Comment by origin_path

3 years ago

The main lesson here seems to be about the value of incrementalism. This project was always pitched as self driving cars in one leap, which set huge expectations and more or less ruled out just shipping incrementally better driver assistance year after year. Probably because the space became dominated by Google and Page-ist futurism, so they weren't able to sell anything that would be expected to come with an actual car because "automotive supply chain" didn't fit with Google's self image. And then everyone got scared that they'd succeed and tried to copy their approach.

The mining site thing is probably where things should have begun, or highway driving only, but the question is, could a company have motivated ML and programming experts with such a humble starting point? Or was the hole-in-one approach required to build a team that could get anywhere at all?

4 comments

origin_path

Reply
lll-o-lll  3 years ago

It’s because self driving doesn’t provide much value until you can remove the human. If you can remove the human, last mile package delivery is suddenly super cheap; taxis are cheap; trucking is much cheaper; share car services are transformed as a business model.

With the human still required you get nothing, so why invest at all?

  • origin_path  3 years ago

    I think a lot of people would be happy to put their feet up once they'd navigated onto a highway, or have their car go find a parking spot by itself, or come back from one. The sort of stuff Tesla is trying to do is in many ways far more practical except they keep diverting resources to FSD and end up not nailing thee more prosaic conveniences.

    • lll-o-lll  3 years ago

      The cost benefit analysis just doesn’t add up though. I can see this adding a small amount to the purchasing decision of a prospective buyer, but without “remove the human” as the end goal these technologies do not provide the return on investment. No one has a crystal ball though, and a few years ago “autonomous vehicles” looked promising. It’s just that we are now looking at the sunken cost fallacy when it hasn’t worked out.