This market's resolution piggybacks on Jeff Atwood and John Carmack's bet documented on the Coding Horror blog: https://blog.codinghorror.com/the-2030-self-driving-car-bet/
However they resolve their bet is how this market resolves.
1,000
3.00Sorry, But I think there's a serious problem with the description of this market. A "Level 5" car is generally defined to not have geographic boundaries that it's able to operate in. (It might have legal boundaries, like only being authorized to operate in one country or one state). So saying "at least one city" doesn't make sense.
Look at the bottom right hand box. "Anywhere".
@DavidFWatson What I'm saying is that either a car is able to drive autonomously throughout the entire United States, (or some other country or State), or it doesn't qualify as Level 5.
@DavidFWatson I think you are correct about the definition of L5, but other factors might prevent the ride in practice. Regulatory issues for example or simply that the service provider doesn't want their vehicles to end up in the boonies.
I'm more bothered by the fact that it's one of the top cities. Driving in LA on a sunny day seems like a much easier problem than in LA during a snowstorm when lane markings might be covered.
@Enlil I don’t think it snows in LA.
But also, I think you’re also not giving a good example of which situations are hard to handle when HD maps are available, the car doesn’t need to see the lanes to know where they are. Hard problems are ones involving new/dynamic objects. Hand signals from people directing traffic, following cones set up for construction. Doing that stuff in the rain when the LiDAR barely works. That’s the hard stuff I think.
@dreev Those rides have humans monitoring and ready to intervene remotely, afaik. Questionable to consider that level 5.
@ErikBjareholt Correct, Waymo is currently level 4, not level 5. But I don't think humans are intervening at all on typical trips and I don't think the humans can take direct control when the car is already moving.
Ha, I think @firingmanager meant "the market" as in consumers in the marketplace of self-driving cars and @DavidFWatson meant "the market" as in this Manifold market. So, yes, I think you're both correct!
I'm not sure we're as close on, specifically, level 5 autonomy as we seem. The 2023 state of the art with Waymo and Cruise in places like Phoenix and San Francisco is level 4. I think it's another big leap to having no restrictions on where the car can go and no restrictions on conditions under which the car can drive.
Level 5 autonomy might be an AI-complete problem, so to speak, like where AGI is a prereq. Maybe not literally but there needs to be almost nothing a human driver can deal with that the AI can't. And there's a quite long tail of things in that category to painstakingly work through. I wouldn't be surprised if it's commonplace to see cars with no one in the driver's seat all over the world and yet none of it counts as level 5.
With current Waymo and Cruise cars there are still situations like road closures or traffic cops making hand signals where the car can't figure out what to do and has to ask for human help.
@dreev when you phrase the question as "available in a major city" I assume it still means there can be some restrictions outside of the city, right? The car must be level 5 within a major city... Or did I misunderstand the phrasing of the question?
@mjmandl Level 5 within the city, yeah. I see the ambiguity though. Does an inability to leave a restricted area mean it's not truly level 5? I'm comfortable saying that for the spirit of this question, that restriction doesn't count, as long as it can operate in the entire city, including non-public roads, reconfigured roads it hasn't seen before, etc.
Or we could just make a reasonable distinction between can't and won't.
How about this: it has to be able to do anything a maximally rule-following taxi driver can do. That can include not leaving the city.