That being said, there's probably still a place for busses during rush hour in a lot of places.

The main reason why non-dense places don't see good bus service is because of labor costs. In developed countries labor is easily the biggest source of cost when it comes to a bus, so agencies will run one 40-passenger bus every half hour rather than three 15-passenger buses every ten minutes. Removing labor makes the latter scenario affordable, and if buses are coming every ten minutes average wait time is about five minutes, which is comparable to Lyft or Uber today in some places.

I mean today it is a no-go decision from be, simply because it takes a minimum of twice as long to go to work by bus as by my car, but with a smarter route, that may not be an issue.

To take a bus from point A to point B, I have to walk to point C, wait for the bus, get on, wait to be dropped of at point D, cross the intersection to point E, wait for a transfer bus, get on, wait to be dropped off at point F, and finally walk to point B.

With this system I call a cab, wait for it at point A, get on, get dropped of at point B. Rather than my time being fragmented into 3 periods of walking interspersed with 4 periods of waiting, it is just 2 periods of waiting, which makes it much easier to make use of the time. In my experience riding buses, only about a third of the time spent is usable (except for long distance commuter routes), while nearly all the time spent in a cab is usable.

Even if the shared cab took a bit longer than a bus, I would still choose it over a bus for the improved productivity / enjoyment.

Like busses and trains? Seriously. Public transit.

The issue is a geometric one. There is only a fixed road capacity. Consuming that road capacity with deadheading vehicles is the worst usage.

IE, it can't go point to point, and is instead stuck in a movement pattern that isn't changing.

I think it would be feasible for the self-driving taxi companies to have a small fleet of cars in each neighborhood that could respond rapidly to take people for short trips.

By definition, public transit can only serve a specific segment of the commuting population.

That's not inherent to public transport. You can easily imagine a public network of bus transit where users send in a request for transit an hour or two beforehand; software coalesces the set of requests into a dynamic set of routes and stops; and tells the user to turn up at a stop (relatively) near the origination point and drops them off at a stop (relatively) near their destination point.

Whether it's economic or feasible is a separate issue. But funding mechanism, transport form factor, and autonomy are all distinct parts of a transportation method, and it's a mistake to think of "private autonomous cars" as competing with "legacy human-driven buses with set routes."

If "on demand mass transit via large multipeople vehicles (IE busses)" were feasible, then you'd already see companies like Uber doing it.

On demand point to point, where you don't have to wait a dozen stops to get where you want, is just strictly better than the alternatives in many many ways.

Sure. Get around this by having preset schedules. Or infer it. These details matter, but you can point out similar issues for public transit as it exists today and for private car ownership as it exists today: both are commonly used, despite drawbacks.

> If "on demand mass transit via large multipeople vehicles (IE busses)" were feasible, then you'd already see companies like Uber doing it.

Not at all. Uber and Lyft are premised on having a large base of independent contractors to perform all the labor and provide all the capital. "Large multipeople vehicles" would require complicated integrations with different municipalities, or for Uber/Lyft to purchase their own fleet of buses. Neither aligns well with their current businesses.

> On demand point to point, where you don't have to wait a dozen stops to get where you want, is just strictly better than the alternatives in many many ways.

Sure. And shared rides is strictly better in one very important way: cost. I'm much less cost-sensitive than the average American, and I take Lyft Line far more often than a regular Lyft.

It's not about what's best, but best in certain situations. I value latency sometimes, and cost sometimes.

Current public transportation is rigidly designed for capturing massive economies of scale (still good), while a dynamic swarm of self-driving vehicles can fulfill last-mile demand with the benefits of economies of scale (market opportunity).

There is a fixed road area to work with.

Links? I haven't found any way to practically share Lyfts for long-ish trips. I don't see any evidence Lyft or Uber will ever do much with shared-rides.

For self-driving cars, people wanting to save money will drive their own cars before they share rides. If you have a car, you've sunk a lot of money into it and so the per-ride cost is fairly small. Even a self-driving car has to price-in wear-and-tear.