WEBVTT - Benedict Evans Talks About the Tech Bubble 0:00:02.200 --> 0:00:06.800 This is Masters in Business with Barry Ridholts on Bloomberg Radio. 0:00:09.280 --> 0:00:11.920 This week on the podcast, I have an extra special guest. 0:00:12.160 --> 0:00:14.720 His name is ben Nick Evans, and he is a 0:00:14.760 --> 0:00:19.400 partner at the famed technology venture capital firm in Silicon 0:00:19.560 --> 0:00:25.439 Valley Andrews and Horowitz, perhaps known to the aficionado as 0:00:25.520 --> 0:00:29.720 a sixteen z Um. I have actually some swag from 0:00:30.000 --> 0:00:34.720 Andreas and Harrowitz. I've blue a sixteen z hat which 0:00:34.800 --> 0:00:38.360 matches my car and I just leave that in the car, 0:00:38.520 --> 0:00:41.360 Um when I'm driving around trying to look cool to 0:00:41.440 --> 0:00:45.640 a very select group of tech geeks. Um, if you 0:00:45.720 --> 0:00:50.320 are at all interested in the development of technology, of ecosystems, 0:00:50.400 --> 0:00:57.200 of of autonomous everything, of smartness, of really where the 0:00:57.280 --> 0:01:01.840 puck is going to be the future of technology, you 0:01:01.880 --> 0:01:05.560 will find this to be a fascinating conversation. He absolutely 0:01:05.640 --> 0:01:09.880 has an encyclopedic knowledge of what's taken place and why 0:01:10.280 --> 0:01:15.680 and has tremendous insight into the likely direction that uh, 0:01:15.720 --> 0:01:18.800 this space is going. So, with no further ado, my 0:01:18.959 --> 0:01:26.920 conversation with Andres and Horowitz is Benedict Evans. I'm Barry 0:01:26.959 --> 0:01:30.440 rid Hilts. You're listening to Master's in Business on Bloomberg Radio. 0:01:30.920 --> 0:01:34.160 My special guest today is Benedict Evans. He is a 0:01:34.200 --> 0:01:38.080 partner at the legendary venture capital firm Andrees and Horowitz, 0:01:38.120 --> 0:01:41.760 where he works as an analyst and does a weekly 0:01:42.120 --> 0:01:48.080 consumer newsletter blog What have You, covering everything from mobile 0:01:48.120 --> 0:01:52.880 platforms to AI to autonomous everything. He is also a 0:01:52.960 --> 0:01:57.360 recommended follow. He began his career as an equity analyst 0:01:57.760 --> 0:02:01.440 in investment banking, moving on to rategy and business development 0:02:01.800 --> 0:02:07.240 at such August firms as Orange and NBC Universal. Benlict Evans, 0:02:07.320 --> 0:02:09.560 Welcome to Bloomberg. Thank you for having me. Do do 0:02:09.600 --> 0:02:12.280 you prefer Benlicht or Ben or Benedictio? Is better? S? A? 0:02:12.680 --> 0:02:16.800 Okay for sure? Um, there is an advantage to having 0:02:16.800 --> 0:02:21.160 an unusual first or last name for SEO purposes. I'm 0:02:21.200 --> 0:02:23.799 convinced if my name was Barry Smith, no one would 0:02:23.880 --> 0:02:26.080 know who I am. But we can talk about s 0:02:26.080 --> 0:02:29.680 c O a little later. So you graduate Cambridge in 0:02:31.680 --> 0:02:36.040 right in the midst of the dot com and technology boom, 0:02:36.520 --> 0:02:39.959 not too long before the bust. What was it like 0:02:40.120 --> 0:02:44.040 coming out into the banking world in the midst of 0:02:44.440 --> 0:02:47.840 that era. Well, so I actually joined an equity capital 0:02:47.840 --> 0:02:50.440 markets team. So we were doing europ and tech I 0:02:50.480 --> 0:02:53.040 P A S and Safe for about nine months. I thought, 0:02:53.200 --> 0:02:55.160 this is going to be a fun career, right, this 0:02:55.240 --> 0:02:57.440 is easy. You just throw it out there and everybody 0:02:57.480 --> 0:03:01.119 buys it. He's some random thing, and a sovereign wealth 0:03:01.160 --> 0:03:04.400 fund puts in an order for three extra floats. And 0:03:04.440 --> 0:03:08.560 then in about well early spring of two thousand, everything 0:03:08.600 --> 0:03:11.840 started going down and we sort of forget now that 0:03:12.000 --> 0:03:13.840 we talked about the dot com bubble. But you know, 0:03:13.880 --> 0:03:16.359 particularly in Europe, there was also a mobile bubble. But 0:03:16.400 --> 0:03:18.600 there was also a mobile bubble, and there was also 0:03:18.680 --> 0:03:21.040 the fixed line broadband fiber bubble. So you kind of 0:03:21.080 --> 0:03:23.400 had three bubbles that kind of had happened more or 0:03:23.480 --> 0:03:26.160 less independently for different reasons, and kind of merged and 0:03:26.240 --> 0:03:28.920 combined into one enormous thing. I would then when then 0:03:28.960 --> 0:03:31.240 the market went down for three years, I would argue 0:03:31.240 --> 0:03:33.400 you had a semi conductor bubble, you had a software bubble, 0:03:33.400 --> 0:03:35.600 you had a hardware bubble, you had a storage bubble. 0:03:35.640 --> 0:03:38.840 It all started a band with bubble. In the US, 0:03:38.880 --> 0:03:41.160 we had fiber uptick and telecom bubble, where you had 0:03:41.600 --> 0:03:43.560 you had a tech bubble, but you also had a 0:03:43.560 --> 0:03:46.960 mobile bubble. Which is very only for very distant reasons, 0:03:47.000 --> 0:03:48.400 was happening at the same time. It was happening at 0:03:48.400 --> 0:03:50.000 the same time, and then there was a fiber bubble, 0:03:50.040 --> 0:03:51.520 and yet so they all combined and we got this 0:03:51.640 --> 0:03:54.040 enormous kind of explosion and then an enormous pot and 0:03:54.080 --> 0:03:56.120 the market went down without stopping more or less for 0:03:56.160 --> 0:03:58.840 three years. So were you always the tech guy or 0:03:59.000 --> 0:04:01.240 or did you was your background more finance or something 0:04:01.360 --> 0:04:04.240 well above? So I did my degree in history, which 0:04:04.560 --> 0:04:08.080 is really an analytic subject. So here is a bunch 0:04:08.120 --> 0:04:11.040 of information, try and work out something interesting or meaningful 0:04:11.120 --> 0:04:13.920 or insightful to say about it, or something original and 0:04:13.960 --> 0:04:16.360 interesting to say about it. And so I went from 0:04:16.400 --> 0:04:19.159 that into sort of fairly conventional routing into invest in 0:04:19.160 --> 0:04:21.960 banking UM, and sort of quickly decided that e c 0:04:22.160 --> 0:04:24.400 M was kind of project management and whereas research was 0:04:24.440 --> 0:04:26.560 probably something I was likely to be better at. And 0:04:26.600 --> 0:04:28.360 so I went into at research and I did did 0:04:28.640 --> 0:04:30.960 did UM. So I went and did my European mybile 0:04:30.960 --> 0:04:33.040 stocks for a couple of years, and that was in 0:04:33.080 --> 0:04:35.680 London or New York and London. And did you did 0:04:35.680 --> 0:04:37.320 you spend any time in New York or did you 0:04:37.360 --> 0:04:40.720 go straight London to well, I was on Valley. Uh No, So, 0:04:40.839 --> 0:04:42.800 I mean I came to New York for work and 0:04:42.800 --> 0:04:45.200 I worked for NBC Universal, so I used to come 0:04:45.200 --> 0:04:47.159 over here. Um, but now the first time I actually 0:04:47.200 --> 0:04:48.920 worked properly in the States was to go to work 0:04:48.960 --> 0:04:53.160 for a sixty and oh so London with really just 0:04:53.360 --> 0:04:55.560 brief sojourns in New York City. Yeah, I mean I 0:04:55.600 --> 0:04:57.040 worked for Orange, So I went to Paris and I 0:04:57.160 --> 0:04:59.400 worked for NBC Universal. So I came to New York, 0:04:59.440 --> 0:05:02.920 but that my job job spot was in London. And 0:05:02.960 --> 0:05:05.960 so you're covering telecom starts, recovering Mobile, You're covering a 0:05:06.000 --> 0:05:10.440 lot of the technology related space as an analyst. What 0:05:10.480 --> 0:05:14.000 was that like in the midst of wild overvaluation and 0:05:14.040 --> 0:05:17.360 collapsing share price? So it's it's kind of interesting. I mean, 0:05:17.360 --> 0:05:19.240 there's there's there's a couple of different things that one 0:05:19.279 --> 0:05:22.400 of them is it's interesting in hindsight to to look 0:05:22.400 --> 0:05:24.560 at them in hindsight in that they were the go 0:05:24.680 --> 0:05:28.000 go growth, exciting, dynamics, sexy, disruptive companies. Me and Mobile 0:05:28.040 --> 0:05:31.520 had gone from nothing in oh my god, everyone on 0:05:31.520 --> 0:05:33.240 Earth is going to have one of these things, and 0:05:33.279 --> 0:05:36.360 they were really kind of like they thought of themselves 0:05:36.360 --> 0:05:41.080 as poets and they connected everybody and that was it. 0:05:41.480 --> 0:05:43.160 And you know, there's obviously there's still a lot of 0:05:43.200 --> 0:05:45.360 stuff going on in emerging markets, but like everyone Knework 0:05:45.360 --> 0:05:46.840 got a phone and that was it. And then they 0:05:46.839 --> 0:05:49.760 went from being amazing. They went from being Google to 0:05:49.800 --> 0:05:54.080 being water companies in like a year two years effect. 0:05:54.360 --> 0:05:56.240 But so you look at the all the stuff we 0:05:56.240 --> 0:05:57.880 do on our smartphones, now, this was all kind of 0:05:57.880 --> 0:06:00.440 in concept videos and presentations from the mob All operated 0:06:00.480 --> 0:06:03.600 is in you're going to do all of this stuff, 0:06:03.600 --> 0:06:05.080 and they were going to do it all, and of 0:06:05.120 --> 0:06:07.360 course it all happened, but it wasn't done by them, 0:06:07.760 --> 0:06:12.440 And so that perspective of how kind of I think 0:06:12.440 --> 0:06:14.000 there's a lot of echoes of that. For example, if 0:06:14.000 --> 0:06:16.160 you look at cars now, you have this all this 0:06:16.279 --> 0:06:18.480 future ology of what's going to happen and what cars 0:06:18.520 --> 0:06:20.560 will be and how everything will change, and you have 0:06:20.640 --> 0:06:22.560 kind of the renderings of the glass cars, which is 0:06:22.600 --> 0:06:24.640 like the renderings of the glass phones in two thousands 0:06:24.640 --> 0:06:27.520 when there were no phones with color screens, and you think, well, 0:06:27.520 --> 0:06:30.479 you might get like three quarters of this, right, But 0:06:30.640 --> 0:06:32.520 like the last quarter of it is going to be 0:06:32.560 --> 0:06:34.760 where all the money is. And that's the difference between 0:06:34.760 --> 0:06:37.400 it being knock you and Microsoft in it being companies 0:06:37.440 --> 0:06:40.520 you've never heard of like Apple and Samsung and Google 0:06:40.600 --> 0:06:42.880 or forgotten like Apple um and no one had ever 0:06:42.920 --> 0:06:45.160 heard of Google. And so it's interesting to look back 0:06:45.160 --> 0:06:46.839 and think, how did we think about what was going 0:06:46.880 --> 0:06:49.080 to happen then, and how much of it was right 0:06:49.120 --> 0:06:50.800 and wrong? And what would you have had to have 0:06:50.839 --> 0:06:53.000 said then in order to get it right. So you're 0:06:53.080 --> 0:06:58.880 describing a form of historical futurism because when we've what 0:06:58.960 --> 0:07:03.400 we've seen over the years is expectations of the future 0:07:03.920 --> 0:07:08.960 turn out to be wildly either over optimistic or pessimistic, 0:07:09.320 --> 0:07:11.440 but rarely right on the nose. Is that a fair 0:07:11.960 --> 0:07:13.640 And I think that's right. I mean, as I said, 0:07:13.680 --> 0:07:15.880 I think there's sort of several bits. There's sort of 0:07:15.880 --> 0:07:19.200 several lessons you can learn from, like not so much 0:07:19.240 --> 0:07:20.640 to drop the bubble per se, but the kind of 0:07:20.640 --> 0:07:22.640 the way people thought about it. And so there was 0:07:22.680 --> 0:07:24.400 this whole idea in the early nineties of the thing, 0:07:24.440 --> 0:07:28.080 this thing called the Information super Highway, which the whole 0:07:28.200 --> 0:07:30.120 just the name conveys the fact that it was going 0:07:30.160 --> 0:07:31.920 to be kind of centrally controlled, and it would be 0:07:31.960 --> 0:07:35.080 the cable company and News Corporation and twenty Century Fox 0:07:35.080 --> 0:07:37.200 and the New York Times, and they would kind of 0:07:37.240 --> 0:07:39.200 get together every six months and decide what you were 0:07:39.200 --> 0:07:42.160 going to have and how. And of course what we 0:07:42.200 --> 0:07:44.600 had instead was permission in this innovation and the open Internet, 0:07:44.640 --> 0:07:46.320 and there was no central authority, and anyone could do 0:07:46.360 --> 0:07:48.840 what they wanted in just to some extent. And so 0:07:49.000 --> 0:07:50.880 that is interesting to compare that now with the way 0:07:50.880 --> 0:07:52.720 we for example, people talk about cars all the way 0:07:52.760 --> 0:07:56.120 people talk about what's happening to TV now. Um, I 0:07:56.160 --> 0:07:59.640 think the kind of the but as I just said, 0:07:59.680 --> 0:08:01.440 the fact that you could have got you could have 0:08:01.480 --> 0:08:04.440 guessed like eighty or nine of it and it's still 0:08:04.480 --> 0:08:07.120 have missed the important parts is also fascinating. Like you 0:08:07.120 --> 0:08:09.960 could have said in nineteen in two thousand, okay, everyone 0:08:09.960 --> 0:08:11.840 will have one of these things. Everyone will have Internet 0:08:11.880 --> 0:08:14.080 on the phone. It will be a real operating system 0:08:14.120 --> 0:08:17.680 and not a future phone. It will be open Internet. Um, 0:08:17.760 --> 0:08:20.640 therefore it will not be the carriers that would have 0:08:20.760 --> 0:08:23.120 can describe today's well perfectly, But you would have still 0:08:23.120 --> 0:08:24.960 have said, well then it'll be Microsoft K Nokia that 0:08:25.000 --> 0:08:28.080 do that. What's amazing is there was this a T 0:08:28.240 --> 0:08:31.720 and T commercial. I very vividly remember. I think Tom 0:08:31.760 --> 0:08:35.080 Selleck did the voice over and he talks about what 0:08:35.200 --> 0:08:38.400 the future will be like, and I think the punchline 0:08:39.000 --> 0:08:41.600 on the beach you will. And it turned out to 0:08:41.640 --> 0:08:43.920 be dead right, except for the fact that a T 0:08:44.040 --> 0:08:46.319 has or more nothing to do with it exactly. I mean. 0:08:46.320 --> 0:08:47.920 One of the ways I described this when I talked 0:08:47.920 --> 0:08:51.000 to Talcos now is it's as though a municipal water 0:08:51.080 --> 0:08:53.680 company looked at the mineral water business and they said, 0:08:53.800 --> 0:08:56.160 you know, come on, we've got brand, we've got water, 0:08:56.280 --> 0:08:58.199 we've got trusted. You wouldn't buy water from a company 0:08:58.240 --> 0:09:01.480 you don't trust. We should be doing And they hire 0:09:01.559 --> 0:09:04.439 McKinsey and wolf Fallens and they build the whole thing, 0:09:04.480 --> 0:09:06.440 and like two years later, that first pallet of water 0:09:06.480 --> 0:09:08.319 balls into Walmart and it goes onto the shelves and 0:09:08.320 --> 0:09:09.719 they look at it next to yether of two D 0:09:09.880 --> 0:09:12.360 brands and they think, hang on a second, something not 0:09:12.480 --> 0:09:17.480 quite right here. Um, So, how did you go from 0:09:17.520 --> 0:09:20.520 working in London or Paris as a telecom man? Was 0:09:20.600 --> 0:09:24.160 too I know, I'm going to move six thousand miles 0:09:24.160 --> 0:09:27.400 away and become a imagineer. And I just asked them 0:09:27.440 --> 0:09:28.840 for a job. I don't know why people say it's 0:09:28.880 --> 0:09:30.280 hard to get a job in venture. You know, you 0:09:30.280 --> 0:09:32.360 go ask them for a job, they say, yes, okay, 0:09:32.400 --> 0:09:34.800 that's it. Yeah, it was easy. How did you how 0:09:34.800 --> 0:09:37.960 did you? Well? First hear of instance, so well, so 0:09:38.000 --> 0:09:40.520 there's a process here. And as I mentioned, I actually 0:09:40.520 --> 0:09:42.720 did my degree in history, which is analysis, and so 0:09:42.880 --> 0:09:44.440 the question was sort of picking up and looking for 0:09:44.480 --> 0:09:46.240 things where you can apply that to analysis, and so 0:09:46.440 --> 0:09:49.319 I left university and became a south Side analyst, and 0:09:49.600 --> 0:09:52.520 is sort of many people will know that, for many 0:09:52.600 --> 0:09:54.560 various reasons, stop being a fun thing to do, or 0:09:54.559 --> 0:09:56.200 even a thing that anyone could do. So I think 0:09:56.200 --> 0:09:58.520 when I was in telecoms, merl Lynch, you have had 0:09:58.559 --> 0:10:01.800 something like fifty um up in telecoms analysts, and the 0:10:01.880 --> 0:10:03.320 last time I looked, I think they had less than 0:10:03.360 --> 0:10:06.680 half a dozen. So that industry ceased to be fun. 0:10:06.960 --> 0:10:10.040 And so I left and went and worked in strategy, 0:10:10.240 --> 0:10:13.480 first at Orange and then at um NBC Universal and 0:10:13.559 --> 0:10:16.800 Channel four again both in London. So from sort of 0:10:16.800 --> 0:10:18.920 sitting on the outside in sitting on the inside out 0:10:18.920 --> 0:10:20.200 trying to work out what we do and how do 0:10:20.200 --> 0:10:22.280 we how do we understand this? And it's kind of 0:10:22.280 --> 0:10:24.160 interesting because the kind of the questions changed and suddenly 0:10:24.200 --> 0:10:25.959 you've got a spreadsheet that lists all the stuff you've 0:10:25.960 --> 0:10:28.280 been trying to work out from the from the public financials. 0:10:28.480 --> 0:10:30.439 But then you're just trying to work out different stuff. 0:10:30.480 --> 0:10:32.199 But the process is the thought process at the state. 0:10:32.559 --> 0:10:34.559 Because I did that for a while. Um. I left 0:10:34.679 --> 0:10:37.560 NBC Universal the week that General Electric share price fell fifty, 0:10:37.960 --> 0:10:41.440 which is obviously a called direct causal relationship. Sometimes you know, 0:10:41.559 --> 0:10:44.120 things happened and you could just draw a straight line 0:10:44.160 --> 0:10:46.600 from A to B exactly. I recall seeing the Wall 0:10:46.600 --> 0:10:50.080 Street General Peace, Ben lick Evans departs NBC Universal gu 0:10:50.200 --> 0:10:53.240 started out in their exactly we went up up by half, 0:10:53.240 --> 0:10:55.040 and that the same amount the next week. Of course, 0:10:55.640 --> 0:10:58.960 um so I left when the financial that is in 0:10:58.960 --> 0:11:01.679 the financial crisis. It was so yeah, sound about that, 0:11:02.280 --> 0:11:04.760 an old and boring story of what happened there. Um 0:11:04.800 --> 0:11:07.000 and then I went as I worked as a consultant 0:11:07.040 --> 0:11:09.240 in London, so I was doing a lot of sort 0:11:09.240 --> 0:11:13.840 of strategy consultancy and sort of producing research reports forum 0:11:14.160 --> 0:11:16.560 around the European media and telecom space, and so I 0:11:16.640 --> 0:11:19.440 was writing both about kind of what's happening in fashion 0:11:19.480 --> 0:11:22.880 magazines and you know record retail and DVD retail. Would 0:11:22.880 --> 0:11:26.400 also water Google and Facebook doing what's happening with smartphones, 0:11:26.960 --> 0:11:29.080 And so I started a blog, and more or less 0:11:29.080 --> 0:11:30.719 the same time, I went on Twitter at that time, 0:11:30.760 --> 0:11:32.160 which he was kind of new, and so you could 0:11:32.160 --> 0:11:34.120 were quite easy to get noticed if you're doing stuff, 0:11:34.520 --> 0:11:36.360 and I was kind of writing stuff. The way I 0:11:36.360 --> 0:11:39.040 describe it is basically I would write stuff that a 0:11:39.040 --> 0:11:41.720 cell side analyst or a senior strategy person or somewhat 0:11:41.720 --> 0:11:45.560 at Apple could write, except that either they couldn't public it, 0:11:45.600 --> 0:11:48.440 publish it, or they would be writing for a very 0:11:48.480 --> 0:11:50.840 different audience. They'd be writing for, you know, buy cell 0:11:50.880 --> 0:11:55.120 whole audience or um or they wouldn't have the analytic 0:11:55.160 --> 0:11:56.959 background to write it. So like a sell side analyst 0:11:57.000 --> 0:11:59.000 could write this stuff, but they're writing for a different audience. 0:11:59.000 --> 0:12:01.400 A senior person at a company knows all of this stuff, 0:12:01.400 --> 0:12:03.000 but they are not used to writing stuff and they 0:12:03.000 --> 0:12:05.480 can't publish it anyway. Somebody and McKinsey knows all this 0:12:05.520 --> 0:12:06.960 stuff that they're not allowed to say that's not going 0:12:07.000 --> 0:12:10.079 to work. Um. So I was kind of an interesting, 0:12:10.200 --> 0:12:12.800 kind of kind of niche in the vendor kind of 0:12:12.800 --> 0:12:15.480 little segment on your vend diagram of somebody who had 0:12:15.480 --> 0:12:17.840 an analytic and strategy background was used to writing it 0:12:17.880 --> 0:12:19.720 that about this stuff and explaining it, could say it 0:12:19.720 --> 0:12:22.440 in public at a time when very few other people 0:12:22.440 --> 0:12:25.520 were doing it. And so for those reasons, UM, I 0:12:25.640 --> 0:12:27.440 sort of got noticed. And you know, I spent like 0:12:27.480 --> 0:12:29.640 two or three years writing a bog posts and getting 0:12:29.640 --> 0:12:31.640 like a hundred page vs a months, and then I 0:12:31.679 --> 0:12:33.560 went through a period where I was getting a couple 0:12:33.559 --> 0:12:35.840 of thousand page vis a day. Um. And that sort 0:12:35.840 --> 0:12:38.240 of happened in the course of two thousand thirteen. And 0:12:38.280 --> 0:12:40.320 then and I kind of picked up my my bag 0:12:40.360 --> 0:12:41.760 and thought, well, what else do I want to do? 0:12:41.800 --> 0:12:43.400 What do I want to do next? Where could I 0:12:43.440 --> 0:12:45.719 deploy this? And you know, you you have those conversations 0:12:45.720 --> 0:12:47.480 at various stages in your life. And I thought, well, 0:12:47.800 --> 0:12:50.040 I should go and do you do this in venture capital? 0:12:50.040 --> 0:12:51.640 And where's the place to go and do venture capital? 0:12:51.640 --> 0:12:55.120 And you the global cluster with San Francisco, and what 0:12:55.200 --> 0:12:58.160 was the kind of firm where it felt like this 0:12:58.240 --> 0:13:01.000 kind of innovative approach to its explaining things and adding 0:13:01.040 --> 0:13:03.679 value in public would work, and A sixteen Z was 0:13:03.720 --> 0:13:06.440 thought of at the top of that list, and you 0:13:06.480 --> 0:13:08.680 reached out to them as opposed to visus. So I 0:13:08.800 --> 0:13:11.440 got a couple of introductions. I went in and I said, well, 0:13:11.720 --> 0:13:13.440 this is sort of what I do. Is there a 0:13:13.440 --> 0:13:15.280 place for this in this firm? Does this fit within 0:13:15.320 --> 0:13:17.960 the firm? Is it's useful? And they kind of said yeah, 0:13:18.360 --> 0:13:23.000 just like that or less. So um, lots of other 0:13:23.040 --> 0:13:26.840 people blogers should be inundating a sixteen Z with their 0:13:26.960 --> 0:13:29.800 resumes and saying, look, I've been doing this also. You know, 0:13:29.840 --> 0:13:32.040 I don't think you need a resume. It's it's kind 0:13:32.040 --> 0:13:33.880 of like, you know, if you want a sales job, 0:13:33.920 --> 0:13:35.120 you should be able to get the meeting. If you 0:13:35.120 --> 0:13:36.719 can't get the meeting, you shouldn't have a sales job. 0:13:36.760 --> 0:13:38.120 And it's kind of the same. You prove that you 0:13:38.160 --> 0:13:40.200 can do the job by doing it. That makes that 0:13:40.240 --> 0:13:44.400 makes a lot of sense. So the focus that you 0:13:44.480 --> 0:13:48.840 have today is no longer telecom the way it was, 0:13:49.320 --> 0:13:53.760 but certainly the concept of mobility as a source of 0:13:54.480 --> 0:13:59.320 um possible changes in technology is a key factor. What 0:13:59.360 --> 0:14:01.360 do you focus on these things? Well, I think talking 0:14:01.400 --> 0:14:03.960 about mobile now is a bit like talking about PCs 0:14:04.040 --> 0:14:06.600 ten years ago. Right, Yes, this is the center of everything, 0:14:06.600 --> 0:14:09.920 but it's happened, and so we're not arguing about iOS 0:14:10.040 --> 0:14:12.040 versus Android, or whether everyone's going to have one of 0:14:12.080 --> 0:14:13.920 these things, or apps versus the web, all of those 0:14:14.000 --> 0:14:16.280 kind of things, though, they're not interesting conversations anymore. It's 0:14:16.320 --> 0:14:18.480 like arguing, you know, is everybody get onto going to 0:14:18.520 --> 0:14:20.720 get onto the web? Well? Yes, Now what next question? 0:14:21.520 --> 0:14:24.880 So you look? So there's one answer is well, what 0:14:24.920 --> 0:14:26.640 are the next questions? What are the next kind of 0:14:27.080 --> 0:14:30.280 mega trends that are happening. The second is there's a 0:14:30.280 --> 0:14:33.000 lot of conversation around what happens with that stuff? What 0:14:33.120 --> 0:14:35.560 happens with the stuff that we've already had so ten 0:14:35.600 --> 0:14:38.000 years ago? What do we do with broadband and browsers? 0:14:38.040 --> 0:14:40.520 So well, let's talk about search, let's talk about social, 0:14:40.560 --> 0:14:42.680 Let's talk about what happens once everyone has a PC. 0:14:43.120 --> 0:14:46.040 Now what happens once everyone has a smartphone and everyone 0:14:46.120 --> 0:14:48.440 is on Facebook? So we have those kinds of conversations 0:14:48.440 --> 0:14:50.480 about the world, where the world that we've just built, 0:14:50.480 --> 0:14:51.920 what happens with that, and then we think, well, what 0:14:51.960 --> 0:14:56.800 are the next things? So machine learning, autonomous cars, mixed reality, cryptocurrency, 0:14:57.160 --> 0:14:59.160 What are the next fundamental trends that will shape the 0:14:59.200 --> 0:15:01.400 tech industry the way first the PC and then mobile 0:15:01.440 --> 0:15:03.560 shape the tech industry in the last twenty or thirty years. 0:15:03.920 --> 0:15:06.760 I think then within that there's a kind of you know, 0:15:06.960 --> 0:15:09.520 for what I try and do as a question of 0:15:09.600 --> 0:15:12.120 looking for the arguments or looking for the questions. So 0:15:12.760 --> 0:15:15.600 what are the places where there's an on the one hand, 0:15:15.680 --> 0:15:17.320 On the other hand, is it going to look like 0:15:17.360 --> 0:15:18.840 this or is it going to look more like that? 0:15:18.880 --> 0:15:20.720 Are they going to be winner not? Will there be 0:15:20.760 --> 0:15:24.040 autonomous cars? Yes not? Will it be in fifteen or 0:15:24.040 --> 0:15:27.120 twenty years? Who knows, don't know. It's more like are 0:15:27.120 --> 0:15:28.840 they going to be winner? Takes all effects in this? 0:15:28.920 --> 0:15:30.360 Is it going to look like Android? Or is it 0:15:30.400 --> 0:15:32.440 going to look like abs where there's like a widget 0:15:32.440 --> 0:15:35.240 that everyone buys and it doesn't matter. Is what's going 0:15:35.280 --> 0:15:37.480 to happen in mixed reality? Is this going to be 0:15:37.520 --> 0:15:40.000 something that's going to be every hardware manufacturer? Is it 0:15:40.000 --> 0:15:42.280 going to be super concentrated? How do we think about 0:15:42.360 --> 0:15:44.720 machine learning? How do we try and understand what that 0:15:44.800 --> 0:15:47.960 might change inside big companies, and so kind of there's 0:15:48.000 --> 0:15:49.720 like what are the topics and then how do we 0:15:49.760 --> 0:15:52.200 work out what the kind of the useful questions to 0:15:52.240 --> 0:15:55.120 talk about might be within that? So you're less extra 0:15:55.240 --> 0:15:58.760 bolending current trends out into some future date and instead 0:15:58.960 --> 0:16:03.000 thinking about, well, here's what we are several steps beyond that. 0:16:03.080 --> 0:16:06.560 What are going to be the subsequent developments that this 0:16:06.640 --> 0:16:10.360 might lead to. You're really several steps ahead, that's right. 0:16:10.400 --> 0:16:12.280 I mean that's partly a consequence of the state of 0:16:12.280 --> 0:16:15.240 the industry. So like five years ago, okay, what's it 0:16:15.320 --> 0:16:17.560 was a horse race who's gonna win? But is Apple 0:16:17.600 --> 0:16:19.240 and Google? What's going to happen? Is Apple going to 0:16:19.320 --> 0:16:22.280 survive even though Android has got this open story and 0:16:22.280 --> 0:16:24.240 this open source story and so on? So is there 0:16:24.240 --> 0:16:25.640 going to be one for Apple? Is there going to 0:16:25.680 --> 0:16:27.040 be win for a third entry? And what will happen 0:16:27.040 --> 0:16:28.760 to BlackBerry? Will black Bey killing on? And in niche? 0:16:28.800 --> 0:16:30.240 Will Windows going to be able to break in? So 0:16:30.360 --> 0:16:33.560 all of that was it was the motorating I'm sorry Windows, Yeah, yah, yeah, 0:16:33.760 --> 0:16:36.760 I don't I don't recall that. Yeah, yeah, that's it's 0:16:36.760 --> 0:16:39.640 a painful subject for some people. So there's it was 0:16:39.640 --> 0:16:42.440 a horse race Steve, Steve Bomber in particular. Yeah, yeah, 0:16:42.440 --> 0:16:46.200 he's three chairs. But that was you are you're calling 0:16:46.200 --> 0:16:48.240 the place. That's that the American phrase, you're you're going 0:16:48.320 --> 0:16:50.400 to work out what was going on? Now you're not. 0:16:50.480 --> 0:16:52.640 That's we know what happened. And we don't have those 0:16:52.720 --> 0:16:55.200 kind of day by day tactical questions around autonomer's cars 0:16:55.200 --> 0:16:57.520 because like there aren't in your automo's cars. We don't 0:16:57.520 --> 0:16:59.960 have those kind of day by day questions around mixed 0:17:00.000 --> 0:17:02.400 reality because you can't buy a mixed reality headset yet, 0:17:03.240 --> 0:17:05.719 mixed reality glasses yet. So the character of the question 0:17:05.800 --> 0:17:08.600 changes because we're at a different point in the S curve. Basically, 0:17:08.600 --> 0:17:10.600 when the S curve is going near vertical, then you're 0:17:10.600 --> 0:17:12.200 trying to work out, oh my god, what's going on? 0:17:13.160 --> 0:17:14.560 Is the rocket ship going to blow up? When is 0:17:14.600 --> 0:17:17.280 it going to start flattening out. Where we are now 0:17:17.440 --> 0:17:19.520 is we've got the oldest curve has flattened out right 0:17:19.520 --> 0:17:21.080 at the top of the of the scale, and we're 0:17:21.080 --> 0:17:22.520 talking about what you can on the top of it, 0:17:22.720 --> 0:17:24.159 and the new S curves are kind of under the 0:17:24.240 --> 0:17:26.600 radar if I can mix my metaphors. One of the 0:17:26.640 --> 0:17:28.840 ways I described this, it's kind of a good Manhattan metaphor. 0:17:28.840 --> 0:17:30.880 It's like you walk past your construction site every day 0:17:30.880 --> 0:17:32.800 for six months and there's a bunch of construction worker 0:17:32.840 --> 0:17:34.840 is kind of scout standing around, scratching their backside, is 0:17:34.880 --> 0:17:36.760 not doing anything. You think all these guys are lazy. 0:17:37.040 --> 0:17:38.840 Then you walk past on Monday morning and they put 0:17:38.840 --> 0:17:40.760 fifteen stories of steel frame up me You think, wow, 0:17:40.760 --> 0:17:44.679 they're really busy over the weekend. You're missing the slow 0:17:46.000 --> 0:17:47.480 And then there's a period at the end where they're 0:17:47.520 --> 0:17:49.640 like putting the facade on and doing all the fit out, 0:17:49.720 --> 0:17:51.879 and again that looks so boring and so mobile is 0:17:51.920 --> 0:17:53.840 at the stage where you know you're putting the facade 0:17:53.840 --> 0:17:56.200 down and you're doing the fit out and the other 0:17:56.200 --> 0:17:57.840 staff are still kind of hulling the ground and you 0:17:57.880 --> 0:18:00.960 can't really see what's going on. Um And so that 0:18:01.000 --> 0:18:04.200 means the character the questions change. In the early stages 0:18:04.240 --> 0:18:07.720 of investments. There there is no data, there's no discounted 0:18:07.720 --> 0:18:12.080 cash forlor model. You're really dealing with two founders and 0:18:12.119 --> 0:18:15.439 a PowerPoint presentation and some numbers which are more or 0:18:15.520 --> 0:18:21.199 less best guesses. Yeah, so the way, it's an interesting 0:18:21.240 --> 0:18:23.520 shift in being an equity analyst because you're right at 0:18:23.560 --> 0:18:25.480 one end of the risk profile. You know, if you 0:18:25.520 --> 0:18:27.520 have tourn of buying tea bills is at the other end. 0:18:28.040 --> 0:18:30.800 You're at the stage where half of the deals you 0:18:30.840 --> 0:18:34.200 do will return less than invested capital. And that's the plan, 0:18:35.320 --> 0:18:39.000 and five pent will produce more than a ten x 0:18:39.040 --> 0:18:41.760 return and that gets you a kind of a three 0:18:41.880 --> 0:18:45.400 x return over ten years. And so everything you do 0:18:45.600 --> 0:18:48.120 has to be capable of being amazing, and if it's 0:18:48.160 --> 0:18:50.200 capable of going from two people at a power point 0:18:50.400 --> 0:18:52.960 to being an amazing thing worth hundreds of millions of 0:18:52.960 --> 0:18:54.959 billions of dollars, it kind of has to be implausible 0:18:55.000 --> 0:18:56.879 and crazy, and they have to be a bunch of 0:18:56.920 --> 0:18:59.840 reasons why it might not work. And therefore what you 0:19:00.040 --> 0:19:02.760 have to do is sort of suspend disbelief and not think, 0:19:02.800 --> 0:19:04.840 here are all the reasons why this might not work, 0:19:04.880 --> 0:19:07.280 But what if this did work, what would it be 0:19:07.720 --> 0:19:10.080 and are these the people who can make that happen, 0:19:10.720 --> 0:19:13.679 Which is a very very different approach than thinking about, 0:19:14.160 --> 0:19:17.200 all right, will this new widget or this new management 0:19:17.200 --> 0:19:20.719 team or whatever sell enough to move the earnings pre 0:19:20.760 --> 0:19:25.159 share calculus of this publicly trading exactly. I think the 0:19:25.240 --> 0:19:27.160 kind of the key way to think about Silicon Valley 0:19:27.240 --> 0:19:30.879 is it's a machine for running experiments, and most of 0:19:30.920 --> 0:19:33.840 the experiments won't work, and that's the plan. And yes, 0:19:33.960 --> 0:19:35.960 you know you might. You know, you could do some 0:19:36.080 --> 0:19:38.200 an experiment that's clearly a terrible idea and it was 0:19:38.240 --> 0:19:39.600 never going to work, and you could kind of mess 0:19:39.600 --> 0:19:40.840 it up and blow up your lab, and you know, 0:19:40.840 --> 0:19:42.679 people will look down on you for that, but no 0:19:42.680 --> 0:19:44.040 one will look down on you for the fact that 0:19:44.040 --> 0:19:46.320 you ran an experiment in a produced negative result that 0:19:46.400 --> 0:19:49.199 was just okay, well we tried, and do you do 0:19:49.200 --> 0:19:51.480 the do you run the experiment, right, that's a different question. 0:19:51.680 --> 0:19:54.080 But you ran the experiment, it didn't work, Okay, that's fine. 0:19:54.640 --> 0:19:57.920 And the ones that do work justify the whole exercise 0:19:57.960 --> 0:20:00.960 and pay for the whole exercise and produce mobile phone, 0:20:01.040 --> 0:20:03.880 saw produced Apple and Google and so on. So let's 0:20:03.920 --> 0:20:06.680 let's let's zoom in on that a little bit. When 0:20:06.680 --> 0:20:08.840 you're at this side of the funnel, when you're looking 0:20:08.880 --> 0:20:13.080 at these really early stage companies, if half of them 0:20:13.240 --> 0:20:16.679 are effectively money losers for you, for the farm as 0:20:16.720 --> 0:20:22.120 an investment. How do you conceptualize what you're looking at? 0:20:22.359 --> 0:20:25.720 Is it is it the founders themselves? Is it the idea? 0:20:26.359 --> 0:20:29.280 Is it something that, hey, this is just so crazy 0:20:29.280 --> 0:20:32.560 it might work. What what is the thought process like when, 0:20:33.200 --> 0:20:37.199 especially where you sit, where there's an endless stream of 0:20:37.240 --> 0:20:41.440 people coming to Silicon Valley to pitch ideas to vcs. Well, 0:20:41.520 --> 0:20:44.480 so I think most venture capitalists look at this stuff 0:20:44.600 --> 0:20:46.760 mostly in the same way. You can argue a little 0:20:46.760 --> 0:20:50.479 bit about emphasis, um, but there is the question of 0:20:50.520 --> 0:20:54.960 what is the market opportunity here? And um? Are these 0:20:55.040 --> 0:20:57.119 the people who are going to be able to work 0:20:57.119 --> 0:21:00.040 out find discover that market opportunity because they never to 0:21:00.119 --> 0:21:02.520 PE's not going to be exactly the original idea. It's 0:21:02.520 --> 0:21:04.360 going to be something sort of adjacent to that. You'll 0:21:04.440 --> 0:21:06.879 kind of twist around and find it pivot shall we? 0:21:07.440 --> 0:21:09.160 Not so much that I'm more iterate. I would say 0:21:09.200 --> 0:21:11.439 pivoty is more like, Okay, we built a whole company 0:21:11.440 --> 0:21:13.159 on that promise and that didn't work at all, so 0:21:13.280 --> 0:21:15.840 now let's try something else. Um, that's not really the 0:21:15.880 --> 0:21:18.600 same thing. It's more kind of well, we were kind 0:21:18.600 --> 0:21:20.720 of that and we sort of moved around, we got, 0:21:20.760 --> 0:21:23.280 we made, we found it worked. There's this whole sort 0:21:23.280 --> 0:21:26.040 of concept in silicon value product market fit, and so 0:21:26.160 --> 0:21:27.720 you have sort of a product and you're trying to 0:21:27.800 --> 0:21:29.520 work out, well, what the market being, what would the 0:21:29.520 --> 0:21:32.240 product be around this space until we can find something 0:21:32.280 --> 0:21:35.960 that meshes and takes off. And so the earlier you are, 0:21:36.400 --> 0:21:39.560 the more in a sense you're betting on the ability 0:21:39.600 --> 0:21:42.800 of the founders to find that, But you're also betting 0:21:42.800 --> 0:21:45.360 on is this a great market and is there some 0:21:45.480 --> 0:21:47.439 angle or some way that they're going to find in 0:21:47.560 --> 0:21:49.560 order to make in order to turn this into a thing. 0:21:49.800 --> 0:21:52.920 So what I'm hearing from you as future growth prospects 0:21:53.040 --> 0:22:00.120 or and valuation is well, well so, um, it depends. 0:22:01.040 --> 0:22:04.840 But the more progress you make and the more numbers 0:22:04.880 --> 0:22:07.600 you have, the more that you start looking at metrics 0:22:07.760 --> 0:22:10.800 and the less that you start thinking about potential. Um. 0:22:10.960 --> 0:22:13.520 So it's sort of the super super early stage. Then 0:22:13.560 --> 0:22:16.480 the valuations all tend to look the same as you 0:22:16.520 --> 0:22:19.520 go further in. Then you start getting much more specific 0:22:19.600 --> 0:22:21.399 about well, how well are they doing and what do 0:22:21.440 --> 0:22:23.640 we think this looks like? Um. Then it gets sort 0:22:23.640 --> 0:22:25.040 of gets kind of case by case, and you know, 0:22:25.080 --> 0:22:27.159 you get up to a company that's got billions of 0:22:27.200 --> 0:22:29.160 dollars a revenue, and then you're doing dcs, and you're 0:22:29.200 --> 0:22:32.800 doing multiples like anybody else, but at the super early stage. 0:22:32.800 --> 0:22:34.600 In a sense, doing a DCF on two people with 0:22:34.640 --> 0:22:36.840 a power point is just an exercise in self deception. 0:22:37.040 --> 0:22:39.080 Sure you could do it. You could. You could sort 0:22:39.080 --> 0:22:42.480 of say, well, if they managed to sell this thing 0:22:42.520 --> 0:22:44.680 to two billion people, well there will be some value 0:22:44.720 --> 0:22:47.040 that trying to do a dcattle that. Well, like, what 0:22:47.080 --> 0:22:48.600 does that get me? It doesn't tell me anything. You're 0:22:48.640 --> 0:22:53.240 just making Yeah, and the things that really work, um, 0:22:53.400 --> 0:22:55.280 in a sense, it doesn't kind of matter if this 0:22:55.400 --> 0:22:57.160 is you know, if you've got you know, take give 0:22:57.160 --> 0:22:59.960 me a kind of hypothetic hypothetical example. You've got to 0:23:00.000 --> 0:23:03.040 fifty million dollar seed fund and you make a significant 0:23:03.040 --> 0:23:05.639 investment in something that turns out to be worth fifty 0:23:05.640 --> 0:23:08.560 billion dollars. Does it really matter if it's worth forty 0:23:08.560 --> 0:23:11.879 billion or sixty billion. The amount of your assist in 0:23:11.920 --> 0:23:13.960 the start, the amount of the amount that it's returned 0:23:14.000 --> 0:23:16.600 your fund is sign is sufficient that those numbers don't 0:23:16.600 --> 0:23:21.280 really make any difference. So so that that's really fascinating 0:23:21.280 --> 0:23:23.959 and end Mark and reason when when we said then 0:23:24.040 --> 0:23:28.440 had a conversation said something similar. But looking back at 0:23:28.440 --> 0:23:31.760 a perspective of twenty years of doing this, and if 0:23:31.760 --> 0:23:33.359 you go back and if you would have paid double 0:23:33.440 --> 0:23:36.600 for everything, it wouldn't mean any difference whatsoever. Yeah, I 0:23:36.600 --> 0:23:39.040 think there's there's a famous investor whose name I think, 0:23:39.359 --> 0:23:41.240 I think I should think it was somebody in Hollywood, 0:23:41.240 --> 0:23:43.160 of famous producer. He said something, if I had said 0:23:43.200 --> 0:23:44.800 yes to all the ones I said no to, and 0:23:44.840 --> 0:23:46.200 no to all the ones I've said yes to, you, 0:23:46.200 --> 0:23:48.359 I would have come out exactly the same place. So 0:23:48.400 --> 0:23:49.880 there's always, you know, this is kind of the index 0:23:50.040 --> 0:23:52.200 indexing story. There's always you can always kind of push 0:23:52.240 --> 0:23:55.200 these arguments and say, well then we should just nobody 0:23:55.240 --> 0:23:57.240 knows anything, And there's a little bit more to it 0:23:57.280 --> 0:24:01.200 than that. Let's talk a little bit of out technology. 0:24:01.760 --> 0:24:04.320 You you have a quote I really like, you have 0:24:04.400 --> 0:24:07.760 several quotes I really like. Um, And let's start with 0:24:08.080 --> 0:24:11.120 one or two of thesegncy where they go. All social 0:24:11.200 --> 0:24:14.080 apps grow until you need a news feed or news 0:24:14.119 --> 0:24:17.960 news feeds grow until you need an algorithm. All algorithmic 0:24:18.000 --> 0:24:20.560 feeds grow until you get fed up seeing the wrong stuff. 0:24:20.920 --> 0:24:25.000 And leave for a new app with less information overload. 0:24:25.280 --> 0:24:28.960 So is the nature of all tech Rinse, lather, repeat 0:24:29.080 --> 0:24:34.000 and and something becomes an incumbent becomes successful and the 0:24:34.040 --> 0:24:35.960 new guys are just going to come up and eat 0:24:35.960 --> 0:24:38.440 their launch. So I have notes through a blog post 0:24:38.600 --> 0:24:42.920 in my phone which is called something like um technology determinism. 0:24:42.960 --> 0:24:45.119 And so there are as it might be half a 0:24:45.160 --> 0:24:48.520 dozen things which are just steady processes, and now half 0:24:48.560 --> 0:24:51.120 a dozen things that of cycles. So steady process would 0:24:51.119 --> 0:24:53.720 most obvious one would be More's law. Uh, there was 0:24:53.760 --> 0:24:56.920 a process of technology news from research labs to start 0:24:57.000 --> 0:24:59.600 ups to pick companies. Um. You know, you can imagine 0:24:59.600 --> 0:25:01.960 like kind of half a dozen of those. Um. Then 0:25:01.960 --> 0:25:04.439 there are cycles. And so there are cycles where you 0:25:04.480 --> 0:25:07.159 go from bundling to unbundling, You go from the client 0:25:07.280 --> 0:25:10.639 to the server and back again. You go from the 0:25:10.720 --> 0:25:14.760 public markets. We had conglomerates and then deconglomerization and then 0:25:14.800 --> 0:25:19.480 re conglomerates. Exactly exactly. Colleague, you at Steven Sinofsky who 0:25:19.680 --> 0:25:22.400 used to run office. He says, all products expand until 0:25:22.400 --> 0:25:25.919 they can edit photographs. That's very funny, like word can 0:25:26.080 --> 0:25:28.840 edit photographs, can edit photographs? All products expand until they 0:25:28.840 --> 0:25:31.159 can edit it. That's the endpoint of software. And so 0:25:31.200 --> 0:25:33.200 there were kind of there and it's a comment about, 0:25:33.200 --> 0:25:34.800 you know, feature creep or whatever it is you want 0:25:34.800 --> 0:25:36.359 to say. And so there were these kind of inevitable 0:25:36.400 --> 0:25:39.080 processes or inevitable pieces of logic that kind of flow through. 0:25:39.280 --> 0:25:41.560 Now the one that you were sort of quoted, particularly 0:25:41.800 --> 0:25:44.040 in sort of an observation about I'm actually I wrote 0:25:44.160 --> 0:25:45.720 a blog post about this last week. It's sort of 0:25:45.720 --> 0:25:48.440 a combination of dune Bar's number and zarka Berg's law. 0:25:48.840 --> 0:25:50.919 Dunebar's number is like, you know, like a hundred and 0:25:50.920 --> 0:25:53.120 fifty or two hundred people, well enough that you would 0:25:53.160 --> 0:25:56.200 trend them on Facebook at the way least. And you've 0:25:56.240 --> 0:25:58.800 got these social apps which make it easy to post 0:25:58.880 --> 0:26:01.920 stuff and share stuff. And because it's one too many, 0:26:01.960 --> 0:26:03.960 you're not emailing it to someone or texting it to someone. 0:26:04.000 --> 0:26:05.680 You can close quite a lot because you don't feel 0:26:05.720 --> 0:26:07.400 like you're kind of imposing on people to do that. 0:26:07.960 --> 0:26:11.040 But you've got two hundred friends and they post five 0:26:11.040 --> 0:26:13.680 things a day, Okay, Now you've got a thousand things 0:26:13.680 --> 0:26:16.480 a day in your news feed, and you can't read notes, 0:26:17.240 --> 0:26:19.359 and so this is the logic that gets Facebook to 0:26:19.600 --> 0:26:22.880 producing what is you know, what's called an algorithmic feed, 0:26:22.880 --> 0:26:25.240 which is just like engineers speak for, let's try and 0:26:25.280 --> 0:26:27.000 work out which of your friends we care about, and 0:26:27.040 --> 0:26:28.840 maybe we should put those at the top. And let's 0:26:28.840 --> 0:26:30.600 work out that you like these kind of things. You 0:26:30.640 --> 0:26:32.879 don't really like new stories from the Guardian. You prefer 0:26:32.960 --> 0:26:34.560 to look at pictures of baby So let's put the 0:26:34.560 --> 0:26:36.240 babies in front of the new stories from the Guardian. 0:26:36.520 --> 0:26:37.959 And you kind of you, you kind of you come 0:26:38.000 --> 0:26:41.160 up and you create that. And so now you've got, 0:26:41.200 --> 0:26:43.600 instead of, so to speak, a random sample, which is 0:26:43.640 --> 0:26:45.879 I open the app, what have people posted in the 0:26:45.960 --> 0:26:47.919 last hour, because I'm not going to scroll past the 0:26:47.960 --> 0:26:50.399 last hour's worth of post So actually it's random, the 0:26:50.480 --> 0:26:52.520 random point fact of being what time did open the app. 0:26:52.960 --> 0:26:56.280 So that's the linear feed. The chronological feed is random. 0:26:56.320 --> 0:26:57.639 So then you said, well, maybe we should put the 0:26:57.640 --> 0:27:00.480 stuff that's important at the beginning. But then you think, okay, 0:27:00.480 --> 0:27:03.000 but it's now you are arguing, well, why isn't the 0:27:03.000 --> 0:27:05.000 Guardian all the New York Times at the top. That 0:27:05.000 --> 0:27:06.719 should be up at the top, because that's important as 0:27:06.760 --> 0:27:09.240 a public benefit to that and you got this wrong. 0:27:09.480 --> 0:27:11.480 My friend posted this thing I wanted to see, and 0:27:11.480 --> 0:27:14.280 I didn't see it, and so you get that sense of, well, 0:27:14.320 --> 0:27:17.000 maybe this isn't actually working, and you have Russians trying 0:27:17.040 --> 0:27:18.719 to game it, and you have all sorts of problems 0:27:18.720 --> 0:27:21.240 with trying to make that feed work, and so then 0:27:21.280 --> 0:27:23.000 you then you can say, well, actually, what I want 0:27:23.040 --> 0:27:24.840 to do is if I really care about this stuff 0:27:24.840 --> 0:27:26.199 and I want people to see it, I'll send them 0:27:26.200 --> 0:27:27.840 a text message, I'll do it and WhatsApp, I'll do 0:27:27.880 --> 0:27:31.880 it on Facebook Messenger. But then I've got fifteen parallel 0:27:31.880 --> 0:27:34.560 conversations or twenty parallel conversations with people. And then we 0:27:34.560 --> 0:27:37.480 start creating WhatsApp groups where twenty people from school or 0:27:37.560 --> 0:27:39.000 can work and all talk to each other. And then 0:27:39.040 --> 0:27:41.000 everyone is like posting more and more stuff, and you think, 0:27:41.320 --> 0:27:43.000 I really like to have a screen in this app 0:27:43.040 --> 0:27:44.959 that just showed me like the important stuff from all 0:27:44.960 --> 0:27:46.920 of these chats that I would see it, and maybe 0:27:47.000 --> 0:27:49.800 that should be sorted by which are the ones that 0:27:49.840 --> 0:27:51.399 I want to see, And so like you kind of 0:27:51.400 --> 0:27:54.679 create the same process over and over again. Now this 0:27:54.760 --> 0:27:56.679 is an opinion. This may be entirely be wrong. This 0:27:56.800 --> 0:27:58.679 may not be how it works, but you can certainly 0:27:58.680 --> 0:28:01.959 sort of see that problem that if you create tools 0:28:01.960 --> 0:28:04.400 that let everyone you've ever met share anything they've they've 0:28:04.400 --> 0:28:06.320 ever been interested in, then you're not going to be 0:28:06.359 --> 0:28:08.560 able to read it all. Well, this leads me to 0:28:09.119 --> 0:28:13.240 the very related quote of Facebook's engineering effort goes to 0:28:13.280 --> 0:28:16.440 stuffing more noise into your news feed and the other 0:28:16.440 --> 0:28:19.280 fift is working on ways to filter it out, or 0:28:19.280 --> 0:28:21.320 it's another expression at the same point. This is this 0:28:21.480 --> 0:28:23.479 is classic joke and I read in a book by 0:28:23.480 --> 0:28:26.240 Casting Glany, which is that somebody's dug a hole to 0:28:26.359 --> 0:28:28.959 build a to create the foundation to build put up 0:28:28.960 --> 0:28:31.000 a building, and they asked their friend what should I 0:28:31.000 --> 0:28:32.680 do with all this earth that's come out of the whole, 0:28:32.760 --> 0:28:34.399 and their friends says, We'll just dig another hole and 0:28:34.440 --> 0:28:38.280 put it in that, And well, it's just not gonna 0:28:38.280 --> 0:28:40.200 work like that. You can't do it like that. And 0:28:40.280 --> 0:28:42.400 if you've created a system that lets anybody you've ever 0:28:42.440 --> 0:28:44.360 met send you anything that they feel like sending, then 0:28:44.360 --> 0:28:45.840 you're not going to be able to fiddle. And there's 0:28:45.840 --> 0:28:48.080 not like some magic algorithm that's going to make that work. 0:28:48.080 --> 0:28:50.440 You're and you're ever gonna have a sample. So I 0:28:50.480 --> 0:28:52.120 want to talk to you about some of your favorite 0:28:52.120 --> 0:28:56.280 technologies and future projects. Do we want to discuss it 0:28:56.400 --> 0:28:59.640 all what's been going on with Facebook this year and 0:29:00.320 --> 0:29:06.040 everything from Russian bots to the um just changes with 0:29:06.120 --> 0:29:10.000 Cambridge Analytics and scraping. First of all, I want to 0:29:10.040 --> 0:29:12.560 say that UM Oxford Analytics is actually not nearly as 0:29:12.600 --> 0:29:17.400 good as Candy Channelized. Okay, for sure, a little little 0:29:17.440 --> 0:29:21.840 almamount of joke, but how much. I'm kind of surprised 0:29:21.880 --> 0:29:27.560 at how shocked people are about this full disclosure. I 0:29:27.600 --> 0:29:32.280 have not been a Facebook heavy user ever. I chafed 0:29:32.320 --> 0:29:34.760 at everything they've ever asked me to give them. They 0:29:34.760 --> 0:29:36.760 don't have my real birthday, they don't have my real phone, 0:29:36.760 --> 0:29:39.239 they don't have my real email address, they don't have 0:29:39.320 --> 0:29:42.200 my real anything. So anyone who wants to scrape other 0:29:42.280 --> 0:29:44.600 than where I went to college, where I went to 0:29:44.640 --> 0:29:48.320 grad school, and and a couple of jobs I've worked 0:29:48.320 --> 0:29:50.959 at that are public, they could scrape what they want 0:29:51.040 --> 0:29:54.200 from me. They're they're getting nothing. But I think this 0:29:54.280 --> 0:30:00.360 whole headache was readily foreseeable by anyone who who forget 0:30:00.440 --> 0:30:04.120 reading Facebook's terms of use. No, you're not entitled to 0:30:04.160 --> 0:30:06.680 any of that private information. I'm not sharing that with you. 0:30:07.040 --> 0:30:09.400 I'll find what I what I'm interested in my own 0:30:09.880 --> 0:30:12.240 or am I just outside of their demographic I'm an 0:30:12.240 --> 0:30:17.400 old fogy And so I think there's a bunch of 0:30:17.720 --> 0:30:22.760 kind of unresolved feeling about Facebook. Is it private or public? 0:30:23.280 --> 0:30:25.280 If I post stuff in the news feed, will my 0:30:25.320 --> 0:30:28.120 friends see it or not? What does that mean? Do 0:30:28.160 --> 0:30:29.840 I want to share? Do I want to share this 0:30:29.920 --> 0:30:33.200 stuff or not? And like, we sort of understand that 0:30:33.280 --> 0:30:35.520 if you search on Google for something, it'll show you 0:30:35.680 --> 0:30:37.240 or try to show you what you asked for, even 0:30:37.280 --> 0:30:39.920 if it's something you shouldn't have searched for. We don't 0:30:40.080 --> 0:30:43.320 really feel like if I my racist uncle posts that 0:30:43.400 --> 0:30:46.239 story on Facebook, should I see it or not? How 0:30:46.280 --> 0:30:48.640 do we think about that? Well, if you like it 0:30:48.840 --> 0:30:51.560 or reposted. But we also think we also sort of 0:30:51.560 --> 0:30:53.640 think or maybe Facebook shouldn't be showing that at the 0:30:53.680 --> 0:30:55.520 top of the list. Well, well it is my he 0:30:55.600 --> 0:30:58.600 is my uncle, and he did post it. So we 0:30:58.760 --> 0:31:01.920 have a bunch of sort of of I don't think 0:31:01.920 --> 0:31:04.960 we have a clear sentiment about put its other extreme, 0:31:05.160 --> 0:31:07.000 we are confident that are back. We are comfortable with 0:31:07.040 --> 0:31:08.680 the fact that our banks know how much money we have. 0:31:09.440 --> 0:31:11.640 For sure, we're comfortable that our mobile operators and know 0:31:11.640 --> 0:31:13.400 where our phone is and that there's a kind of 0:31:13.400 --> 0:31:16.240 a legal apparatus around that if the police need to know, 0:31:16.320 --> 0:31:18.080 then they can find out. But it's just not available 0:31:18.120 --> 0:31:20.360 to anybody. I think we sort of have that feeling 0:31:20.360 --> 0:31:24.240 around Google mostly. I don't think we have like a 0:31:24.280 --> 0:31:28.440 resolved feeling around Facebook. Um. Now, this particular story is 0:31:28.440 --> 0:31:30.600 sort of fascinating because of how many pieces of there 0:31:30.720 --> 0:31:34.640 there are to that. So Facebook creates this developer platform 0:31:34.840 --> 0:31:37.360 that allows you to install an app that can access 0:31:37.360 --> 0:31:40.560 your information and also access information about your friends. And 0:31:40.600 --> 0:31:42.400 there's a sort of a bunch of reasons why that 0:31:42.440 --> 0:31:44.920 would be useful, like I want to create in a 0:31:44.960 --> 0:31:47.120 calendar entry, or I want to you know, I want 0:31:47.160 --> 0:31:49.840 to see who else is using this app. Um. And 0:31:50.000 --> 0:31:53.040 a large part of like the advocacy around the tech 0:31:53.080 --> 0:31:55.840 industry with wall gardens are bad, Facebook is bad because 0:31:55.880 --> 0:31:57.680 it's closed. People being need to be able to get 0:31:57.680 --> 0:31:59.480 their information out. You need to be able other people 0:31:59.520 --> 0:32:01.920 need to be able to use this to innovate. So 0:32:02.080 --> 0:32:04.680 there was this whole sort of ideological argument um that 0:32:04.720 --> 0:32:06.840 Facebook would be evil for not doing this, and Facebook 0:32:06.880 --> 0:32:10.360 needed to do this, So you create this platform. Um, 0:32:10.400 --> 0:32:13.080 it turns out that people are able to exploit that 0:32:13.280 --> 0:32:16.080 and do stuff with it that was not really anticipated. 0:32:16.400 --> 0:32:19.719 You think it's not anticipated because from my perspective, I 0:32:19.760 --> 0:32:23.400 look at it as by design that Okay, so yes 0:32:23.440 --> 0:32:25.520 and no, So let's get let's go to an analogy. 0:32:25.600 --> 0:32:30.520 So do you remember word macro viruses? Um? Sure about? 0:32:30.920 --> 0:32:34.840 So office is supposed to be an open development environment. 0:32:35.040 --> 0:32:37.480 The people suing them before antitrust is suing them from 0:32:37.520 --> 0:32:39.760 making it closed in half for third parties to work with. 0:32:40.200 --> 0:32:42.120 So they create all these API s and they create 0:32:42.160 --> 0:32:44.560 this whole macro language. One of the things on page 0:32:44.560 --> 0:32:47.160 fifteen of the textbook is you can make a macro run. 0:32:47.160 --> 0:32:50.160 When you open the document page forty eight of the textbook. 0:32:50.520 --> 0:32:52.280 You can get it to look at your email addresses. 0:32:52.400 --> 0:32:56.160 Page seventy two. You can get it to send an email. Okay, 0:32:56.200 --> 0:32:58.239 so I get an email word document, I open it, 0:32:58.240 --> 0:32:59.920 it emails a copy of itself to everyone in might 0:33:00.080 --> 0:33:04.959 s book at. Okay, that's not what we expected, but 0:33:05.080 --> 0:33:07.360 it's really that is not its intended but it's not 0:33:07.400 --> 0:33:10.920 its intended purpose exactly. But all of the individual A 0:33:11.000 --> 0:33:13.239 p I s were there, and so you've got this 0:33:13.320 --> 0:33:15.840 period from Microsoft where they're thinking, okay, so are we 0:33:15.880 --> 0:33:18.640 supposed to not have macros that we're supposed to be closed? Now, 0:33:18.840 --> 0:33:20.440 how do we think about this? And they had to 0:33:20.440 --> 0:33:22.959 go through this like hundred and eighty degree turn as 0:33:23.000 --> 0:33:25.120 they went from thinking we should make it as easy 0:33:25.160 --> 0:33:28.040 as possible for anybody to do this to what would 0:33:28.040 --> 0:33:32.800 happen if it's going to moderate my language? What would 0:33:32.800 --> 0:33:37.080 happen if a bad person, um what my seven year 0:33:37.080 --> 0:33:40.080 old would call a dingly head, decided to read the 0:33:40.120 --> 0:33:42.800 textbook Because it's not like you found bugs in it. 0:33:42.920 --> 0:33:44.680 They're doing stuff it was in the textbook, that was 0:33:44.720 --> 0:33:46.520 in the manual, but you weren't expecting them to use 0:33:46.560 --> 0:33:48.640 them in those ways. And I think there's a very 0:33:48.680 --> 0:33:52.040 strong parallel there with Whatness Cambridgewn Litigo was doing, which 0:33:52.120 --> 0:33:53.760 was you were able to install an app. The app 0:33:53.800 --> 0:33:55.840 can get your information, The app can ask for your 0:33:55.840 --> 0:33:57.959 friends information. If you say yes, it will get your 0:33:57.960 --> 0:34:00.600 friends information. Yeah, but we didn't expect that people would 0:34:00.640 --> 0:34:04.560 use that to exportrate eighty million people's profiles, just as 0:34:04.600 --> 0:34:06.400 we didn't expect someone would make a word document that 0:34:06.400 --> 0:34:08.359 could email a copy of itself to me. So now, 0:34:08.440 --> 0:34:10.479 so now let me I'm gonna go toe to toe 0:34:10.480 --> 0:34:12.960 on technology with you, which is clearly a mistake on 0:34:13.040 --> 0:34:18.160 my pod. But Microsoft notorious for having all these weak 0:34:18.239 --> 0:34:23.840 security setups and easily exploitable that is such a foreseeable issue. 0:34:24.480 --> 0:34:28.200 Granted this little bit of hindsight bioso in hindsight, right, 0:34:28.400 --> 0:34:35.239 but you know Microsoft is home of the exploitable security era. 0:34:36.040 --> 0:34:39.239 My relationship with Amazon and with Apple is that I 0:34:39.360 --> 0:34:42.560 pay them for stuff and I expect a different level 0:34:43.160 --> 0:34:45.640 of trust and a different level of Hey, I'm already 0:34:45.640 --> 0:34:48.520 giving you money, don't exploit my personal data for other reasons. Well, 0:34:48.600 --> 0:34:50.480 let me let me give you a hypothetical. Um, you 0:34:50.480 --> 0:34:52.839 can install an app on your smartphone. It can pop 0:34:52.880 --> 0:34:55.120 up a box and ask for your friends, and there's 0:34:55.120 --> 0:34:57.200 an awful lot of sensible reasons, like you're playing a game, 0:34:57.200 --> 0:34:58.839 who else? Which of your friends are playing? The game? 0:34:59.200 --> 0:35:01.880 Installed the Instagram? Who are your friends? So you can 0:35:01.920 --> 0:35:03.759 follow them? Like we were asked if you want to 0:35:03.800 --> 0:35:08.520 split a fair with somebody, any basic logical reasons why 0:35:08.560 --> 0:35:10.399 you wanted to do that? Okay, So that app has 0:35:10.440 --> 0:35:15.879 just downloaded six hundred people's home addresses. Is that a breach? Well, 0:35:15.960 --> 0:35:19.759 it has a downloaded their email address or address or 0:35:19.760 --> 0:35:22.520 have they looked at the home address? And then people 0:35:23.320 --> 0:35:25.279 second order question it needs their email address or their 0:35:25.280 --> 0:35:27.120 phone number to work. That's the idea. There's there is 0:35:27.200 --> 0:35:30.319 no other identify, so it's got six d people's email addresses. Okay, 0:35:30.360 --> 0:35:33.080 is that a breach of privacy? Is that Apple's faults? 0:35:33.920 --> 0:35:38.560 Well maybe a little bit. I would say it is not. 0:35:38.680 --> 0:35:41.200 But Apple popped up They said, here is this capability. 0:35:41.239 --> 0:35:42.879 Apple pops up the thing. Do you want this app 0:35:42.920 --> 0:35:45.760 to to have your address or not? And the sort 0:35:45.840 --> 0:35:48.040 of a point where it's not kind of black and white. 0:35:48.040 --> 0:35:51.160 It's not like somebody hacked into Google and gave them email. 0:35:51.719 --> 0:35:54.000 And so I think Facebook has been sort of they 0:35:54.640 --> 0:35:56.920 they have this record of pushing this thing to the 0:35:57.000 --> 0:36:00.680 kind of the outer envelope um for the last fifteen years. 0:36:00.760 --> 0:36:03.560 And I think what's happened here slightly ironically, because they 0:36:03.560 --> 0:36:05.640 actually closed off all of these APIs like a couple 0:36:05.640 --> 0:36:07.600 of years ago. So sort of what happened is like 0:36:07.640 --> 0:36:10.400 the stable door was open if you were a developer, 0:36:10.440 --> 0:36:12.040 if you were in tech, if you went to the 0:36:12.120 --> 0:36:14.200 Developer event. They stood up on stage and said, hey, 0:36:14.200 --> 0:36:16.120 look we leave our stables store always doors open so 0:36:16.200 --> 0:36:18.759 you can do all this stuff. Isn't this great? And 0:36:18.880 --> 0:36:21.040 their doors stable doors are opened for like five years 0:36:21.600 --> 0:36:24.040 and then they maybe this isn't a good idea. We're 0:36:24.080 --> 0:36:25.239 going to close the door now. And a bunch of 0:36:25.280 --> 0:36:27.480 people said, oh, evil, Facebook, you shouldn't close the doors. 0:36:27.520 --> 0:36:29.960 You should allow free open access. And now here we 0:36:30.000 --> 0:36:31.640 are in two thousand and eighteen and people go, wow, 0:36:31.680 --> 0:36:33.279 a lot of people went into the stable and stole 0:36:33.320 --> 0:36:39.120 all the horses. What a shark? And Facebook, Yeah, but 0:36:40.239 --> 0:36:43.080 like we near, the doors were open. So there's a 0:36:43.120 --> 0:36:44.680 bunch of you can kind of see this as they 0:36:44.719 --> 0:36:46.359 try and work out how we talk about this, how 0:36:46.400 --> 0:36:48.640 do we think about this stuff. We have been speaking 0:36:48.640 --> 0:36:51.919 with Benlic Devans of Andres and Horowitz. If you enjoy 0:36:52.000 --> 0:36:55.080 this conversation, be sure and check out our podcast extras, 0:36:55.280 --> 0:36:58.560 where we keep the tape rolling and continue discussing all 0:36:58.680 --> 0:37:04.080 things technology. We love your comments, feedback and suggestions right 0:37:04.200 --> 0:37:07.480 to us at m IB podcast at Bloomberg dot net. 0:37:08.120 --> 0:37:10.759 Check out my daily column on Bloomberg Vie dot com. 0:37:10.960 --> 0:37:14.160 You could follow me on Twitter at Ri Halts, I'm 0:37:14.200 --> 0:37:18.520 Barry Hults. You're listening to Masters in Business on Bloomberg Radio. 0:37:31.560 --> 0:37:34.239 Welcome to the podcast, Bennetick, Thank you so much for 0:37:34.280 --> 0:37:37.800 doing this. You're you're talking about stuff that I really 0:37:37.840 --> 0:37:45.000 find endlessly fascinating, including, UM, the responsibility of Facebook to 0:37:45.840 --> 0:37:49.040 either have an open platform and what that risk entails, 0:37:49.640 --> 0:37:53.360 or to control their own APIs and control who access 0:37:53.520 --> 0:37:58.280 what from your UM feed, what nobody is talking about. 0:37:58.440 --> 0:38:01.520 And by the time this airs, he will have Mark 0:38:01.640 --> 0:38:07.480 Zuckerberg will have already um done his congressional testimony. But UM, 0:38:08.160 --> 0:38:11.680 there's still a tremendous amount of responsibility on the individual 0:38:11.800 --> 0:38:15.920 user who willingly said, hey, here's a ton of private 0:38:15.960 --> 0:38:19.480 personal information about my me, try not to mess it up. 0:38:20.000 --> 0:38:23.160 I was always too skeptical. I was always too why 0:38:23.200 --> 0:38:26.239 do you want this information? Can kind of noted it 0:38:26.239 --> 0:38:28.839 from the other end, which was I think people have felt. 0:38:28.920 --> 0:38:31.120 I think for years and years and years people have 0:38:31.160 --> 0:38:33.560 just sort of presumed that nothing on Facebook with public 0:38:33.600 --> 0:38:36.680 with private, right, that's correct, and therefore just sort of 0:38:36.680 --> 0:38:38.720 treated it as a public forum. If you were smart, 0:38:38.800 --> 0:38:40.680 that's what you presumed, or if you were I should 0:38:40.719 --> 0:38:43.279 say knowledgeable. That should have been your work, even think 0:38:43.520 --> 0:38:45.399 I didn't even think beyond that. So, like I've heard 0:38:45.400 --> 0:38:47.840 people say things like I don't use Facebook. I wouldn't 0:38:47.840 --> 0:38:49.560 say that on Facebook Messenger because I don't want it 0:38:49.560 --> 0:38:51.880 to be public, and of course favorit. Messager isn't public. 0:38:51.880 --> 0:38:55.319 I mean technically Facebook can, but anybody can screenshot and 0:38:55.320 --> 0:38:58.000 anybody can. Yeah. Well that's like people thought it was 0:38:58.000 --> 0:38:59.920 his public and stuff on their on their public profile, 0:39:00.080 --> 0:39:02.799 and I went, I joined Facebook, and I forget when 0:39:02.800 --> 0:39:05.399 it launched in the UK, but everyone's profile was public. 0:39:05.440 --> 0:39:06.759 I mean it was public if you were to if 0:39:06.760 --> 0:39:09.440 to like if you're in the London network, so anyone 0:39:09.480 --> 0:39:11.759 could join the London network, so it was public. And 0:39:11.800 --> 0:39:13.719 then so there was a period of like a year when, 0:39:13.880 --> 0:39:16.160 like everybody I've been at school, at university, we've had 0:39:16.160 --> 0:39:17.920 a profile on Facebook and it was public and I 0:39:17.920 --> 0:39:19.279 could go and look, and then everyone kind of turned 0:39:19.280 --> 0:39:22.480 the privacy things on. But I just and even then 0:39:22.560 --> 0:39:25.439 it's not for I don't feel like you would say 0:39:25.480 --> 0:39:29.000 stuff on Facebook that you expect to be secret. That's 0:39:29.040 --> 0:39:30.800 what I'm getting at, And I don't think many people 0:39:30.880 --> 0:39:33.279 ever actually thought that, which comes again to my point 0:39:33.320 --> 0:39:35.759 about sort of unresolved feelings, you can kind of you 0:39:35.800 --> 0:39:39.279 can go to the extreme and say, um, everything on 0:39:39.320 --> 0:39:42.200 Facebook should have been completely private and everyone should have 0:39:42.280 --> 0:39:44.759 understood that and everyone you know, that's not realistic. But 0:39:44.840 --> 0:39:46.840 I don't think that's actually very realistic. I think it 0:39:46.920 --> 0:39:48.719 was much more kind of nuanced and fuzzy than that. 0:39:48.840 --> 0:39:50.560 And I think you could argue everything in a WhatsApp 0:39:50.600 --> 0:39:52.600 group should be private, but I think that's a very 0:39:52.640 --> 0:39:54.919 different kind of form to posting on your news feed. 0:39:55.040 --> 0:39:59.480 But when I ask people, um, you know, sometimes you 0:39:59.480 --> 0:40:02.120 have to come out the discussion from an oblique angle. 0:40:02.719 --> 0:40:04.840 And I like to use radio as an example. I 0:40:04.880 --> 0:40:08.560 ask people what is radio cell and invariably they say advertising, 0:40:08.920 --> 0:40:12.239 And that's the wrong answer. The advertisers of the buyers. 0:40:12.960 --> 0:40:16.560 Radio sells an audience, and Facebook more or less has 0:40:16.600 --> 0:40:19.200 the same business model. So I find this interesting. There's 0:40:19.200 --> 0:40:20.920 a sort of there's a common meme even kind of 0:40:20.920 --> 0:40:23.680 in the tech industry where people say Facebook sales for information. 0:40:24.200 --> 0:40:26.319 Now they sell and you as a part of an audience. Well, 0:40:26.320 --> 0:40:28.919 it's interesting because in in a literal sense, if you're 0:40:28.920 --> 0:40:31.120 an advertising on Facebook, you don't get given a zip 0:40:31.120 --> 0:40:34.440 file of the profile of everybody who saw that. So 0:40:34.560 --> 0:40:36.799 in a literal sense, it's give not true. In a 0:40:36.880 --> 0:40:39.080 kind of a metaphorical sense, well, they're sort of they're 0:40:39.080 --> 0:40:41.680 selling the fact that they know like even like kind 0:40:41.680 --> 0:40:43.799 of a metaphorical sense. That kind of struggles me. I 0:40:43.840 --> 0:40:46.759 struggle with that as a statement, But I think the 0:40:46.760 --> 0:40:49.799 fact that people have those conversations reflects against sort of 0:40:50.080 --> 0:40:52.920 unresolved feelings about how we should think about this stuff. 0:40:52.920 --> 0:40:54.799 It's the same with this idea that's going around now 0:40:54.960 --> 0:40:56.800 that all the Facebook's problems can be blamed on the 0:40:56.800 --> 0:40:58.800 ad model, and you think, well, if they had a 0:40:58.840 --> 0:41:01.719 subscription model, would they not be having to how to 0:41:01.760 --> 0:41:03.960 develop a platform? Would you not have been posting your 0:41:03.960 --> 0:41:06.360 personal stuff to your news feed? Like why would that 0:41:06.400 --> 0:41:08.879 have made any difference? And you would you would still 0:41:08.920 --> 0:41:11.480 have the same opportunity for apps to come in and 0:41:11.520 --> 0:41:15.560 scrape that data before they closed the barn door. Yeah, 0:41:15.600 --> 0:41:17.279 And I mean I was talked about this because somebody, 0:41:17.320 --> 0:41:19.640 I think you say on Twitter's somebody who's had a journalism, 0:41:19.800 --> 0:41:21.520 have had a journalism in school, and they posted like 0:41:21.560 --> 0:41:23.600 an explanation of what came don We've been doing it. 0:41:23.840 --> 0:41:27.960 This is three weeks ago, and it's just it was 0:41:28.040 --> 0:41:30.120 interesting simply that they were. Three weeks on, we're still 0:41:30.120 --> 0:41:33.080 having people are still writing explanations of what happened because 0:41:33.120 --> 0:41:35.120 it's all sort of fuzzy and a pay can no 0:41:35.120 --> 0:41:38.719 one quite understandable. It's complex, and people like simple narratives 0:41:38.719 --> 0:41:41.839 with defineable good guys and bad guys. That makes it 0:41:41.880 --> 0:41:45.960 easier to to do. Nuance sort of gets lost on 0:41:45.960 --> 0:41:50.160 on cable television to the least. All right, let's talk 0:41:50.160 --> 0:41:55.200 about some of your favorite technologies. Autonomy of filling the blank, 0:41:55.239 --> 0:42:00.120 autonomous cars, autonomous uh, whatever, how do we how do 0:42:00.160 --> 0:42:04.640 you see the development of AI and autonomous everything? Have 0:42:04.760 --> 0:42:07.520 you got another hour? I do? I don't know if 0:42:07.560 --> 0:42:11.520 we are are, so I think what can we say 0:42:11.520 --> 0:42:13.319 about this? So, first of all, the reason that we're 0:42:13.320 --> 0:42:16.480 talking about autonomous cars now is because of what we 0:42:16.560 --> 0:42:19.520 call AI, which really means this new technology called machine learning, 0:42:19.600 --> 0:42:22.359 or technology that just started working called machine learning, which 0:42:22.400 --> 0:42:24.880 offers the prospect that a bunch of problems around autonomous 0:42:24.920 --> 0:42:26.680 cars might be solvable in a way that they really 0:42:26.680 --> 0:42:29.399 didn't seem to be easily solvable before. So, in that sense, 0:42:29.440 --> 0:42:31.600 you could say autonomy is a spinoff of machine learning 0:42:31.600 --> 0:42:33.120 all the break the advanced. The fact that we're interested 0:42:33.120 --> 0:42:34.960 in autonomy is a spinoff in machine learning, but there's 0:42:34.960 --> 0:42:36.960 altos of other stuff that machine learning does as well 0:42:37.400 --> 0:42:40.759 as we go to autonomy. UM. The sort of the 0:42:40.800 --> 0:42:42.719 way I kind of talk about this is I have 0:42:42.719 --> 0:42:44.920 a slide with a picture of a horseless carriage from 0:42:45.840 --> 0:42:48.880 It's a carriage with no horse, but otherwise it's nothing's changed. 0:42:49.200 --> 0:42:51.560 And I think that's what I hear when people say 0:42:51.640 --> 0:42:55.400 driverless car, that you've taken the steering wheel out maybe, 0:42:55.440 --> 0:42:57.319 but like nothing else has changed and it will still 0:42:57.400 --> 0:43:00.560 drive around like like autonomous cars mean us that drive 0:43:00.600 --> 0:43:03.239 like people, but without making mistakes or making the speed limit. 0:43:03.280 --> 0:43:05.520 And that's that's just a really shortsighted way of thinking 0:43:05.520 --> 0:43:07.520 about this. I think there's kind of two kind of 0:43:07.520 --> 0:43:09.560 building blocks to think about. The first is you can 0:43:09.680 --> 0:43:11.799 if you have when we have a fully autonomous world 0:43:11.880 --> 0:43:15.239 in end decades forty years, depending on what you think 0:43:15.239 --> 0:43:18.200 the s cards will look like. UM. And of coursely 0:43:18.280 --> 0:43:20.400 you have periods where some places will be much go 0:43:20.560 --> 0:43:22.320 much quicker, So like you might you might say Manhattan 0:43:22.360 --> 0:43:25.759 is autonomous only in the week in twenty thirty or something. Um. 0:43:26.320 --> 0:43:30.040 But at that point there's no accidents, well, certainly much 0:43:30.120 --> 0:43:33.279 less well basically no accidents because all the accidents are 0:43:33.280 --> 0:43:36.080 caused by human error, and you have much less congestion 0:43:36.160 --> 0:43:38.280 because you don't have traffic waves, you don't have accidents 0:43:38.320 --> 0:43:40.600 which cause a third of congestion or something. UM. You 0:43:40.640 --> 0:43:42.920 can have vehicles on freeways driving a hundred miles and 0:43:42.960 --> 0:43:45.839 our two apart from each other. So you radically change 0:43:45.840 --> 0:43:48.120 what congestion looks like. You change radically change all the 0:43:48.200 --> 0:43:50.640 vehicles look like. So you could have a vehicle that 0:43:50.840 --> 0:43:52.840 will never you know, if you're got calling an on 0:43:52.880 --> 0:43:55.799 demand vehicle in Manhattan, it will not go over twenty miles. 0:43:55.800 --> 0:43:58.480 An how it could be a golf cart. Um. If 0:43:58.520 --> 0:44:00.319 you're going to go to JFK, they and you would 0:44:00.320 --> 0:44:03.000 send a different vehicle. And so you can radically just 0:44:03.080 --> 0:44:04.759 redesign the vehicle in the same way that when we 0:44:04.760 --> 0:44:06.840 got rid of the horse, you radically redesigned the vehicle. 0:44:07.280 --> 0:44:09.520 And you can also radically redesign the city or change 0:44:09.520 --> 0:44:11.360 your assumptions about the city in the same way that 0:44:11.400 --> 0:44:13.239 we did whom we got rid of the horse, um, 0:44:13.360 --> 0:44:17.040 except that you're no longer trying to design a city 0:44:17.480 --> 0:44:20.840 that will constrain the nineteen year old guy on a 0:44:20.840 --> 0:44:24.280 souped up Chevy Tomorrow. Um, you're designing a city based 0:44:24.320 --> 0:44:26.279 on You can create rules. You can tell the cars 0:44:26.280 --> 0:44:27.719 where to go and what they can do or not do. 0:44:27.960 --> 0:44:29.880 You can have dynamic real time road right saying that 0:44:29.920 --> 0:44:31.600 you can say, do you want to pay a lot 0:44:31.640 --> 0:44:33.440 to get there in fifteen minutes or you're willing to 0:44:33.440 --> 0:44:35.839 pay less to get there in thirty minutes. You can 0:44:35.880 --> 0:44:38.080 tell the cars at any given instant which road to 0:44:38.120 --> 0:44:40.080 say should be taking and how fast they should be going. 0:44:40.560 --> 0:44:42.680 And so what you have is like a change in 0:44:42.680 --> 0:44:45.359 the structure of a city that has more a lot 0:44:45.400 --> 0:44:47.560 in common with like the way the city changes a 0:44:47.600 --> 0:44:50.440 result of the car. UM kind of the example I 0:44:50.480 --> 0:44:53.560 often give, which works well since we're in Manhattan is 0:44:53.600 --> 0:44:57.200 imagine if you live in Brooklyn and it's November and 0:44:57.239 --> 0:44:59.560 you want to go to that cool new restaurant in Manhattan. 0:45:00.000 --> 0:45:01.439 How are you going to get there? Where you could 0:45:01.440 --> 0:45:04.040 walk ten minutes in the rain to the subway station, 0:45:05.360 --> 0:45:07.759 you could get a cab UK well, presuming you can 0:45:07.800 --> 0:45:09.359 get a cab. Presume you can get a cab O care. 0:45:09.400 --> 0:45:12.319 It's going to cost you what dollars each way you 0:45:12.320 --> 0:45:14.640 could drive, then one of you can't drink on the 0:45:14.640 --> 0:45:16.040 way back, and you're going to have to park, and 0:45:16.080 --> 0:45:18.239 you have to pay for parking in little twenty minutes 0:45:18.280 --> 0:45:20.839 to find a place to park there. Let's not go 0:45:20.880 --> 0:45:23.680 out of that place, okay, Now go to a fully 0:45:23.719 --> 0:45:27.440 autonomous world. You raise your watch, you say, hey, alex 0:45:27.520 --> 0:45:29.640 or I need a car. The nip hod that's around 0:45:29.640 --> 0:45:32.880 the block stops outside your door within thirty seconds, and 0:45:33.000 --> 0:45:35.600 there is no congestion. The pod just takes you there. 0:45:35.719 --> 0:45:38.160 It drops you off outside the door um if it's 0:45:38.200 --> 0:45:40.719 your pod, um, and then it goes and waits for 0:45:40.760 --> 0:45:42.680 you somewhere where parking is cheap, or it waits for 0:45:42.680 --> 0:45:44.520 you somewhere else. If it's an on demand pod, it 0:45:44.560 --> 0:45:48.040 immediately starts driving other people around, so there's no parking, remember, 0:45:48.320 --> 0:45:50.319 and then you can go home when you can drink 0:45:50.320 --> 0:45:52.120 as much as you like. I mean, remember all those 0:45:52.160 --> 0:45:56.480 those photographs of European and East Coast cities from before cars, 0:45:56.560 --> 0:45:58.200 and you think where the streets are all twice as 0:45:58.239 --> 0:46:00.759 wide because you don't have cars hop down both sides 0:46:00.800 --> 0:46:03.719 of the street. What we could go back to, and 0:46:03.800 --> 0:46:06.239 so you have like kind of radical change in how 0:46:06.280 --> 0:46:09.520 we think about what the city is. Therefore, in what 0:46:09.600 --> 0:46:11.759 does a gas station mean? Very obviously, but also what 0:46:11.800 --> 0:46:13.560 does big box we tell mean? Where are you willing 0:46:13.560 --> 0:46:17.040 to shop? What does your commute look like? You know? Today? 0:46:17.120 --> 0:46:19.800 So I live in San Francisco, which calls itself a city, 0:46:19.840 --> 0:46:21.959 and I work in Menlo Park, which is an office park. 0:46:22.120 --> 0:46:25.600 Forty five minutes drive south if there's not much traffic, Um, 0:46:25.640 --> 0:46:27.439 I should say an hour, but I'll admit to forty 0:46:27.480 --> 0:46:32.719 five minutes south. Now, supposing there really is no traffic, well, 0:46:32.760 --> 0:46:35.120 then I can get there in half an hour. But 0:46:35.360 --> 0:46:38.399 supposing I'm able to read instead of drive, I might 0:46:38.400 --> 0:46:40.720 be able to live further away and have it spend 0:46:40.760 --> 0:46:42.400 longer in the car because I don't need to just 0:46:42.440 --> 0:46:44.719 sit there staring at the road all day. So where 0:46:44.719 --> 0:46:46.400 do you live, where do you commute? Where does the 0:46:46.480 --> 0:46:49.759 retail go? All those sorts of changes that happened, like 0:46:50.000 --> 0:46:51.799 the stuff that happened as a result of cars, which 0:46:51.800 --> 0:46:53.600 basically what I'm talking about. Like, there's this great saying 0:46:53.600 --> 0:46:55.720 that it was easy to predict mass ownership of cars, 0:46:55.880 --> 0:46:57.799 but hard to put itt Walmart. It's hard to put 0:46:57.800 --> 0:47:01.000 it drive through a t M s like that kind 0:47:01.040 --> 0:47:04.200 of stuff. Those second and third auto consequences will flow 0:47:04.200 --> 0:47:07.799 out of this stuff. Let's talk about smart fill in 0:47:07.840 --> 0:47:12.239 the blanks, smart homes, smartphone, smart cars, smart speakers. What 0:47:12.440 --> 0:47:17.520 is it about the the preface smart that suddenly everybody 0:47:17.560 --> 0:47:22.960 wants to move in that direction? So so software and okay, 0:47:23.000 --> 0:47:24.719 so three or four blocks to talk about here. The 0:47:24.760 --> 0:47:27.520 first block is smartphone supply chain. One and a half 0:47:27.520 --> 0:47:30.520 billion smartphones sold last year. All of those chips are 0:47:30.520 --> 0:47:32.480 available as like a fire hose of stuff to make 0:47:32.520 --> 0:47:34.680 stuff with. So before smartphones, if you wanted to put 0:47:34.680 --> 0:47:37.279 computing into something, you had to use PC components, So 0:47:37.320 --> 0:47:39.959 an a t M is a PC. But those are big, 0:47:40.000 --> 0:47:42.279 and they're heavy, and they need mains power, they need 0:47:42.320 --> 0:47:45.080 power and so on. Smartphone components much cheaper, much smaller, 0:47:45.160 --> 0:47:47.120 much lighter. So suddenly you can make a connected door lock, 0:47:47.160 --> 0:47:50.759 and like you can get the parts really easily. UM 0:47:50.840 --> 0:47:53.839 plus um UBER called as internet plus machine learning, means 0:47:53.880 --> 0:47:56.319 that like a camera can actually be able to tell 0:47:56.320 --> 0:47:58.400 if there's something moving or not. So you've got all 0:47:58.480 --> 0:48:01.040 this stuff. Suddenly the stuff was not working. I think 0:48:01.080 --> 0:48:04.520 The best analogy for this is like our grandparents could 0:48:04.560 --> 0:48:06.879 have told you how many electric motors they owned. There 0:48:06.920 --> 0:48:09.000 was one in the car, they had a vacuum cleaner, 0:48:09.040 --> 0:48:10.879 there was one in the fridge. They owned maybe five 0:48:10.920 --> 0:48:13.480 electric motors in total, Like there was only one electric 0:48:13.480 --> 0:48:15.800 motor in the car. Let's start a major that was it. Today, 0:48:15.880 --> 0:48:17.480 who has a clue have any electric motors are in 0:48:17.480 --> 0:48:19.480 your car? There's like twenty And if you tell your 0:48:19.520 --> 0:48:21.640 grandfather that you can press a button to adjust your 0:48:21.640 --> 0:48:23.319 wing moor, he'd hit you on the back of the head. 0:48:23.440 --> 0:48:25.560 But that's just how it worked, because that's just how 0:48:25.600 --> 0:48:27.879 the technology got deployed. The same thing in your home. 0:48:28.120 --> 0:48:29.560 There was a period when everything was going to have 0:48:29.600 --> 0:48:32.520 a DC motor. Now everybody has you like you have 0:48:32.600 --> 0:48:34.840 an you have a microwave, You maybe have a toaster, 0:48:34.920 --> 0:48:37.160 you may have a kettle, you have a blender. Nobody 0:48:37.200 --> 0:48:38.880 has all of those things. Everyone in East Agea has 0:48:38.880 --> 0:48:41.040 a rice cooker. Everyone in the UK has a kettle. 0:48:41.080 --> 0:48:42.920 In America, you maybe don't have a kettle, but you 0:48:42.960 --> 0:48:46.200 have a coffee machine. But nobody has an electric carving knife. 0:48:47.080 --> 0:48:49.000 And so what happens is like there are those underlying 0:48:49.000 --> 0:48:51.480 components as cheap commodities were in this period of trying 0:48:51.480 --> 0:48:52.960 to work out what you should do with them and 0:48:53.000 --> 0:48:55.359 how they should all get plugged together, and should they 0:48:55.400 --> 0:48:56.960 all be the same system or not sure they all 0:48:56.960 --> 0:48:59.680 talked to Alexa or not everything, And then you've got 0:48:59.680 --> 0:49:02.640 like the austrial logic. So like somebody the Samson board 0:49:02.680 --> 0:49:04.800 sat down and said, everything we sell must have the 0:49:04.840 --> 0:49:07.759 Samsing Voice Assistant, because then people will be more likely 0:49:07.800 --> 0:49:09.680 to buy the Samson fridge that sorts to the Samsing 0:49:09.719 --> 0:49:12.280 dish watcher when it talks to the samthing this, um, 0:49:12.280 --> 0:49:14.399 how's it working out? Well? Then you so it's great 0:49:14.440 --> 0:49:16.160 you see this at CS because you can see like 0:49:16.200 --> 0:49:17.960 the fridge people thought this is a fantastic I mean 0:49:18.000 --> 0:49:20.160 metaphorically speaking, the fridge people thought, this is a fantastic 0:49:20.239 --> 0:49:23.760 idea to dish watch. Your people are like, don't damn it. Okay, 0:49:23.960 --> 0:49:25.920 we'll put the voice assistant in the dish wash but 0:49:27.760 --> 0:49:30.120 a that and be they want to sell it to 0:49:30.160 --> 0:49:32.640 people who also in an LG fridge and a sub 0:49:32.719 --> 0:49:35.160 zero fridge. So they've got the Samson voice is this, 0:49:35.320 --> 0:49:37.880 and they've got home kit and they've got Alexa in 0:49:37.920 --> 0:49:39.600 it because they only need to sell dish watches. They 0:49:39.640 --> 0:49:42.200 don't care about the Greek strategy. It's like when Everythingy 0:49:42.200 --> 0:49:44.319 product had a memory stick, and there was some and 0:49:44.400 --> 0:49:46.560 like the Sony group had said everything's gonna have have 0:49:46.600 --> 0:49:48.400 a memory stick, and some bits of Sony thought this 0:49:48.440 --> 0:49:51.000 is great, and some bits were like, God, damn it, um, 0:49:51.040 --> 0:49:52.160 what are we gonna do? What are we going to 0:49:52.239 --> 0:49:54.880 do with this? And as with an electric carving knife 0:49:54.960 --> 0:49:57.400 versus the blender, like some of this stuff will make them, 0:49:57.440 --> 0:49:59.520 so some of it won't. It would be quite nice 0:49:59.560 --> 0:50:01.960 to be able to say to my oven, um, okay, 0:50:02.000 --> 0:50:05.279 pre heat the oven to three degrees, as opposed to 0:50:05.360 --> 0:50:08.080 having to learner the interview and walk over and touch it. 0:50:08.560 --> 0:50:10.759 I have my apartment, I have an infrared sensor in 0:50:10.760 --> 0:50:12.640 the bathroom. I walk into the bathroom, the light comes on. 0:50:13.360 --> 0:50:16.200 It's fantastic. My parents hated because they get up at 0:50:16.200 --> 0:50:18.440 four o'clock in the morning, then the lights come on. 0:50:18.480 --> 0:50:21.239 I think you program that because it's not anything. It's 0:50:21.280 --> 0:50:23.839 just literally it is just a dumile sensor from forty 0:50:23.880 --> 0:50:26.480 years again. So so here's where the smart world starts 0:50:26.520 --> 0:50:29.600 to get more interesting. So I have a long, twisty 0:50:29.719 --> 0:50:32.719 driveway because we're Our house is said off the main road, 0:50:33.280 --> 0:50:37.120 and we have lights along it. And the first phase 0:50:37.320 --> 0:50:39.560 was having a switch that put the lights on and 0:50:39.600 --> 0:50:41.920 put the lights off. And then the next phase was 0:50:41.960 --> 0:50:45.040 having a timer that has the lights going on at 0:50:45.040 --> 0:50:48.680 seven a pm and off at eleven pm. But the 0:50:48.719 --> 0:50:51.880 problem with that is seven pm is light sometimes of 0:50:51.880 --> 0:50:56.160 the year and dark. I thought, no, no, nothing like 0:50:56.200 --> 0:51:00.839 that's a contemporary house. But the new switch that is 0:51:00.880 --> 0:51:06.360 literally going in this weekend is built into the lights 0:51:06.360 --> 0:51:10.520 which as you can set your latitude and that lets 0:51:10.840 --> 0:51:14.320 whatever you said as your like going on thirty minutes 0:51:14.360 --> 0:51:17.960 after sunset will change throughout the year regardless and you 0:51:18.000 --> 0:51:21.120 don't have to deal with it. So that's our sort 0:51:21.160 --> 0:51:24.120 of smart application of software. Are like, we don't have 0:51:24.160 --> 0:51:26.920 any you know, your your phone changes its clock when 0:51:27.000 --> 0:51:30.319 the time don't changed automatically things the satellites. So I 0:51:30.320 --> 0:51:32.000 think there's a there's a kind of question to where 0:51:32.000 --> 0:51:33.880 will the complexity set And this is like the electric 0:51:33.840 --> 0:51:35.600 carving knife. There will be stuff that just doesn't make 0:51:35.640 --> 0:51:38.440 sense the other you do electric carving knife have been 0:51:38.480 --> 0:51:40.520 around for decades and they never would the other the 0:51:40.560 --> 0:51:42.759 other really so well. Some people have a like I 0:51:42.800 --> 0:51:44.360 can show you where they took a chunk out of 0:51:44.360 --> 0:51:49.239 my finger. The other extreme is like the other. The 0:51:49.239 --> 0:51:51.200 other way to think about this is it's old saying 0:51:51.239 --> 0:51:53.680 that basically a computer should never ask you a question 0:51:53.680 --> 0:51:55.640 that it ought to be able to work out on itself. 0:51:56.640 --> 0:51:59.080 And that used to mean, like you plug the printer in, 0:51:59.120 --> 0:52:00.840 the computer should know what the printer is. It shouldn't 0:52:00.880 --> 0:52:03.279 ask you what printer it is. Um, then it means 0:52:03.360 --> 0:52:05.000 your phone doesn't ask where you are when you call 0:52:05.040 --> 0:52:07.640 a car because it's got GPS and it knows. Then 0:52:07.680 --> 0:52:09.479 it means like the light should know what the times 0:52:09.520 --> 0:52:11.239 it is, so the light should just go on right, 0:52:11.400 --> 0:52:14.200 that's right. And the chair that the tension point and 0:52:14.320 --> 0:52:16.840 all of those is, is it actually more hassle to 0:52:16.880 --> 0:52:19.680 configure the thing to do that? Sometimes it is, sometimes 0:52:19.760 --> 0:52:22.279 it is, sometimes it isn't. So my beef about Siri, 0:52:22.480 --> 0:52:26.760 which was the leading voice app to begin with before 0:52:27.560 --> 0:52:31.200 Alexa began to eat its lunch, is asking you questions 0:52:31.200 --> 0:52:34.439 that it should, from the context be able to figure out. 0:52:34.760 --> 0:52:38.600 This is just paying a database, paying a location, um 0:52:39.160 --> 0:52:42.520 software ping something it has access to all that stuff. 0:52:42.719 --> 0:52:45.760 This so this is a different sense of the word smart. 0:52:46.280 --> 0:52:48.920 So let's talk about voice for a minute. Okay, So 0:52:50.000 --> 0:52:52.359 we have this thing called machine so okay, the right 0:52:52.400 --> 0:52:54.840 way putting this. So when you talk to a computer, 0:52:55.360 --> 0:52:58.239 the three things that are going on. Step one is 0:52:58.320 --> 0:53:01.040 it is taking the audio wave form and turning that 0:53:01.120 --> 0:53:04.120 into text. So it's just transcribing it and turning it 0:53:04.160 --> 0:53:06.719 into words, which I think is basic math and and 0:53:07.000 --> 0:53:09.480 well no it was. And this is something where it 0:53:09.640 --> 0:53:11.879 used to work okay, like three quarters of the time, 0:53:12.400 --> 0:53:16.440 but you remember using dictation apps like National actually speaking, 0:53:17.120 --> 0:53:20.200 it's sort of work three quarters. There are more and 0:53:20.200 --> 0:53:22.359 it would get better like half a point every year. 0:53:22.800 --> 0:53:25.280 A machine learning comes along and it goes from working 0:53:25.280 --> 0:53:29.120 three quarters at the time to working at the tent 0:53:29.160 --> 0:53:30.600 at the time. And so machine learning gives you this 0:53:30.680 --> 0:53:34.000 radical change in that it also then there's a second piece, 0:53:34.040 --> 0:53:36.239 which is you needed to go from the text to 0:53:36.640 --> 0:53:39.360 a structured query. You need to actually work out what 0:53:39.440 --> 0:53:41.360 the verb and the nan is and what is his asking, 0:53:41.400 --> 0:53:45.160 which is a completely different computation. Um. And so when 0:53:45.200 --> 0:53:47.400 you talk to your computer, that's it's two very almost 0:53:47.480 --> 0:53:50.520 unrelated things going on. Machine learning. Also, this is called 0:53:50.600 --> 0:53:55.759 natural language processing. UM. Machine learning also made that work way, way, 0:53:55.800 --> 0:53:59.640 way way better. UM. Then you have the third problem 0:53:59.760 --> 0:54:02.440 we're is, Okay, I've created this structured query, do I 0:54:02.480 --> 0:54:06.759 have anything to give it to? And you should have. Well, 0:54:07.960 --> 0:54:11.000 so this is the problem, which is what you've actually 0:54:11.040 --> 0:54:13.799 got here is an IVYR. You've got a voice tree, 0:54:13.960 --> 0:54:16.880 Press one for this, press to for that. Machine learning 0:54:16.920 --> 0:54:18.759 means you can now and we've we've all kind of 0:54:18.800 --> 0:54:20.799 experienced this, saying you phone the airline and they say, 0:54:20.880 --> 0:54:22.319 tell us what you want to ask about, and there's 0:54:22.320 --> 0:54:25.200 only fifteen things you could ask so they can recognize 0:54:25.200 --> 0:54:27.080 it the text, and then they can work out what 0:54:27.120 --> 0:54:28.759 you're asking about and they can read you to the 0:54:28.840 --> 0:54:32.440 right number inside that they're pretty awful also, But what 0:54:32.560 --> 0:54:34.560 we don't have. Machine learning gives us a way of 0:54:34.600 --> 0:54:38.719 automating the transcription and getting the natural language processing to work. 0:54:39.239 --> 0:54:41.600 But you might ask it any one of three or 0:54:41.600 --> 0:54:45.600 four million things, and so effectively, what I mean when 0:54:45.640 --> 0:54:48.040 I say a structured query is you can fill in 0:54:48.040 --> 0:54:50.719 a dialog box by talking to the computer, and the 0:54:50.719 --> 0:54:53.920 computer can work out which dialog box you were asking for. 0:54:54.840 --> 0:54:58.440 But somebody has to have made the dialog box by hand. Okay, 0:54:58.520 --> 0:55:00.960 so you know if you're if you're a phone company, 0:55:01.080 --> 0:55:04.160 so if you're a phone manufacturer and you have a 0:55:04.200 --> 0:55:07.680 few million or maybe even a few billion queries, but 0:55:07.800 --> 0:55:10.839 how uses your frame of reference? But so you can 0:55:10.920 --> 0:55:15.160 make the top fifty categories, all the top thousand categories. Well, 0:55:15.400 --> 0:55:17.920 and if you're gonna ask something, really is the curve 0:55:18.000 --> 0:55:21.799 gets really really steep really quickly, and you don't know 0:55:21.840 --> 0:55:26.120 what you can ask. So you cannot ask. You can 0:55:26.160 --> 0:55:28.239 get the top five things, and you can say you 0:55:28.239 --> 0:55:30.640 can ask for weather, you can ask for timer, you 0:55:30.680 --> 0:55:32.960 can ask for a unit conversion, you can ask for 0:55:32.960 --> 0:55:35.239 the time in a different city. You can ask me 0:55:35.280 --> 0:55:37.160 to play music and you'll get the music right roughly 0:55:37.200 --> 0:55:38.919 half of the time. This is you know, you ask 0:55:38.960 --> 0:55:40.920 it to play an album and it plays you like 0:55:41.000 --> 0:55:44.520 a weird cover of it because it doesn't understand um. 0:55:44.560 --> 0:55:47.920 But then I was like, what are the thousand things 0:55:48.000 --> 0:55:51.399 you might ask your ear to do? And how many 0:55:51.400 --> 0:55:55.040 of those get really complex really quickly? Like can you 0:55:55.080 --> 0:55:59.000 rebook my meeting this afternoon? Okay, that's not a simple query, 0:55:59.280 --> 0:56:01.560 that's like five hundred things that you need to know 0:56:02.719 --> 0:56:05.239 that And so the problem is as I said. What 0:56:05.280 --> 0:56:07.120 you have with a voice assistant is you have an 0:56:07.120 --> 0:56:09.759 IVR that will always understand what you asked. Machine learning 0:56:09.800 --> 0:56:11.960 means it will always correctly root you to the right number. 0:56:12.480 --> 0:56:14.400 What it doesn't have is a way to have somebody 0:56:14.440 --> 0:56:17.600 at that number automated automate what that person at the 0:56:17.680 --> 0:56:20.200 number will do. And so you have to create those 0:56:20.200 --> 0:56:22.360 dialog boxes one by one. So give you an example. 0:56:22.880 --> 0:56:25.600 Sirie learns how to do cricket. You know it didn't 0:56:25.760 --> 0:56:28.840 somebody at Apple sat down and wrote the cricket module. Okay, 0:56:28.880 --> 0:56:32.960 now Sirie can do hurling. Okay, Well, somebody at Sirie 0:56:33.040 --> 0:56:35.279 at Apple had to sit down and write that, and 0:56:35.320 --> 0:56:38.799 then Siri can do call me a uber, right, Well, 0:56:38.840 --> 0:56:40.840 someone at Uber had to write that, and then it 0:56:40.880 --> 0:56:44.560 can do um tell me whether my flight is delayed? Right. 0:56:44.640 --> 0:56:47.440 Someone had to write that, And every single one of those, 0:56:47.719 --> 0:56:49.920 some human being has to sit down and spend an 0:56:49.960 --> 0:56:52.400 afternoon writing those. And the problem with this is it 0:56:52.440 --> 0:56:55.440 doesn't scale. And in a sense, if you could do that, 0:56:55.480 --> 0:56:58.000 you'd have made how nine thousand and how nine thousand 0:56:58.120 --> 0:57:00.359 is not the aggregate of like some of you sitting 0:57:00.400 --> 0:57:02.520 and writing those one at a time. That's that other 0:57:02.600 --> 0:57:05.160 thing that's machine learning, a big data and a bunch 0:57:05.160 --> 0:57:07.200 of things. So that's the gap. You know, we have 0:57:07.239 --> 0:57:09.440 a way to get the transcription and the NLP to 0:57:09.560 --> 0:57:11.480 work completely accurately, Well, we don't have. It is a 0:57:11.520 --> 0:57:14.479 way to get to automatically make a system that could 0:57:14.480 --> 0:57:18.280 answer any question you could possibly answer human being to 0:57:18.320 --> 0:57:20.480 think of what all of those would be. So that 0:57:20.680 --> 0:57:24.040 is a description of So let me answer answer this 0:57:24.480 --> 0:57:30.040 question another way. So imagine in the nineteenth century someone 0:57:30.040 --> 0:57:33.320 tries to make a mechanical horse. There's no I mean, 0:57:33.320 --> 0:57:35.840 you've seen the pattern drawing things. There's no law of 0:57:35.840 --> 0:57:38.640 physics that says you can't make a mechanical horse. It's 0:57:38.680 --> 0:57:41.200 just that the degree of complexity required to get that 0:57:41.240 --> 0:57:43.520 to work was impossible in the nineteenth century, and arguably 0:57:43.520 --> 0:57:45.480 it's impossible now. I mean, Boston Dynamics are now trying 0:57:45.520 --> 0:57:47.000 to do this pretty close. But that's like a hundred 0:57:47.000 --> 0:57:49.760 and fifty years later, um and so and so. What 0:57:49.800 --> 0:57:51.360 you can actually do is you could make a bicycle 0:57:51.400 --> 0:57:52.960 and the steam engine, and a bicycle is a lot 0:57:52.960 --> 0:57:54.720 simpler than a horse, and it can't do everything that 0:57:54.720 --> 0:57:55.920 a horse can do. But it turns out that was 0:57:55.960 --> 0:57:58.520 kind of useful anyway. The same thing with with AI. 0:57:58.640 --> 0:58:00.880 What people tried to do with AI until a couple 0:58:00.880 --> 0:58:02.760 of years ago was they would try and write rules. 0:58:03.520 --> 0:58:05.800 So if you wanted to recognize a cat, what you 0:58:05.880 --> 0:58:08.200 do is you do edge detection, and you do text 0:58:08.240 --> 0:58:10.479 your analysis, and you try and make something that looks 0:58:10.480 --> 0:58:12.600 for two eyes in roughly the right place relative to 0:58:12.640 --> 0:58:14.600 two ears, and you try and make something that tries 0:58:14.640 --> 0:58:16.960 to look for legs and hopefully, and you try and 0:58:17.000 --> 0:58:18.160 work out how the hell are we going to tell 0:58:18.160 --> 0:58:20.160 the difference between a cat and a dog? And it 0:58:20.200 --> 0:58:23.240 would sort of work, like three quarters at the time, 0:58:23.520 --> 0:58:25.400 it would never really work for the same reason that 0:58:25.440 --> 0:58:28.840 the mechanical horse would sort of work but never actually work. 0:58:29.320 --> 0:58:31.120 And then what machine learning does is say no, no, no no, 0:58:31.120 --> 0:58:33.840 no no. You've given a million pictures labeled cat and 0:58:33.840 --> 0:58:35.680 a million pictures of labeled dog, and you let the 0:58:35.720 --> 0:58:38.520 machine ripe the rules. That's what machine learning means. The 0:58:38.600 --> 0:58:41.760 machine generates a millionaire statements that will allow it to 0:58:41.800 --> 0:58:44.120 calculate this difference between a cat picture and a dog 0:58:44.160 --> 0:58:50.000 picture with ninety two point seven percent accuracy, and that 0:58:50.320 --> 0:58:52.840 gives us a breakthrough of a whole class of problem. 0:58:52.960 --> 0:58:56.880 It solves image recognition itselves, speech itselves, language, itselves, a 0:58:56.920 --> 0:58:59.920 whole kind of level of pattern recognition. When it doesn't 0:59:00.040 --> 0:59:02.920 do is give you a human ten year old. It 0:59:02.960 --> 0:59:06.200 doesn't give you general intelligence. Basically, what you've built is 0:59:06.240 --> 0:59:10.920 you've built an enormous spreadsheet with a million IF statements 0:59:11.080 --> 0:59:13.920 meshed together and linked together. And you can give this 0:59:14.000 --> 0:59:16.800 spreadsheet a picture of a cat and a picture, and 0:59:16.800 --> 0:59:18.280 it will tell you whether the cat in it or not, 0:59:18.600 --> 0:59:21.840 and that's all it will do. So another way of 0:59:21.880 --> 0:59:23.760 thinking about this is like, we have this fantasy of 0:59:23.800 --> 0:59:25.720 domestic robots that will walk around your home and do 0:59:25.760 --> 0:59:28.400 the housework. We have domestic robots. It's called a washing machine. 0:59:29.160 --> 0:59:31.400 That's what machine learning gives you. It gives your washing machine. 0:59:32.040 --> 0:59:35.120 It gives your washing machine that can recognize family pictures. 0:59:35.160 --> 0:59:37.280 It gives your washing machine that can tell me is 0:59:37.280 --> 0:59:39.600 there strange behavior happening on this network. It gives your 0:59:39.600 --> 0:59:42.240 washing machine that can recognize handwriting. It gives your washing 0:59:42.240 --> 0:59:44.520 machine that can do X. But it can only do 0:59:44.600 --> 0:59:46.800 that one thing. It can't do anything else, and it 0:59:46.840 --> 0:59:49.960 can't clean. You can't wash your dishes, And so that's 0:59:49.960 --> 0:59:52.200 what machine only gives it gives you this step change 0:59:52.200 --> 0:59:54.560 in capability. So one of the way I'm kind of 0:59:54.600 --> 0:59:56.280 I'm sort of giving you a stack of metaphors here. 0:59:56.280 --> 0:59:57.680 I think a really good way to think about what 0:59:57.720 --> 1:00:00.680 machine learning gives us is it's like thinking about relations databases. 1:00:01.400 --> 1:00:04.280 It's a relational databases gives you this amazing step change 1:00:04.280 --> 1:00:06.919 in what you can do with computers. Suddenly you can say, 1:00:06.960 --> 1:00:10.560 show me X by Y Bloomberg. Going would not exist 1:00:10.600 --> 1:00:13.720 without this, SAP would not exist without this, just in times, 1:00:13.720 --> 1:00:16.600 supply chains would not exist without this totally transformed our world. 1:00:17.200 --> 1:00:22.000 But everything you use now but it's not AI. It is. 1:00:22.040 --> 1:00:24.400 But it's like if it's AI, then everything is AI. 1:00:25.360 --> 1:00:27.560 It doesn't get you how nine thousand. Let me jump 1:00:27.600 --> 1:00:31.120 into some of my favorite questions. Tell us the most 1:00:31.200 --> 1:00:34.800 important thing that people who work with you don't know 1:00:34.880 --> 1:00:39.120 about you? I hate personal questions? Is that true? I 1:00:39.160 --> 1:00:40.720 don't know. I find I kind of look at this, 1:00:40.720 --> 1:00:42.680 this email, this, I didn't puzzled by these kind of questions. 1:00:42.680 --> 1:00:44.360 I don't know, what is it the people who work 1:00:44.400 --> 1:00:46.000 with that, we have no idea. They probably don't know 1:00:46.080 --> 1:00:49.160 have a dog. There you go, Well, because sometimes it's 1:00:49.200 --> 1:00:53.120 revealing some people have told us deep dark secrets from 1:00:53.320 --> 1:00:57.480 decades ago, and case there is a joke that the 1:00:57.480 --> 1:01:00.040 most secure form of encryption is anything you say in 1:01:00.080 --> 1:01:02.440 the second hour of a podcast. Here we are, I 1:01:02.480 --> 1:01:06.040 can say it now, and it's you, me, and a 1:01:06.080 --> 1:01:08.120 handful of friends that are hearing that nobody else is 1:01:08.160 --> 1:01:10.840 gonna hear. Um, tell us about your early mentors who 1:01:10.880 --> 1:01:14.960 affected the way you looked at technology and venture inversiting. 1:01:15.840 --> 1:01:18.560 I would say it's not really about how I look 1:01:18.600 --> 1:01:21.919 at technology. It's more sort of mental processes and ways 1:01:21.960 --> 1:01:24.840 of thinking about things and ways to try and answer questions. 1:01:25.400 --> 1:01:27.760 And so the stuff I did at university, you know, 1:01:27.800 --> 1:01:30.400 people talking to me about medieval history, and it's not 1:01:30.640 --> 1:01:32.919 this is no, it's not that didn't happen. This didn't happen. 1:01:33.360 --> 1:01:36.120 It's more, okay, how do you actually understand what we're 1:01:36.120 --> 1:01:38.400 trying to think about here? So, like I remember like 1:01:38.440 --> 1:01:41.160 my first or second, like you know, the first. The 1:01:41.200 --> 1:01:42.680 way it came Bridge history works is you do an 1:01:42.960 --> 1:01:44.480 You get given an essay title and a reading list, 1:01:44.520 --> 1:01:45.960 and you come back in a week with your essay 1:01:46.120 --> 1:01:49.600 and that's it. You don't have to get lectures. And 1:01:49.760 --> 1:01:51.440 so I think. My first essay was the one that, 1:01:51.480 --> 1:01:53.480 of course everyone screws up, and it was like was 1:01:53.560 --> 1:01:56.280 the King of England more or less powerful after ten 1:01:56.440 --> 1:01:59.560 exty six? And you know I gave the long answer 1:01:59.560 --> 1:02:01.360 and my professor said, look, you have to think about 1:02:01.400 --> 1:02:03.080 this is what does it actually mean to be powerful? 1:02:03.120 --> 1:02:04.760 And power is the ability to get people to do 1:02:04.840 --> 1:02:06.880 something they don't want to do. And so you can 1:02:06.880 --> 1:02:09.400 give all this other stuff about all the sophistication of 1:02:09.480 --> 1:02:11.800 the Anglo Saxon monarchy and they had their own much 1:02:11.800 --> 1:02:15.560 more sophisticated coinage and blah blah blah. No that was 1:02:15.600 --> 1:02:17.640 two years later, But like, who is more Is he 1:02:17.680 --> 1:02:20.120 actually more powerful? William the conqueror can tell people to 1:02:20.160 --> 1:02:21.320 do stuff and they have to do it. And the 1:02:21.400 --> 1:02:23.800 King of the sax the king before the conquest, King 1:02:23.840 --> 1:02:26.200 of England, like no one really paid any attention to 1:02:26.240 --> 1:02:28.440 what he said except for the coin like the complicated coinage. 1:02:28.440 --> 1:02:31.320 So my point is like it's not so much that 1:02:31.440 --> 1:02:34.360 someone told me how this is how to think about technology. 1:02:34.920 --> 1:02:38.120 It's more how to sort of think systematically and try 1:02:38.160 --> 1:02:40.600 and break up problems and try and look for peace 1:02:40.640 --> 1:02:43.200 of components and try and look kind of look at 1:02:43.240 --> 1:02:46.760 things in those kinds of ways quite personally. Tell us 1:02:46.760 --> 1:02:50.720 about some of your favorite books, be they fiction, nonfiction, technology, 1:02:50.840 --> 1:02:53.800 or what have you. Um, I don't know. I'm sort 1:02:53.840 --> 1:02:56.600 of I'm always reading and I'm always buying books, and 1:02:56.600 --> 1:02:59.240 those two process as are completely independent. I read almost 1:02:59.240 --> 1:03:02.479 no business books or technology books, partly because I feel 1:03:02.480 --> 1:03:04.800 that all business books are basically a short magazine article 1:03:04.840 --> 1:03:08.880 padded out to There are some authors in particular I 1:03:08.880 --> 1:03:11.880 can name, well, that's their entire over. Well, if you 1:03:11.960 --> 1:03:14.000 feel like if you've written fifteen business books on the 1:03:14.040 --> 1:03:16.160 same theme, then like you're either a very bad writer 1:03:16.280 --> 1:03:18.600 or very good writer, depending on your very good marketer, 1:03:18.920 --> 1:03:22.000 Well that too. Perhaps, I know. I think I read 1:03:22.040 --> 1:03:24.360 things that intrigued me and interest me. The last thing 1:03:24.400 --> 1:03:27.880 I read with the Iliad, which I've never read before. Wait, 1:03:27.920 --> 1:03:31.120 the history major never read the Iliad. Well, so there's 1:03:31.120 --> 1:03:33.200 this great line from Italia Calvino where he says the 1:03:33.280 --> 1:03:35.280 number of something like the number of essential works is 1:03:35.320 --> 1:03:37.240 so great that nobody, no matter how much they're reading, 1:03:37.280 --> 1:03:40.280 has read more than a fraction. Um, you know, literature 1:03:40.280 --> 1:03:42.680 as a sample, you can't read all the books. There 1:03:42.720 --> 1:03:44.160 was a period when you could have seen all the movies. 1:03:44.160 --> 1:03:45.560 Now you can't see all the movies, you can't listen 1:03:45.600 --> 1:03:47.960 to all the music. You just have to take samples 1:03:48.040 --> 1:03:50.280 and take do things that stimulate you or make you 1:03:50.320 --> 1:03:53.200 think in interesting ways, particularly things that aren't about what 1:03:53.240 --> 1:03:57.000 you work on. So give us another example from that's 1:03:57.160 --> 1:04:01.000 that's newer than two thousand years old. Um, I don't know. 1:04:01.800 --> 1:04:04.320 By the way, this is everybody's favorite question, the one 1:04:04.600 --> 1:04:07.840 I get. I read books and then I can't remember 1:04:07.880 --> 1:04:09.400 what it is? What? What? What? The last book I 1:04:09.480 --> 1:04:11.920 read was, Um, what stands out to you? What has 1:04:12.000 --> 1:04:14.520 really read? Really? The Iliad really stood out for me. 1:04:14.600 --> 1:04:15.920 That was a kind of a couple of weeks ago, 1:04:15.960 --> 1:04:20.280 I read this fascinating book about medieval manuscripts called Encounters 1:04:20.360 --> 1:04:23.600 with Fascinating Manuscripts or something like that by college librarian 1:04:23.600 --> 1:04:25.880 in Cambridge, and it's one of these sort of new 1:04:26.040 --> 1:04:27.920 kind of genre of books which kind of looked like 1:04:27.920 --> 1:04:30.240 a big hardback nonfiction book. But they're all in color 1:04:30.280 --> 1:04:31.720 all the way through, which is funny to do with 1:04:31.760 --> 1:04:34.240 the development of printing technology, which is there's there's loads 1:04:34.240 --> 1:04:36.080 of books that have color illustrations all the way through, 1:04:36.080 --> 1:04:37.960 which is but aren't art books. It's kind of interesting 1:04:38.120 --> 1:04:40.000 and it's just sort of about how books evolved. So 1:04:40.000 --> 1:04:44.360 it's kind of fascinating observation that books or codeses used 1:04:44.400 --> 1:04:46.800 to be rectangular because you made them out of papyrus. 1:04:46.800 --> 1:04:48.800 So you would lay your sheets, you know, you lay 1:04:48.840 --> 1:04:52.160 the strips of plant kind of cross weights like plywood, 1:04:52.200 --> 1:04:53.920 and you'd make them square shaped because you had the 1:04:54.520 --> 1:04:56.080 strips of all the same length, So that produced to 1:04:56.120 --> 1:05:00.720 square shaped piece of papyrus. And then you moved to um, 1:05:00.720 --> 1:05:03.640 what's it called animal skin? What's the word? I've forgotten 1:05:03.640 --> 1:05:06.439 the word? My mind is gone blank now too, um Um. 1:05:06.480 --> 1:05:10.240 You moved to parchment from animal skins. And there's this 1:05:10.280 --> 1:05:13.840 great line. Animals tend to be oblong, so you get 1:05:13.880 --> 1:05:16.560 oblong pieces and then you fold them up and that 1:05:16.600 --> 1:05:19.080 produces an oblong book. And so when you go from 1:05:19.120 --> 1:05:22.040 papyrus department, you go from square books to oblong books. 1:05:22.400 --> 1:05:24.400 And here we are years later and the books are 1:05:24.400 --> 1:05:27.800 still oblong, and that's just the residual of the earlier 1:05:27.840 --> 1:05:31.440 technology of parchment. Yeah, exactly. I think those sorts of 1:05:31.520 --> 1:05:33.600 I mean, coming back to you know, why is this interesting? 1:05:33.600 --> 1:05:36.720 I do think those sorts of how did stuff happen, 1:05:36.840 --> 1:05:39.160 and why did the what's the thread of causation and 1:05:39.200 --> 1:05:42.280 how things can be completely random is sort of interesting 1:05:42.280 --> 1:05:44.680 and how do these does this stuff change how we 1:05:44.720 --> 1:05:46.600 think about it? So on the flight here, I've read 1:05:46.640 --> 1:05:48.920 a book about kind of lighting tech, the evolution of 1:05:48.960 --> 1:05:51.240 lighting technology in the eighteenth and nineteent century, as we 1:05:51.280 --> 1:05:53.360 go from candles to light to kind of lamps that 1:05:53.440 --> 1:05:55.360 hold the candle and produce five times more light, how 1:05:55.440 --> 1:05:58.160 we go to now or the gas lights to to 1:05:58.560 --> 1:06:01.520 um earlier electric light which used weren't very bright, and 1:06:01.520 --> 1:06:03.920 then use tungsten filament and again, so there's the evelation 1:06:03.960 --> 1:06:06.400 of light. This observation that stuck in my mind and 1:06:06.400 --> 1:06:08.200 I and now need to go and check which is 1:06:08.240 --> 1:06:12.640 an assertion that Dutch seventy century Dutch pictures have windows 1:06:12.680 --> 1:06:16.240 but no cuts, because why would you keep the light out. 1:06:16.400 --> 1:06:18.040 Everybody wants to light and you want the light in. 1:06:18.160 --> 1:06:19.600 And of course, as we all sort of no, light 1:06:19.680 --> 1:06:22.320 was really expensive and used to be a luxury and 1:06:22.320 --> 1:06:24.240 now it's not. But I just thought that was a 1:06:24.280 --> 1:06:27.440 really revealing way of thinking about how your whole sense 1:06:27.480 --> 1:06:30.200 of the world changes when lighting is cheap. Why would 1:06:30.240 --> 1:06:33.560 you have curtns that that's quite fastening. The book How 1:06:33.600 --> 1:06:36.720 We Got to Now has there's six inventions, one of 1:06:36.760 --> 1:06:40.240 which is lighting, and the process of watching the price 1:06:40.400 --> 1:06:45.320 plummet over over the centuries is really is really fastening, 1:06:45.640 --> 1:06:47.600 all right. The interesting thing that comes out of that 1:06:47.720 --> 1:06:50.600 is that when light was expensive, state of symbol was 1:06:50.680 --> 1:06:53.320 to stay up all night. So rich people would get 1:06:53.360 --> 1:06:55.280 up at lunchtime and go to bed at l five 1:06:55.280 --> 1:06:57.200 in the morning because they could afford to, because they 1:06:57.200 --> 1:07:01.080 could afford candles. Everybody else gets up with the croaker, think, um, 1:07:01.120 --> 1:07:02.280 what do you do for fun? What do you do 1:07:02.360 --> 1:07:06.760 to relax outside of the office in San Francisco? Um? 1:07:06.880 --> 1:07:10.240 So I have a seven year old and a dog, 1:07:10.760 --> 1:07:15.000 so my time is not my own understood completely? What 1:07:15.120 --> 1:07:17.560 sort of advice would you give to a millennial or 1:07:17.640 --> 1:07:21.920 someone who just graduated college that was interested either in 1:07:21.920 --> 1:07:26.640 a career in technology or a venture capital So I 1:07:27.240 --> 1:07:29.200 sort of struggle to give an advice party because my 1:07:29.280 --> 1:07:32.400 career has been a series of random lurches and kind 1:07:32.400 --> 1:07:36.120 of companies going out of business. I won't apply for 1:07:36.120 --> 1:07:38.880 a job at cling Back. Don't worry, I think there's 1:07:38.920 --> 1:07:42.680 a The only thing I've sort of observed is there is? Um, 1:07:42.720 --> 1:07:44.919 there is the kind of things that you are good 1:07:44.960 --> 1:07:47.320 at doing and the kind of mental processes that you 1:07:47.360 --> 1:07:49.880 are good at. Are you good at people or not? 1:07:50.000 --> 1:07:53.040 Are you good at process and project management and making 1:07:53.040 --> 1:07:55.520 sure everything gets done right? Or is that not what 1:07:55.600 --> 1:07:58.320 it interests you? Are you good at trying to kind 1:07:58.360 --> 1:08:01.240 of explain an empathy sizing with people? Are you good 1:08:01.240 --> 1:08:04.480 at trying to explain and discuss and argue with people 1:08:04.600 --> 1:08:08.160 and persuade people? Um? What are the things that you're 1:08:08.200 --> 1:08:10.400 good at doing? And those are not necessarily what was 1:08:10.440 --> 1:08:14.080 your degree in, and they aren't necessarily specific to two jobs. 1:08:14.200 --> 1:08:17.000 So like I have contemporaries who went off and become 1:08:17.200 --> 1:08:19.519 what England we call a barrister, which is basically a 1:08:19.600 --> 1:08:21.760 lawyer who only goes to court is in England and 1:08:21.800 --> 1:08:23.639 law is split solicitors to be the paper what barristers 1:08:23.680 --> 1:08:25.960 get called as well. Exactly that's all, but that's all 1:08:25.960 --> 1:08:28.880 the effectively and it's a completely different profession. Um, And 1:08:28.920 --> 1:08:31.360 I have contemporaries who are barristers, and you know I 1:08:31.360 --> 1:08:33.360 could have done that. I could have gone and argued 1:08:33.360 --> 1:08:34.559 for a living, and you know you kind of you 1:08:34.600 --> 1:08:35.720 go and do you look at the problem and you 1:08:35.760 --> 1:08:37.240 pull it apart and you work out what story to 1:08:37.280 --> 1:08:39.160 tell and how to place added to the law and 1:08:39.160 --> 1:08:41.920 try and persuade the judge and the jury. Um So, 1:08:42.080 --> 1:08:44.240 in a sense, I have those skill sets. On the 1:08:44.280 --> 1:08:46.840 other hand, I look at these contemporaries and it says, 1:08:46.960 --> 1:08:51.080 you know, um, Charlie is probably the leading up and 1:08:51.120 --> 1:08:54.479 coming advocate in complex of your maritime insurance disputes. And 1:08:54.520 --> 1:08:57.840 I think, well, I hope he's paid a lot to that, 1:08:59.000 --> 1:09:02.040 but his interests did in the argument and the explaining 1:09:02.040 --> 1:09:05.720 and the discussion. The fact that it's insurance is kind 1:09:05.720 --> 1:09:08.439 of neither here nor there really, um So, I think 1:09:08.439 --> 1:09:10.080 you have to just kind of work out, well, what 1:09:10.280 --> 1:09:12.720 is it What are the things that fascinate you, and 1:09:12.800 --> 1:09:15.720 what are the kind of mental processes that fascinate you, 1:09:16.160 --> 1:09:18.479 And and our final question, what is it that you 1:09:18.520 --> 1:09:21.240 know about the worlds of technology that you wish you 1:09:21.320 --> 1:09:26.320 knew twenty years ago? Other than like buy Apple or something, No, no, 1:09:26.439 --> 1:09:29.360 nothing like that. That that's you know, anyone could look 1:09:29.400 --> 1:09:32.360 at prices twenty years ago and said hey, by Amazon, 1:09:32.560 --> 1:09:36.639 by Apple. But generally in terms of thinking in terms 1:09:36.680 --> 1:09:39.560 of broad processes, what would have helped you early in 1:09:39.640 --> 1:09:41.760 your career had you figured it out sooner. So that 1:09:41.920 --> 1:09:46.160 actually might be a good way of me framing my 1:09:46.400 --> 1:09:50.280 technology to determinism peace because it's a sort of I mean, 1:09:50.479 --> 1:09:52.519 the other thing people talk about about a lot in 1:09:52.600 --> 1:09:55.800 venture capitalist patent recognition that this sort of looks like 1:09:55.880 --> 1:09:58.280 things that will work. This looks like things that that 1:09:58.360 --> 1:10:01.000 never work and why and the thing that never work 1:10:01.439 --> 1:10:03.320 eventually they do work, and that's where there's a hundred 1:10:03.320 --> 1:10:05.519 billion dollar outcome. So that's also an interesting kind of 1:10:05.520 --> 1:10:09.320 fruitful copy conversation. But it's sort of understanding these patents 1:10:09.479 --> 1:10:11.680 of Okay, you have to think, okay, how is this 1:10:11.720 --> 1:10:15.040 going to fit within this ecosystem? This is what is 1:10:15.040 --> 1:10:18.000 the timing for this? It's those sorts of mechanics of 1:10:18.080 --> 1:10:23.200 the process um that are interesting. I think that's quite fascinating. 1:10:23.840 --> 1:10:26.760 We have been speaking with Ben Nick Evans of Andres 1:10:26.760 --> 1:10:29.800 and Horowitz. If you enjoy this conversation, be sure and 1:10:29.840 --> 1:10:31.559 look up an Inch or down an inch on Apple, 1:10:31.640 --> 1:10:37.000 iTunes or SoundCloud, overcast, Bloomberg dot com wherever your final 1:10:37.080 --> 1:10:39.400 podcasts are sold, and you could see any of the 1:10:39.439 --> 1:10:42.479 other two hundred such conversations we've had over the past 1:10:43.040 --> 1:10:46.920 four years. We love your comments, feedback and suggestions right 1:10:47.000 --> 1:10:49.920 to us at m IB podcast at Bloomberg dot net. 1:10:50.439 --> 1:10:52.320 I would be remiss if I did not thank the 1:10:52.360 --> 1:10:56.240 crack staff that helps us put this podcast together each week. 1:10:56.560 --> 1:10:59.599 Taylor Riggs is our book or slash producer. Medina Parwana 1:11:00.080 --> 1:11:03.640 is our audio engineer slash producer. Mike Batnik is our 1:11:03.680 --> 1:11:07.599 head of research. I'm Barry Retolts. You're listening to Master's 1:11:07.640 --> 1:11:09.679 in Business on Bloomberg Radio.