking about it.

01:33:46 ◼   ► And Casey Johnston. Casey Johnston.

01:33:48 ◼   ► Casey wrote an article in October of last year, whenever it was, and that sort of put

01:33:52 ◼   ► a pin in the topic of like, you know, hey, here's here's a on the record, a site writing

01:33:58 ◼   ► about it, you know, and then she kept hammering at it. And so I think there was a long time

01:34:03 ◼   ► before she codified it. There was like buzz, but she really crystallized it well and then

01:34:07 ◼   ► kept hammering on it.

01:34:08 ◼   ► So it got embarrassing.

01:34:10 ◼   ► I think it gets embarrassing at some point for people who are ostensibly known for building

01:34:16 ◼   ► some of the best hardware in the world.

01:34:18 ◼   ► I love something that's just so terrible.

01:34:21 ◼   ► Yeah, I really did laugh out loud at Casey Johnston's take after this repair program

01:34:30 ◼   ► was announced.

01:34:35 ◼   ► She wrote, "Apple did not immediately return a request from this reporter for comments

01:34:41 ◼   ► on whether repairs may now be done on site at stores to shorten the time computers must

01:34:46 ◼   ► be without their computers, whether the keyboard design is changed such that a repair may eliminate

01:34:51 ◼   ► the problem rather than prop up a faulty design, or whether Apple anticipates releasing updated

01:34:58 ◼   ► hardware that is not so prone to failure at any point in the future.

01:35:03 ◼   ► their keyboards too are broken. I'm a big KZ fan. I have been for a long time.

01:35:11 ◼   ► Just got a sharp… That was just an amazing sentence. Perhaps

01:35:16 ◼   ► their keyboards too are broken. I don't know. It's a weird thing. And I'll tell

01:35:21 ◼   ► you, it's one of those things as the guy who writes Daring Fireball and does the show.

01:35:25 ◼   ► I get… for years and many, 10 years more, I get email from people, readers, and they'll

01:35:31 ◼   ► I'll just say something like,

01:35:33 ◼   ► hey, my kid's going to college in August

01:35:37 ◼   ► and I was gonna get her a new MacBook,

01:35:39 ◼   ► but what should I do?

01:35:40 ◼   ► Should I wait?

01:35:41 ◼   ► I'm worried about this keyboard thing.

01:35:44 ◼   ► We're hoping to get her a keyboard,

01:35:45 ◼   ► we're hoping to get her a MacBook

01:35:46 ◼   ► that'll take her through college for four years.

01:35:48 ◼   ► And I often don't know what to say.

01:35:53 ◼   ► I don't know how to make recommendations for people.

01:35:56 ◼   ► If people say like, hey, should we get a MacBook Pro

01:35:59 ◼   ► or get the regular MacBook, you know,

01:36:01 ◼   ► I don't know how to answer that sometimes.

01:36:05 ◼   ► But in this particular case, I have no idea what to say,

01:36:07 ◼   ► you know, because you might need a computer,

01:36:09 ◼   ► but I have no idea, like, you know,

01:36:12 ◼   ► the timing on new MacBooks coming out

01:36:14 ◼   ► is actually pretty unfortunate,

01:36:15 ◼   ► 'cause June would have been in time for back to school

01:36:18 ◼   ► if they had had them to announce at WWDC,

01:36:20 ◼   ► and September is obviously too late,

01:36:23 ◼   ► 'cause the kids all start school in late August.

01:36:25 ◼   ► So I don't know what to say to people.

01:36:27 ◼   ► It's a tough time.

01:36:29 ◼   ► I wonder, as this gains publicity,

01:36:33 ◼   ► whether we'll start seeing it reflected

01:36:34 ◼   ► in the sales numbers of Macs.

01:36:38 ◼   ► How many people out there are holding off on a new MacBook?

01:36:41 ◼   ► - Honestly, probably not much.

01:36:42 ◼   ► - I don't think so.

01:36:43 ◼   ► It doesn't seem like it so far.

01:36:44 ◼   ► - It's a tempest in a teapot almost always with this stuff.

01:36:47 ◼   ► Like, you look at the video card stuff and like that.

01:36:50 ◼   ► There's a variety of debacles

01:36:52 ◼   ► that Macs have had over the years,

01:36:53 ◼   ► and I just don't think that most people really give a crap,

01:36:56 ◼   ► Especially not the people that are buying at scale.

01:36:59 ◼   ► - And as much as we, meaning like me and you

01:37:01 ◼   ► and everybody who's listening to this show,

01:37:03 ◼   ► we think that this is like a major issue and controversy.

01:37:05 ◼   ► It is in our little tech meme universe.

01:37:10 ◼   ► - And I'm not saying it's not valid, 100% valid, right?

01:37:12 ◼   ► But that's not the point at all.

01:37:13 ◼   ► It's just a matter of like,

01:37:14 ◼   ► there's an external assessment you can do

01:37:16 ◼   ► on whether or not it's gonna have a material impact.

01:37:18 ◼   ► I don't think so, to be honest.

01:37:20 ◼   ► But at the same time, I think it's like,

01:37:22 ◼   ► so when I said there's just two options,

01:37:25 ◼   ► there's really three options.

01:37:26 ◼   ► The 2A and 2B is they introduced new MacBook hardware now

01:37:31 ◼   ► because they were ready to replace this keyboard

01:37:36 ◼   ► or have known about it for a long enough time

01:37:38 ◼   ► to have gotten that done, gotten whatever major fix done

01:37:42 ◼   ► that needs to be done so you don't have to replace

01:37:43 ◼   ► the entire top of your MacBook to fix a key.

01:37:46 ◼   ► Or they weren't ready and so whatever speed bump

01:37:52 ◼   ► that they were going to introduce is delayed.

01:37:55 ◼   ► - Yeah. - Right?

01:37:56 ◼   ► And so if that is, you might see some material impact

01:37:59 ◼   ► because of the school buying season.

01:38:01 ◼   ► - Right. - Right?

01:38:02 ◼   ► And the buying season for people going back,

01:38:04 ◼   ► kids go back to school,

01:38:06 ◼   ► the school year's starting back up again, things like that.

01:38:08 ◼   ► But most of that stuff happens so early in the year,

01:38:11 ◼   ► I just don't, I don't know.

01:38:13 ◼   ► You know, I don't know if we'll see anything there.

01:38:16 ◼   ► - All right, and my last point on this,

01:38:18 ◼   ► and Marco Arment made it very, very succinctly

01:38:21 ◼   ► on the most recent episode of ATP.

01:38:24 ◼   ► But I don't like this four-year limit

01:38:28 ◼   ► on the replacement thing.

01:38:30 ◼   ► And I'm literally talking to you

01:38:32 ◼   ► on a Retina 13-inch MacBook Pro mid-2014

01:38:37 ◼   ► that I think I bought, I don't know,

01:38:40 ◼   ► like September, October of 2014.

01:38:42 ◼   ► So it's nearly four years old.

01:38:44 ◼   ► And it is fantastic.

01:38:47 ◼   ► It's actually my favorite Apple laptop I've ever owned.

01:38:51 ◼   ► and not just like by grading on a curve

01:38:55 ◼   ► by the technologies available in the day.

01:38:59 ◼   ► I'm not comparing a 20-year-old PowerBook screen

01:39:02 ◼   ► to this Retina screen.

01:39:04 ◼   ► I'm just saying in terms of how much I liked the book

01:39:06 ◼   ► compared to a notebook compared to what else

01:39:11 ◼   ► was on the market or what I expected to get out of it.

01:39:13 ◼   ► It's the greatest MacBook I've ever owned.

01:39:16 ◼   ► I love it.

01:39:17 ◼   ► And it is just rock solid.

01:39:19 ◼   ► I have no intention to replace it soon.

01:39:21 ◼   ► and it's four years old.

01:39:22 ◼   ► I don't know that, I think it's very,

01:39:25 ◼   ► especially for the prices that Apple charges,

01:39:27 ◼   ► I think it's very reasonable for everybody

01:39:30 ◼   ► who buys any MacBook that they sell,

01:39:33 ◼   ► right down to the 999 MacBook Air,

01:39:36 ◼   ► to expect to get at least four years out of it.

01:39:39 ◼   ► So I don't know about that.

01:39:42 ◼   ► It just seems like a weirdly arbitrary number

01:39:44 ◼   ► for Apple to have picked.

01:39:45 ◼   ► And it's a number that I think is lower

01:39:49 ◼   ► than the expected lifespan of a MacBook.

01:39:53 ◼   ► Again, if they were selling $200 Chromebooks,

01:39:58 ◼   ► a four-year replacement thing on the keyboard would be fine

01:40:02 ◼   ► because you're lucky if you can get one of those things

01:40:06 ◼   ► to last four years, whereas a $2,000 MacBook Pro,

01:40:09 ◼   ► boy, that seems to me like it ought to be able

01:40:11 ◼   ► to last four years and still be able to type a space bar.

01:40:14 ◼   ► (laughs)

01:40:16 ◼   ► - Right.

01:40:18 ◼   ► So I don't know.

01:40:19 ◼   ► - Yeah, yeah, I don't know.

01:40:20 ◼   ► It really is a depressing outlook

01:40:23 ◼   ► when you look at how long you've been able

01:40:26 ◼   ► to use hardware in the past.

01:40:27 ◼   ► I mean, I used MacBooks for many, many years at a chunk.

01:40:31 ◼   ► And honestly, given that the stuff that I was doing

01:40:36 ◼   ► went more and more towards the web over the years

01:40:38 ◼   ► and continues to do so,

01:40:39 ◼   ► it's just hard to justify buying into a MacBook

01:40:46 ◼   ► that you don't think can last that long.

01:40:47 ◼   ► because I've had some that have been decimated by like liquid damage or other

01:40:51 ◼   ► things that it could have easily still been using, you know, um,

01:40:54 ◼   ► with no problem at all. It's just, you really just open them up, you use them,

01:40:58 ◼   ► they're reliable, everything works great.

01:41:00 ◼   ► And then to have that reputation undercut in this way,

01:41:04 ◼   ► I think it's more about long-term damage than it is about short term.

01:41:07 ◼   ► And in terms of the most vociferous users who then become activist users,

01:41:12 ◼   ► you know? Right. Right. Cause part of the argument is, Oh yeah,

01:41:15 ◼   ► Apple laptops do cost a lot more than the average competition, but they're worth it,

01:41:20 ◼   ► and their build quality is higher, and they last longer. This undercuts all of that. Well,

01:41:26 ◼   ► we'll see if they extend it. I wouldn't be surprised if they wind up, now that they've

01:41:32 ◼   ► broken the seal and done the hardest part, which is just start the program at all, I

01:41:36 ◼   ► wouldn't be surprised if they extend it in some way, as they say quietly. Apple quietly

01:41:42 ◼   ► extends my keyboard repair program. Matthew, we've gone way over the amount of time I told

01:41:48 ◼   ► you I would take and I apologize for that, but I appreciate it so much and I cannot say

01:41:53 ◼   ► how much I enjoyed your exclusive story here behind the scenes with maps. You just knocked

01:42:00 ◼   ► it out of the park again and I just couldn't have done a better job.

01:42:06 ◼   ► Thank you very much, sir. I appreciate it. It means a lot.

01:42:09 ◼   ► Keep them busy in the summer.

01:42:11 ◼   ► busy. Yes, I'm looking forward to relaxing for until the end of the quarter. Yeah. So

01:42:18 ◼   ► everybody can read your fine tweets at Panzer panzer on the on the Twitter and of course,

01:42:26 ◼   ► see your read your fine work and the work of your staff at tech crunch calm