00:02:08 · Ben Thompson, Scott Kirby, John Collison
[00:02:08] S2: the cycle with large capital purchases is actually also a tech story that's relevant with memory. Like, Samsung came to dominate memory by systematically investing into every downturn because it's very cyclical. And obviously it's hugely beneficial today. The flip side of that is everyone wants to invest when it's at its peak. Have we gone full circle and now [00:02:29] S0: Yeah, it's you want to invest when everyone else, when things are down and not invest when things are up. I mean, for us, and probably for like memory, it's a little harder too because our assets are so long lived. For us, our assets are thirty year. I guess for a chip factory it's less because the technology moves away. But it's a thirty year life. So like we're going go through several recessions and we're going to go through several ups and downs. So it's a little harder to do that. But you get better terms, get better deals, you know, like we got really good pricing, you know, for example, during the downturn. [00:03:00] S1: You're the only person ordering here? [00:03:01] S0: Because we're the only one in the world. And we sort of launched the market. Like, you know, for Boeing, they kind of could go to other customers and say, look, United believes you should too. Right.
00:03:01 · Scott Kirby, John Collison
[00:03:01] S0: Because we're the only one in the world. And we sort of launched the market. Like, you know, for Boeing, they kind of could go to other customers and say, look, United believes you should too. Right. [00:03:11] S1: So I remember... It's very striking. I remember that period of people over extrapolating COVID. Like the tech version of that was companies deciding that they're going to be fully remote forever, and a lot of companies decided that, and then were like, wait a second. No, that's not right. Why were people so wrong on the travel side of things? Because [00:03:30] S0: there are two things that are human nature that are mistakes. One is to extrapolate the current into the future. Assume that everything that's happening today is going to continue into the future. And two, trusting people on surveys. Surveys are the absolute worst way to figure out what's going to happen in the future. Because the reason travel experts thought travel, they did surveys and people were terrified. Like everyone's dying. Right. And you ask them, Are you going to travel internationally Yeah. No. It's like, but, and that was true of what they thought at the time. But like once it was over, of course they were going to go back to traveling again. So ignore... And by the way, I said this sort of like, experts live in the world of two standard deviations. They force fit everything into things that they've already seen. And so when things happen outside two standard deviations, when the tails happen, which is where you create all the The frequency
00:03:30 · Scott Kirby, John Collison
[00:03:30] S0: there are two things that are human nature that are mistakes. One is to extrapolate the current into the future. Assume that everything that's happening today is going to continue into the future. And two, trusting people on surveys. Surveys are the absolute worst way to figure out what's going to happen in the future. Because the reason travel experts thought travel, they did surveys and people were terrified. Like everyone's dying. Right. And you ask them, Are you going to travel internationally Yeah. No. It's like, but, and that was true of what they thought at the time. But like once it was over, of course they were going to go back to traveling again. So ignore... And by the way, I said this sort of like, experts live in the world of two standard deviations. They force fit everything into things that they've already seen. And so when things happen outside two standard deviations, when the tails happen, which is where you create all the The frequency [00:04:19] S1: in finance, the frequency of Six Sigma events is...
00:04:19 · John Collison, Scott Kirby
[00:04:19] S1: in finance, the frequency of Six Sigma events is... [00:04:22] S0: That's where I first started. You know, like, know, VAR, valuation at risk is the dumbest model in history. It's stupid. Like, you know, things that shouldn't happen in the history of the universe happen once every five years because it's a dumb, dumb, fundamentally flawed mathematical model that assumes everything is normally distributed. It's not. And it's just the perfect... I had a debate when I was getting my master's degree back in the 90s. This was about black souls. You know, this is such a dumb model for valuing options. Like, because it assumes a normal distribution. The world's not normally distributed. The tails are much fatter in everything that matters. And anyway, that applies to everything, but especially in a crisis. [00:05:04] S1: Black Scholes has a lot of assume a spherical cow like assumptions baked into it. Okay, love hearing you talk about the over fixation on surveys, But like the travel industry, the airline industry loves surveys. United probably surveys people up the wazoo. And definitely at Stripe, find we need to deprogram people a bit because in tech, people love running things via metrics and you can outsource your thinking to the dashboards and And so we're trying to encourage people to... I mean, your opinions should survive contact with reality, but the metrics are not going to tell you what to do. That sounds like...
00:05:04 · John Collison, Scott Kirby
[00:05:04] S1: Black Scholes has a lot of assume a spherical cow like assumptions baked into it. Okay, love hearing you talk about the over fixation on surveys, But like the travel industry, the airline industry loves surveys. United probably surveys people up the wazoo. And definitely at Stripe, find we need to deprogram people a bit because in tech, people love running things via metrics and you can outsource your thinking to the dashboards and And so we're trying to encourage people to... I mean, your opinions should survive contact with reality, but the metrics are not going to tell you what to do. That sounds like... [00:05:38] S0: So we use survey. I like surveys for how did people feel about stuff we've already done. I ignore surveys for what we could do in the future. [00:05:46] S1: AppleFamously does not do a lot of focus group testing on the products. They just like build the products [00:05:50] S0: That's they want to the way I, you know, stuff that we're doing right now. Like today, we rolled out a Starlink tracker. We didn't ask anyone about that. Like, I sit on an airplane and talk to customers and I know they're gonna think that's cool. Yes. And they do think it's cool. [00:06:02] S1: Well, you're getting to the second thing which is people try to over measure tiny steps and it's just like, do the Starlink tracker. If it's not useful, whatever. Doesn't matter.
00:06:02 · John Collison, Scott Kirby, Ben Thompson
[00:06:02] S1: Well, you're getting to the second thing which is people try to over measure tiny steps and it's just like, do the Starlink tracker. If it's not useful, whatever. Doesn't matter. [00:06:10] S0: Yeah, and that's like, and especially in a world of AI, you can do so much so fast. Like, it's really powerful. I have a different view of what we'll probably talk about, but I at least have a different view of where the value is going be created in AI. But you can just try so much so fast. And some of it doesn't work and people don't care. Somebody do, like it takes me to take more time to do a survey or a focus group, and just do it, see if they like it, and if they don't, stop, and if they do, do it, more of it. [00:06:35] S2: How important is your ability to do these sorts of experimentation? We talked, you know, eight months ago, redoing your whole tech stack so you could sort of put all these things on top of it.