Speaking on Personal Clouds in SF, and Robocars in Phoenix

Two upcoming talks:

Tomorrow (April 4) I will give a very short talk at the meeting of the personal clouds interest group. As far as I know, I was among the first to propose the concept of the personal cloud in my essages on the Data Deposit Box back in 2007, and while my essays are not the reason for it, the idea is gaining some traction now as more and more people think about the consequences of moving everything into the corporate clouds.

My lighting talk will cover what I see as the challenges to get the public to accept a system where the computing resources are responsible to them rather than to various web sites.

On April 22, I will be at the 14th International Conference on Automated People Movers and Automated Transit speaking in the opening plenary. The APM industry is a large, multi-billion dollar one, and it's in for a shakeup thanks to robocars, which will allow automated people moving on plain concrete, with no need for dedicated right-of-way or guideways. APMs have traditionally been very high-end projects, costing hundreds of millions of dollars per mile.

The best place to find me otherwise is at Singularity University Events. While schedules are being worked on, with luck you see me this year in Denmark, Hungary and a few other places overseas, in addition to here in Silicon Valley of course.


I for one wish you didn't work for Google. Since they are the dominant Robocar player, it would be nice if you could gossip, dish, speculate, etc. about the company.

It is so encouraging to read your words "The APM industry is a large, multi-billion dollar one, and it’s in for a shakeup thanks to robocars, which will allow automated people moving on plain concrete, with no need for dedicated right-of-way or guideways. APMs have traditionally been very high-end projects, costing hundreds of millions of dollars per mile."
As a civil engineer, road designer, road safety specialist, autonomous vehicle enthusiast and tax payer I am appalled at how this rapidly emerging robocar technology is being ignored in all of our short, mid and long-term transportation planning. I am not aware of a single jurisdiction in North America that has taken account of this technology and its impact on public transportation. (Maybe you have? - I would love to know if somewhere is planning for this - maybe in California?)
At what point will the public realize that they are being asked to get into debt to fund these massively expensive light rail transit (LRT) schemes - that may not even be finished being built by the time that we have robo-taxis running round providing a higher level of service with virtually zero cost to the public purse.
One question that I would like to ask in respect of robo-taxis, and I asked this in my latest presentation to the Canadian Institute of Transportation Engineers (CITE) "Is there an elephant in the room?" - who do you think will end up being responsible for coordinating these robo-taxis? Will the municipalities choose to transfer their transit resources and oversee the cloud-based system for ordering and distributing robo-taxis in the most efficient manner? Or will it best left to the private sector? (although we could end up with numerous fleets competing and therefore not working as efficiently as a single coordinated fleet). Or will we see a mix of public and private rob-taxi fleets?
Your feedback from the APM Conference would be most appreciated - from my work with the Institute of Transportation Engineers I know that there is now a growing awareness, but I am not sure that your ideas will get the warmest of welcomes!
Best wishes for your travels.

I disagree on this point. I think a lot of history clearly shows that numerous competing private fleets would be vastly more efficient than a centralized system. I'm surprised that's even debatable. I mean sure, in theory one central system could be a bit more efficient, but it's rare to see that in practice. There are not the same incentives.

I think the light rails -- which are hugely inefficient, using more energy than the average car per passenger in almost all cities -- are a perfect example of that. You correctly point out how badly the central planners are doing, but ignore that in thinking they will do better running the taxis.

I have often wondered if rail coridoors could have the rails embedded in an otherwise flat paved surface and used for simple automated vehicles, including trains themselves, buses, and private vehicles.
It's hard to believe that current signalling systems couldn't be completely automated and expanded to accomidate additional vehicle types.
As for trains themselves, they appear to be massively taller and heavier than they need to be.
Could we not get rid of the central coridoor put 1 full carriage length lifting door on each side of carriages. This would make each carriage no higher than a car.
Similar to automated cars train should be able to be light becuse their crash likelihood is low.

Is this a method to gain some of the advantages of automation initially in a controlled environment?

Roads with embedded rail cost a bunch more to maintain, but yes they are valuable as you get dual use of the right of way.

There are apparently regulations that make trains be so heavy in part because they might collide with other things on the rails, and a locomotive hitting a light car is super destructive for the light car.

Can one lane of taxis carry as many people per hour as a 150' long train that comes about every minute and a half in each direction during rush hour on the central segment (Copley to Government Center) of the MBTA Green Line?

England apparently has ``competition'' for bus service in a lot of areas which in practice apparently means that if you hold a monthly pass from bus company A, and bus company B's bus happens to show up first, you generally get to waste time waiting for the next bus company A bus to show up. This does not really seem to count as efficient.

Are 10 casual carpool robotaxi companies really going to be able to match up riders as efficiently as one big casual carpool robotaxi company that gets all of the riders to itself?

On that scale, convincing more people to use the carpool service should reduce congestion, which may lead to better return on capital investment if trips complete faster such that a single vehicle can complete more trips in one day.

There's even the question of: if a large employer is having trouble finding enough good employees, does subsidizing the cost of transportation in their region to reduce congestion have the potential to let them get more out of their existing employees as a result of a shorter commute, and can that improved productivity help to pay for not maximizing the transportation revenue they collect from non-employees? (This gets more complicated if the employer is already running a transit system to let their employees work in traffic, at which point the question becomes whether working while in transit is as productive as working in the office.)

No, at rush hour, the long train has more capacity in the fixed space, but the reality is that you only have that frequency on the central segment at rush hour. But you don't have just that, and if you did it would be damned expensive to build with its tunnels and private RoW.

The Boston T, if you look at the DoE's spreadsheet uses 3469 BTUs per passenger mile, about the same as the average car, and way more than an efficient car like a Prius or most of the electric cars even with just one person in them. So whatever the capacity of the T, it's worse for the environment than electric cars would be, and even many gasoline cars.

You may be misunderstanding why competing systems are more efficient than a larger centralized one. Of course the centralized one is more efficient in theory. But it's extremely rare that this is true in reality. You can come up with all sorts of arguments about how one big network would work better in theory. They are not that germane.

Or rather, if they are that powerful, the competing companies start cooperating where it makes sense for them to do so. If company A doesn't have a cab very close to you, it pays company B to send one to you, and makes you happier, and gets reciprocation or appropriate money flow. This happens all the time -- you have probably noticed that lots of the time when you book at ticket on an airline you end up on a codeshare partner. But I am surprised I even have to lay this out. Have people seen a lot of cases where a centralized monopoly is more efficient than several competing companies when there are not really overwhelming natural monopoly factors or economies of scale?

And to top it off, if a street can't handle the demand of private vehicles, just start running buses. The green line could not compete with a fleet of buses with private RoW (or even shared RoW) if you add the bus ability to pass other buses. Even if they were competing buses.

Add new comment