WEBVTT - The Story: An Innovation Playground w/ Astro Teller 0:00:01.480 --> 0:00:03.760 Welcome to tech Stuff. This is the story. 0:00:04.519 --> 0:00:06.880 Each week on Wednesdays, we bring you an in depth 0:00:06.920 --> 0:00:09.240 interview with someone who has a front row seat to 0:00:09.240 --> 0:00:12.960 the most fascinating things happening in tech today. We're joined 0:00:12.960 --> 0:00:22.119 by Astroteller. Astroteller is the captain of Moonshots for X, 0:00:22.400 --> 0:00:28.320 an innovation lab within Alphabet, Google's parent company. To understand Astroteller, 0:00:28.760 --> 0:00:32.760 you have to understand moonshots. Years ago, a moonshot might 0:00:32.800 --> 0:00:35.640 have been slang for a long shot, but in the 0:00:35.640 --> 0:00:39.600 technological age, it's more akin to a lofty goal or 0:00:39.880 --> 0:00:42.960 a giant leap. So to be the captain of Moonshots, 0:00:43.280 --> 0:00:44.479 you sort of have to be. 0:00:44.440 --> 0:00:47.240 The ringleader for pioneers. Now. 0:00:47.440 --> 0:00:49.920 If you've never heard of X, you've probably heard of 0:00:49.920 --> 0:00:52.199 the products that came out of what's known as the 0:00:52.240 --> 0:00:56.400 Moonshot Factory. Google Glass a leap forward for smart glasses 0:00:56.440 --> 0:01:01.160 and computer vision, Weai mow actual self driving tech, and 0:01:01.400 --> 0:01:06.080 Google Brain, a groundbreaking AI research team. All of them 0:01:06.120 --> 0:01:08.840 got their start at X. But X is more than 0:01:08.840 --> 0:01:13.080 an incubator, it's a playground. And thanks to the Moonshop podcast, 0:01:13.280 --> 0:01:15.959 we finally get a bird's eye view of the factory floor. 0:01:17.240 --> 0:01:20.640 Host Astroteller leads us through what feels like an oral 0:01:20.800 --> 0:01:24.440 history about how these innovations came to be from the 0:01:24.480 --> 0:01:27.240 perspective of the people who built them, the problems that 0:01:27.280 --> 0:01:31.200 sparked the idea, the setbacks, the successes, and how each 0:01:31.240 --> 0:01:35.319 project evolved. I had the opportunity to interview Astroteller about 0:01:35.360 --> 0:01:39.160 the Moonshop podcast at south By Southwest. We met at 0:01:39.160 --> 0:01:42.319 the Google office in Austin, Texas, and the first thing 0:01:42.360 --> 0:01:45.600 I wanted to know was what motivated Alphabet to produce 0:01:45.640 --> 0:01:46.319 this podcast. 0:01:47.000 --> 0:01:49.800 Our original excuse for making it was that we're turning 0:01:49.840 --> 0:01:53.080 fifteen years old and just seemed like a nice time 0:01:53.240 --> 0:01:57.600 to stop and look backwards and think about sort of 0:01:57.760 --> 0:02:02.280 where have we come? Just for reminiscence, but also as 0:02:02.320 --> 0:02:05.360 a way of educating ourselves and maybe sharing with other people, 0:02:05.800 --> 0:02:09.640 Like what is a moonshot factory? Does the world only 0:02:09.680 --> 0:02:12.240 need one? Does it need zero? Does it actually need 0:02:12.280 --> 0:02:14.600 a hundred of them? I hope it's the answer, is 0:02:14.600 --> 0:02:19.760 the last, And what can we learn from all of 0:02:19.760 --> 0:02:24.040 these people who've gone through this process. Let me give 0:02:24.080 --> 0:02:27.080 you a really concrete example, because if I stand on 0:02:27.160 --> 0:02:29.800 stage and I say one of the mantras at X 0:02:30.080 --> 0:02:33.880 is get into the real world as fast as possible, 0:02:33.919 --> 0:02:37.760 get in contact with it, and get this sort of painful, complex, 0:02:37.960 --> 0:02:42.160 dirty learning lessons from getting into the world and then 0:02:42.360 --> 0:02:45.360 realizing in all kinds of ways you're wrong. I can 0:02:45.400 --> 0:02:47.919 talk about why that's a good idea. It is sincerely 0:02:47.960 --> 0:02:49.720 one of the mantras and sort of the ways that 0:02:49.760 --> 0:02:55.120 we operated X that's very abstract, it's very philosophical. On 0:02:55.200 --> 0:02:58.280 the first podcast, there's a nice moment where the Wing 0:02:58.400 --> 0:03:01.480 team is talking about the fact that when they got 0:03:01.520 --> 0:03:04.280 out in the world first really doing deliveries, which was 0:03:04.360 --> 0:03:07.240 first in Australia, they were worried that people be annoyed 0:03:07.320 --> 0:03:09.799 by the sound of their drones, and so they had 0:03:09.800 --> 0:03:12.760 worked really hard for the propellers that are operating when 0:03:12.760 --> 0:03:15.000 it's in hover modes, so it's just hovering above a 0:03:15.080 --> 0:03:17.640 house and it's lowering a package on a string for 0:03:17.680 --> 0:03:20.840 a delivery. They were worried that that sound is the 0:03:20.880 --> 0:03:22.720 sound that would bother people. So they've done all this 0:03:22.800 --> 0:03:25.799 work ahead of time to make that as quiet as possible, 0:03:25.960 --> 0:03:27.760 and when I got out there, they found out no 0:03:27.800 --> 0:03:28.800 one cared about. 0:03:28.520 --> 0:03:30.880 That sound because they were excited to be receiving the packages. 0:03:30.960 --> 0:03:35.240 Right exactly, and it was the forward flight this zsh 0:03:35.440 --> 0:03:38.560 right as something was going over houses, you know, sixty 0:03:38.640 --> 0:03:40.680 seventy miles an hour, two hundred feet in the air, 0:03:40.760 --> 0:03:43.920 which we had never considered would be the problem. That 0:03:44.080 --> 0:03:47.440 was actually the sound that people were bothered by. And 0:03:47.480 --> 0:03:49.240 then we went and we did the hard work of 0:03:49.280 --> 0:03:51.360 getting that to be much quieter. And now that doesn't 0:03:51.360 --> 0:03:54.680 bother people either. But hearing the chief engineer for Wing 0:03:54.840 --> 0:03:58.280 unpack the discovery of that and then how they worked 0:03:58.320 --> 0:04:03.240 on it makes this this very abstract idea let's say, 0:04:03.240 --> 0:04:05.480 get in contact with the real world all of a sudden, 0:04:05.640 --> 0:04:07.120 really concrete for people. 0:04:07.480 --> 0:04:09.800 So they say that, you know, a great podcast always 0:04:09.840 --> 0:04:13.440 drops you into a scene, and your podcast drops you 0:04:13.560 --> 0:04:16.120 into quite a fascinating scene as a listener with you 0:04:16.360 --> 0:04:20.039 and Sebastian Thrun, who originally found at Google x and 0:04:20.160 --> 0:04:22.440 hired you in the back of a way, mo I 0:04:22.440 --> 0:04:24.520 mean a project that you know, you guys had worked 0:04:24.520 --> 0:04:26.840 on that really came to life. But I'm wondering if 0:04:26.839 --> 0:04:30.520 you can drop our listeners into a scene, describe what 0:04:30.560 --> 0:04:33.320 it's like to be in the Moonshot factory. I'm thinking 0:04:33.320 --> 0:04:35.120 about roller skates. I'm thinking about the line that you 0:04:35.200 --> 0:04:36.080 cross every morning. 0:04:36.520 --> 0:04:40.440 Sure. So the building itself originally was the first air 0:04:40.480 --> 0:04:44.799 conditioned mall in California. So it's a relatively old mall, 0:04:45.200 --> 0:04:50.320 very high ceilings. We've left it quite raw, polished cement floors, 0:04:50.680 --> 0:04:54.520 the concrete beams on the sides, like this one that's 0:04:54.600 --> 0:04:56.680 right here where we're recording. This has a lot of 0:04:56.680 --> 0:04:59.240 graffiti on it. Not this concrete one because people have 0:04:59.279 --> 0:05:03.800 cleaned this one up, but we've left the original construction 0:05:04.040 --> 0:05:07.200 markings you know, gas line that way, don't drill or 0:05:07.200 --> 0:05:11.719 whatever on the concrete. It's not a decoration for us. 0:05:11.800 --> 0:05:14.000 But it was like a why would we spend time 0:05:14.040 --> 0:05:16.800 cleaning it up? Is that really the money we want 0:05:16.839 --> 0:05:19.800 to spend? B We want the place, and I think 0:05:19.839 --> 0:05:23.360 it successfully feels like a work in progress. And if 0:05:23.440 --> 0:05:27.200 we're as humans as professionals and our projects are constantly 0:05:27.240 --> 0:05:30.320 a work in progress, how can we send a lot 0:05:30.360 --> 0:05:35.000 of unconscious signals to encourage that work in progress mentality? 0:05:35.520 --> 0:05:37.920 And so the walls are a lot of them made 0:05:37.920 --> 0:05:41.720 of plywood. Could we have used something other than plywood? Yes, 0:05:42.160 --> 0:05:45.680 but plywood works fine, and it sends that same signal 0:05:46.000 --> 0:05:50.240 when you come into the main lobby. We have a 0:05:50.279 --> 0:05:54.080 lot of things hanging there on the walls or large robots. 0:05:54.120 --> 0:05:56.040 I guess they're sitting on the floor in some cases, 0:05:56.520 --> 0:06:00.839 but pretty universally, they're not the finished product that we made. 0:06:01.360 --> 0:06:03.960 There's something that we made along the way, and that's 0:06:04.000 --> 0:06:08.360 another opportunity for us to signal to each other and 0:06:08.400 --> 0:06:11.160 to everyone who visits us. We're more proud of the 0:06:11.240 --> 0:06:16.720 process than we are of the outcome, including is you're referencing. 0:06:16.839 --> 0:06:20.080 There's this huge line across the floor and it says, 0:06:20.160 --> 0:06:23.280 in all caps, right by the line, you may never 0:06:23.400 --> 0:06:28.560 cross this line. Exclamation point ever exclamation point, And it's 0:06:28.600 --> 0:06:31.000 a stupid rule. You can't get into the building unless 0:06:31.000 --> 0:06:34.600 you cross the line, and it's a way to help 0:06:34.720 --> 0:06:39.320 people practice breaking stupid rules. We don't want people to 0:06:39.440 --> 0:06:45.599 like cause fraud or embezzle money or hurt somebody, but 0:06:45.839 --> 0:06:49.400 so many rules are actually in our heads. There are 0:06:49.400 --> 0:06:53.360 assumptions about how the world works that we haven't realized 0:06:53.600 --> 0:06:56.599 or just assumptions, and they can and need to be questioned. 0:06:56.600 --> 0:06:58.880 If you're going to do something unusual. 0:06:59.040 --> 0:07:01.280 Can you describe what it feels like to work at 0:07:01.279 --> 0:07:02.320 the Moonshot Factory? 0:07:03.080 --> 0:07:10.200 There is a happy, mild mannickness to being around us 0:07:10.320 --> 0:07:15.600 every day. The Moonshot Factory is a very matrix place where, 0:07:16.240 --> 0:07:18.960 in some great ways and maybe in some not ideal ways, 0:07:18.960 --> 0:07:21.280 we're sort of all in each other's business all the time. 0:07:21.720 --> 0:07:24.200 There's a lot of hey, can I help with that. 0:07:24.280 --> 0:07:26.320 In fact, just as someone was dropping me off in 0:07:26.360 --> 0:07:29.320 this room, we were ending a conversation about ways we 0:07:29.360 --> 0:07:32.560 could be even better at not worrying about whose job 0:07:32.680 --> 0:07:34.680 was what, but we could just like jump in and 0:07:34.760 --> 0:07:37.160 help not only things that are still at X, but 0:07:37.200 --> 0:07:39.840 we were actually talking about something that has left X, 0:07:40.880 --> 0:07:43.120 and people at X were saying, oh, do we still 0:07:43.160 --> 0:07:44.840 need to be helping them? He was like, who cares? 0:07:45.080 --> 0:07:47.360 Like they're part of us? They were a graduate of ours, 0:07:47.520 --> 0:07:49.880 just like, jump in and help. That's a lot of 0:07:49.880 --> 0:07:52.880 what it feels like in the conversations in the hallways 0:07:52.920 --> 0:07:58.440 every day. There's an ethos of helpfulness, of excitement, a 0:07:58.520 --> 0:08:05.320 sense of purpose. The philosophy of experimentation plays out all 0:08:05.360 --> 0:08:08.960 the time. Every day. There's very much a why don't 0:08:08.960 --> 0:08:11.960 we just try it, like instead of talking about it 0:08:12.040 --> 0:08:14.040 when we don't really know what the right answer is. 0:08:14.280 --> 0:08:17.960 And this might be about some hr issue or about 0:08:18.040 --> 0:08:21.760 some public relations issue. It might be a very technical issue. 0:08:21.880 --> 0:08:24.240 It's actually the same thing. It's do we really know 0:08:24.280 --> 0:08:27.440 what the right answer is. Let's make a hypothesis and 0:08:27.480 --> 0:08:31.440 then find the fastest, simplest, cheapest way to test that hypothesis. 0:08:31.640 --> 0:08:34.080 That's kind of how it feels in all the conversations 0:08:34.120 --> 0:08:34.600 all the time. 0:08:34.760 --> 0:08:37.400 So I think the process is fascinating, but the output 0:08:37.440 --> 0:08:39.920 is also fascinating, right, Like I've also been in a 0:08:39.960 --> 0:08:42.680 Way Mow, and you know you have that experience like 0:08:42.720 --> 0:08:46.079 this is totally uncanny, and then it becomes normal very quickly. 0:08:46.600 --> 0:08:51.160 Google Brain obviously also spun out of X and also 0:08:51.280 --> 0:08:55.120 gave birth, in partnership with deep Mind, to last year's 0:08:55.320 --> 0:08:59.200 Nobel Prize in chemistry. What marked out, let's say, Way 0:08:59.320 --> 0:09:04.600 Mow and Brain as huge successes versus the projects that 0:09:04.840 --> 0:09:05.600 didn't graduate. 0:09:05.840 --> 0:09:08.400 What characterizes a success in a failure? 0:09:08.920 --> 0:09:11.960 At least the way it feels at X is we 0:09:12.040 --> 0:09:15.480 start on a whole bunch of unlikely journeys. But I 0:09:15.480 --> 0:09:17.880 don't know that we're much better than random at predicting 0:09:17.920 --> 0:09:19.959 ahead of time, what's going to be a great idea? 0:09:20.040 --> 0:09:22.960 If we were great at predicting what's going to be 0:09:22.960 --> 0:09:25.200 a great idea. We would only work on the great ideas. 0:09:25.880 --> 0:09:28.960 I don't think anybody gets that privilege. There are people 0:09:29.160 --> 0:09:32.200 who pick one thing. They announced that they have the 0:09:32.280 --> 0:09:34.880 right idea, and then they work really hard on it 0:09:34.760 --> 0:09:36.800 and it turns out they did have a great idea. 0:09:37.720 --> 0:09:41.040 That's just survivor bias thought. Those are just the ones 0:09:41.040 --> 0:09:43.280 you hear. There are a thousand people who said they 0:09:43.320 --> 0:09:45.559 had a great idea and worked really hard on it, 0:09:45.760 --> 0:09:48.560 and just they didn't have a great idea it turns out, 0:09:48.760 --> 0:09:51.440 and so they went away. The one person who made 0:09:51.440 --> 0:09:54.959 it is not necessarily smarter. Mostly I believe we believe 0:09:55.000 --> 0:09:58.280 at X they just got lucky. So we start about 0:09:58.280 --> 0:10:01.760 a thousand things per decade, and then we're always looking 0:10:01.800 --> 0:10:05.080 for evidence. Is this really a once in a generation 0:10:05.200 --> 0:10:07.760 opportunity for the world to make the world better and 0:10:07.800 --> 0:10:11.480 to create an enduring business. And when the evidence starts 0:10:11.520 --> 0:10:14.600 to pile up, no, or at least we can't get 0:10:14.640 --> 0:10:17.920 good evidence that the answer is yes, we throw it away. 0:10:18.400 --> 0:10:22.880 So brain and way Mow survived those pressure tests, that's 0:10:22.920 --> 0:10:26.640 really what happened. It wasn't this was the good stuff 0:10:27.000 --> 0:10:30.200 ahead of time. It's more like the evidence piled up 0:10:30.240 --> 0:10:32.960 that it turns out these were particularly good ideas, And 0:10:33.040 --> 0:10:35.640 you know, maybe there were things that we've tried that 0:10:35.720 --> 0:10:38.920 were really good ideas, but we just didn't find the 0:10:39.080 --> 0:10:43.200 right wedge in on the problem. So maybe somebody else will. 0:10:43.520 --> 0:10:46.000 And in a number of cases we've actually when we 0:10:46.040 --> 0:10:49.520 wound something down, we've published to the rest of the 0:10:49.559 --> 0:10:52.120 world everything that we learned so that they could build 0:10:52.160 --> 0:10:52.800 on top of that. 0:11:00.360 --> 0:11:01.200 When we come back. 0:11:01.400 --> 0:11:04.640 How an innovation playground birthed in the era of endless 0:11:04.640 --> 0:11:17.480 optimism maintains its momentum. Stay with us, Welcome back to 0:11:17.520 --> 0:11:20.920 tech Stuff. I'm talking to Astroteller about the new podcast 0:11:21.080 --> 0:11:25.520 out of Alphabet's ex called the Moonshot Podcast. There's an 0:11:25.559 --> 0:11:28.719 interesting moment in the podcast where Sebastian Thrunn, the co 0:11:28.760 --> 0:11:32.480 founder of x, describes a conversation he had with former 0:11:32.520 --> 0:11:35.720 Google CEO Larry Page back in two thousand and five. 0:11:36.840 --> 0:11:39.400 Larry asked Thron to come to Google and make a 0:11:39.440 --> 0:11:43.280 self driving car essentially out of nowhere, and Thron, though 0:11:43.320 --> 0:11:46.360 skeptical of the outcome, took a leap of faith and 0:11:46.400 --> 0:11:52.000 started experimenting, creating what would ultimately become the Moonshop Factory. 0:11:52.080 --> 0:11:53.280 This was during the era. 0:11:53.160 --> 0:11:56.800 Of endless optimism in Silicon Valley, but fifteen years later 0:11:57.240 --> 0:11:59.400 there seems to have been something of a vibe shift 0:12:00.160 --> 0:12:03.480 cuts at old big tech companies, including cuts that have 0:12:03.520 --> 0:12:07.720 affected Eggs, And I wanted to know if Astarteta thought 0:12:07.760 --> 0:12:09.560 this vibe shift affects X. 0:12:09.559 --> 0:12:14.240 Today, we're constantly learning and trying to improve, but I 0:12:14.320 --> 0:12:17.679 hope that that vibe shift hasn't per se changed us 0:12:18.840 --> 0:12:21.520 at all. Let me describe that in another way. From 0:12:21.600 --> 0:12:24.920 the very beginning, we've been trying to find ways to 0:12:25.040 --> 0:12:30.560 make a moonshot factory. That is, keep our audacity really high, 0:12:30.840 --> 0:12:34.960 but find ways to systematize the process. So we've been 0:12:34.960 --> 0:12:37.600 committed to that for fifteen years. And as soon as 0:12:37.640 --> 0:12:40.000 you commit to the factory part of it, not just 0:12:40.040 --> 0:12:43.080 the moonshot part of it, you are pre committing to 0:12:43.200 --> 0:12:47.680 a constant attempt to up the rigor without killing off 0:12:47.720 --> 0:12:51.840 the magic. And so we didn't need a vibe shift 0:12:51.840 --> 0:12:56.040 to get interested in efficiency. We've always been interested in 0:12:56.160 --> 0:12:59.320 how to keep ratcheting up that efficiency. That's not a 0:12:59.360 --> 0:13:02.319 new thing for us. And so that vibe shift of 0:13:02.400 --> 0:13:06.280 anything has aligned the world better with what we were 0:13:06.320 --> 0:13:09.200 already trying to do, which is, we want to be 0:13:09.360 --> 0:13:12.560 creating a great return on investment where the things that 0:13:12.559 --> 0:13:16.560 we produce are worth more than enough to justify the 0:13:16.559 --> 0:13:18.480 money we've spent and the time that it took to 0:13:18.559 --> 0:13:22.839 make ten or fifteen year period. Now it happens that 0:13:22.960 --> 0:13:25.960 Alphabet is very long term in its thinking, which it 0:13:26.000 --> 0:13:29.200 does to its great long term benefits. And you know, 0:13:29.440 --> 0:13:32.480 it has become a very large business by thinking long 0:13:32.559 --> 0:13:36.640 term and by having the bravery to place these much 0:13:36.640 --> 0:13:41.720 longer term bets, and so X has received the support 0:13:41.760 --> 0:13:46.560 from Alphabet, which I'm very grateful for. And in that context, 0:13:46.720 --> 0:13:49.360 we're continuing to do our job. How can we take 0:13:49.640 --> 0:13:53.880 these audacious attempts to find something really great for the 0:13:53.920 --> 0:13:58.840 world that is also you know, significant shareholder value production, 0:13:59.320 --> 0:14:03.520 and to do that as efficiently as possible. Understanding we're 0:14:03.559 --> 0:14:06.120 still going to be wrong ninety nine percent of the time. 0:14:06.800 --> 0:14:08.720 The question is not how do we get it so 0:14:08.760 --> 0:14:11.840 we're only wrong ninety percent of the time. The question 0:14:11.880 --> 0:14:15.000 at X is how do we discover the ninety nine 0:14:15.040 --> 0:14:18.800 percent where we're wrong as fast and as cheaply as possible. 0:14:19.320 --> 0:14:22.200 The better we get at that part of the riggor 0:14:23.000 --> 0:14:26.280 that one percent that comes through will come through more 0:14:26.320 --> 0:14:28.400 and more efficiently because we've spent less and less time 0:14:28.440 --> 0:14:30.240 and money on the stuff where it turned out not 0:14:30.320 --> 0:14:31.200 to be a great idea. 0:14:31.320 --> 0:14:33.000 And one of the things you said was that as 0:14:33.040 --> 0:14:35.520 a manager, you don't want to be telling people I 0:14:35.560 --> 0:14:37.520 want to kill that idea. You want them to come 0:14:37.520 --> 0:14:39.400 to you proactivity and kill their own ideas. 0:14:39.560 --> 0:14:41.520 That's right. So if you worked at X and you're 0:14:41.600 --> 0:14:47.480 building the Teleporter project or whatever, if I believe that 0:14:47.640 --> 0:14:52.080 you can't practice intellectual honesty and work in the betterment 0:14:52.720 --> 0:14:57.400 for X and alphabet of X's overall portfolio, including a 0:14:58.160 --> 0:15:02.520 dispassionate view of your own work, at least periodically, you 0:15:02.560 --> 0:15:06.400 and I are fundamentally in antagonism with each other. You're 0:15:06.440 --> 0:15:09.000 not only not on my team, you're actually working against 0:15:09.120 --> 0:15:12.960 my team. If you're being overly partisan to what you do, 0:15:13.840 --> 0:15:16.280 then why do I even. 0:15:16.080 --> 0:15:16.640 Have you here? 0:15:16.680 --> 0:15:20.200 That's horrible. If we were an actual incubator, I would 0:15:20.200 --> 0:15:23.760 get it, because then we're just this system and you're 0:15:23.760 --> 0:15:27.080 trying to leach energy off of us to launch your thing. 0:15:27.320 --> 0:15:30.480 You're supposed to be partisan to your thing, but that's 0:15:30.520 --> 0:15:32.880 not what we're doing at the Moonshot Factory. All of 0:15:32.960 --> 0:15:35.520 us here, including you if you've joined the Moonshot Factory, 0:15:35.840 --> 0:15:38.920 are working to systematize innovation. And while I hope that 0:15:38.960 --> 0:15:41.760 your teleporter works out, I don't hope that nearly as 0:15:41.840 --> 0:15:44.360 much is that the factory works out. And if you 0:15:44.440 --> 0:15:46.920 aren't on that team, you're not going to be happy 0:15:46.920 --> 0:15:47.320 at X. 0:15:48.240 --> 0:15:54.120 So you, obviously, through an iterative process, taken ideas to fruition, 0:15:54.280 --> 0:15:58.720 whether it's Weimo, Brain Wing or others. What does the 0:15:58.760 --> 0:16:02.360 iterative process of the Moonshelt Fight Tree itself been. I 0:16:02.440 --> 0:16:06.600 think he specifically you spoke about measuring or creating a 0:16:06.640 --> 0:16:09.160 balance between efficiency and magic making. 0:16:10.320 --> 0:16:12.600 I mean yes, and I can give you examples of that. 0:16:12.640 --> 0:16:15.880 But let me tell you about something that has evolved 0:16:16.040 --> 0:16:19.880 for us over time. We've been realizing more and more 0:16:20.320 --> 0:16:23.080 that for at least many of the things that we make, 0:16:23.800 --> 0:16:28.640 landing them outside Alphabet is actually better for Alphabet and 0:16:28.880 --> 0:16:32.880 for the project the proto company as it becomes a company. 0:16:33.400 --> 0:16:37.400 So Alphabet can still have a large minority interest in 0:16:37.440 --> 0:16:41.200 this business. But if it's outside of Alphabet and Alphabet 0:16:41.200 --> 0:16:46.760 doesn't control it, then get it can participate in market 0:16:46.800 --> 0:16:50.480 capital and get strategic partners in a way that's different 0:16:50.560 --> 0:16:53.840 than if it's inside alphabet. It can go faster in 0:16:53.880 --> 0:16:58.200 some ways for being decoupled from Alphabet, which is also 0:16:58.240 --> 0:17:02.240 a complex, very large business. So we're finding ways to 0:17:02.280 --> 0:17:05.280 sort of systematize the landing of things more and more 0:17:05.320 --> 0:17:08.959 outside of Alphabet. And that's something we've learned through the 0:17:09.000 --> 0:17:12.600 process of making these other bets. And it sometimes happens 0:17:12.800 --> 0:17:18.720 that being in a culturally, operationally and legally separate entity 0:17:18.920 --> 0:17:23.800 within Alphabet, weimo wing intrinsic these verily these kinds of 0:17:23.800 --> 0:17:26.840 things that came from X. It works. It's sometimes good 0:17:26.880 --> 0:17:29.639 for them, but it's not for all of them. And 0:17:29.680 --> 0:17:32.879 so that's an example where we've been trying to learn 0:17:32.920 --> 0:17:36.200 ourselves how we can do our job better, systematize our 0:17:36.280 --> 0:17:40.800 process and the conveyor belt for these ideas to optimize 0:17:40.960 --> 0:17:43.000 their chances of being really great for the world. 0:17:44.160 --> 0:17:45.840 I don't want to drag you into politics, but this 0:17:45.920 --> 0:17:49.600 word efficiency has obviously become very very loaded recently, and 0:17:49.640 --> 0:17:52.760 there's a kind of wider debate within the country about 0:17:53.160 --> 0:17:56.679 the value of supporting a government supporting long term science 0:17:56.680 --> 0:18:01.360 and research initiatives versus cutting waste and cutting efficiency. If 0:18:01.359 --> 0:18:03.920 you have one sort of piece of advice to another organization, 0:18:04.080 --> 0:18:06.879 be it government or another company, What would you say 0:18:06.920 --> 0:18:10.400 about what you've learned in terms of balancing those two imperatives. 0:18:11.320 --> 0:18:18.399 Well, first of all, getting efficient is generally a good goal, 0:18:19.200 --> 0:18:22.239 but you have to know what game you're playing. In 0:18:22.280 --> 0:18:25.280 our case, because we believe that we're in the moonshot business, 0:18:25.520 --> 0:18:29.120 it's super important, as I've been describing, that the efficiency 0:18:29.160 --> 0:18:31.880 stays balanced with what we're trying to do, and there 0:18:31.920 --> 0:18:34.560 has to be a lot of exploration and a lot 0:18:34.560 --> 0:18:38.920 of being wrong. So if you propose the teleporter project, 0:18:39.280 --> 0:18:42.560 and I start with here's the thirty reasons, that's stupid. 0:18:43.080 --> 0:18:47.280 One you will never bring up a creative idea ever again. Two, 0:18:47.880 --> 0:18:51.960 it's easy to say that thirty thousand reasons why some 0:18:52.240 --> 0:18:56.480 unusual idea isn't going to work. But then we aren't 0:18:56.480 --> 0:18:58.600 going to go on any adventures, which, by the way, 0:18:58.880 --> 0:19:02.879 I don't believe in Taro cards metaphysically, but there is 0:19:02.920 --> 0:19:07.000 a tarot card poster of the fool. It's the only 0:19:07.040 --> 0:19:10.879 thing on the door of my huddle. Because the activity 0:19:11.280 --> 0:19:14.080 of setting out on a new journey is the activity 0:19:14.160 --> 0:19:17.880 of creation, and that's the job that we're all in 0:19:18.080 --> 0:19:21.480 at X, and so if you ran a widget factory, 0:19:21.640 --> 0:19:24.440 you might have a very different set of goals with efficiency, 0:19:24.480 --> 0:19:26.760 and you might do six sigma and that might be 0:19:26.840 --> 0:19:30.159 reasonable for you. Six sigma is the wrong way to 0:19:30.160 --> 0:19:34.919 think about efficiency at a moonshot factory. And you know, 0:19:34.960 --> 0:19:36.800 I'll leave it to people who are smarter than me 0:19:36.880 --> 0:19:40.840 to figure out how the government should focus on efficiency productively. 0:19:41.080 --> 0:19:52.639 Fair enough, coming up, what Astro tele learned from his 0:19:52.720 --> 0:20:08.439 grandfather about innovation, Stay with us, welcome back to tech stuff. 0:20:08.920 --> 0:20:11.919 Before I had the opportunity to interview Astro Teller at 0:20:11.960 --> 0:20:15.000 the Google office in Austin, Texas, I saw him on 0:20:15.040 --> 0:20:18.760 a live panel moderated by Nicholas Thompson, CEO of the Atlantic. 0:20:19.560 --> 0:20:20.760 At one point, Nick. 0:20:20.600 --> 0:20:22.800 Started to say that government had created a lot of 0:20:22.800 --> 0:20:26.359 innovation throughout the twentieth century. At one point, Nicholas Thompson 0:20:26.400 --> 0:20:29.600 started to ask a question around how government had driven 0:20:29.600 --> 0:20:32.520 a lot of innovation in the twentieth century, and Astro 0:20:32.600 --> 0:20:36.840 quickly interjected, saying, actually government funded research, which in turn 0:20:36.920 --> 0:20:37.640 drove innovation. 0:20:38.119 --> 0:20:39.400 It was a fascinating conversation. 0:20:39.800 --> 0:20:41.480 You'll be able to hear the whole thing because it's 0:20:41.480 --> 0:20:44.880 going to be episode ten of the Moonshot podcast, and 0:20:45.080 --> 0:20:48.520 Astro Teller has first hand knowledge of these relationships because 0:20:48.560 --> 0:20:52.200 his grandfather, Edward Teller, was one of the key members 0:20:52.240 --> 0:20:55.080 of the Manhattan Project, the R and D project that 0:20:55.160 --> 0:20:59.040 developed nuclear weapons during World War Two. So I asked Teller, 0:20:59.480 --> 0:21:02.199 how does he think about the balance between academia and 0:21:02.280 --> 0:21:06.480 deep research, government and private labs like Google Eggs when 0:21:06.480 --> 0:21:07.800 it comes to building a future. 0:21:08.400 --> 0:21:10.719 I think it's a place for all of them. You know, 0:21:10.880 --> 0:21:13.879 there will probably from time to time always be some 0:21:14.160 --> 0:21:19.560 issues which are national security level issues, and it's rational 0:21:19.600 --> 0:21:22.639 for any government, including the United States government, to spend 0:21:22.720 --> 0:21:25.680 money to solve those things in a somewhat cost and 0:21:25.760 --> 0:21:28.440 sensitive way because it's a national security issue. The Manhattan 0:21:28.440 --> 0:21:29.719 Project was an example of that. 0:21:29.800 --> 0:21:31.720 You have toio granfather' not working on that, by. 0:21:31.560 --> 0:21:36.479 The way, frequently. Yeah, there will always be a place 0:21:36.600 --> 0:21:42.360 in all countries for basic science because the raw material 0:21:42.480 --> 0:21:46.679 of training the next generation of people in all the 0:21:46.720 --> 0:21:52.480 stem fields creates this sort of rich soil from which 0:21:52.720 --> 0:21:58.520 really great new things can spring. And then obviously private organizations, 0:21:58.920 --> 0:22:02.800 either ones in history like the Bell Labs or Xerox Park, 0:22:02.920 --> 0:22:06.160 or maybe more modern things like x the Moonshot factory, 0:22:06.560 --> 0:22:09.879 I think also have a place because something has to 0:22:10.000 --> 0:22:15.880 bridge between the Silicon Valley venture world of we can 0:22:15.920 --> 0:22:17.800 see where we're going. It's kind of a ways off, 0:22:17.800 --> 0:22:19.200 but we can see it, and we just want to 0:22:19.280 --> 0:22:23.520 rush there as fast as possible. Academia, Oh my god, 0:22:23.600 --> 0:22:25.879 we found a frictionless surface, but we have no idea 0:22:25.920 --> 0:22:29.159 what this is good for. There's a big gap between 0:22:29.200 --> 0:22:33.359 those two things, and I think that moonshot factories, not 0:22:33.600 --> 0:22:35.879 just the one that we're making, but I think others 0:22:35.920 --> 0:22:39.560 out in the world could over time fill that gap 0:22:39.600 --> 0:22:40.399 really effectively. 0:22:40.960 --> 0:22:42.679 I wonder if you you could talk briefly about your 0:22:42.920 --> 0:22:45.280 grandfather and was that one conversation or one piece of 0:22:45.320 --> 0:22:47.680 advice he gave you from the point of view of 0:22:47.680 --> 0:22:49.560 the Manhattan Project that put you on the path through 0:22:49.600 --> 0:22:50.000 on today. 0:22:51.400 --> 0:22:53.080 I mean, I'd give a few things. First of all, 0:22:53.119 --> 0:22:57.680 he liked to quote Neil's Boorr, who said that an 0:22:57.720 --> 0:23:00.680 expert is someone who's made the majority of the mistakes 0:23:00.680 --> 0:23:04.400 in their field. And I think that that is one 0:23:04.480 --> 0:23:06.600 of the things that helped me lock in on the 0:23:06.720 --> 0:23:11.800 understanding of failure is an inherent part of becoming an expert, 0:23:12.040 --> 0:23:14.719