Dwarkesh Patel: prediction

Transcript context

…How I think about that is that you kind of see a similar structure in anything that's project-based or anything where there's an element of figuring out what you're doing while you're doing it. Compared to a large-scale manufacturing option where you spend a lot of time figuring out what exactly it is that you're building. You spend a lot of time designing it to be built and do your first number of runs through it, then you tweak your process to make it more efficient. There's always an element of tweaking it to make it better, but to some extent, the process of figuring out what you're doing is largely separate from the actual doing of it yourself. For a project-based industry, it's not quite like that. You have to build your process on the fly. Of course, there are best practices that shape it, right? For somebody writing a new software project or anything project-based, like making a movie, they have a rough idea for how it's going to go together. But there's going to be a lot of unforeseen things that kind of come up like that. The biggest difference is that either those things can often scale in a way that you can't with a building. Once you're done with the software project, you can deploy it to 1,000 or 100,000, or 1 million people, right? Once you finish making a movie, 100 million people can watch it or whatever. It doesn't quite look the same with a building. You don't really have the ability to spend a lot of time upfront figuring out how this thing needs to go. You kind of need to figure out a way to get this thing together without spending a huge amount of time that would be justified by the sheer size of it. I was able to dig up a few references for software projects and how often they just have these big, long tails. Sometimes they just go massively, massively over budget. A lot of times, they just don't get completed at all, which is shocking, but because of how many people it can then be deployed to after it's done, the economics of it are slightly different. I see, yeah. There's a famous law in software that says that a project will take longer than you expect even after you recount for the fact that it will take longer than you expect. Yeah. Hofstadter's law or something like that is what I think it is.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.