Yes, that's a nice introduction to the pros and cons.

One of the cons it doesn't mention is that shared cars (whether driverless or not) necessarily increase the number of vehicle miles travelled. If Alice wants to go from A to B, and Chuck wants to go from C to D, then if they both use private cars, the total distance travelled is AB+CD. If they use a shared car (and Alice goes first), the total distance travelled is AB+BC+CD.

Shared cars could reduce the total number of vehicles, the total number of parking spaces required, and the fuel consumption per mile travelled. On the other hand, they're likely to increase the total number of miles travelled, average congestion, and the average time required for travel (because you sometimes need to wait for a car).

You could mitigate some of the negatives by increasing the fleet size and providing carefully-distributed parking, among other things.

But this is clearly one of those cases where you don't get the best outcome unless you do a great deal of planning and preparation up front. Which is just what the article concludes.

Going to eventually kill another whole class of low end jobs that are something of a "job of last resort" for (mostly) desperate men that can't find other work.

"Well, crap, if I can't find a manufacturing job, and i can't get a gig swinging a hammer, I guess I can always go drive a (truck, cab, uber, etc..)"

May not seem like a big thing, but pile it on to the rest of tragic shift for (formerly, or on the edge)lower middle class workers... can't get a job building things in this country, now the service jobs are going away - wow.

Thank your lucky stars that your family was able to get you an education. But if you think it doesn't eventually effect you even peripherally - think again.

Mr. Levy - thanks for the link.

"Some of my friends remind me that road expansions initially reduce congestion but soon attract more drivers and congestion returns."

That old fallacy from the sixties is easily debunked. Simply observe that the number of cars on the road keeps increasing after roadbuilding stops, with the excess spilling over onto adjacent roads as available, or just jamming up. Only a decrease in population (a cardinal anathema to civic boosters and economists) can stop or reverse this trend. More people -> more cars. Curmudgeon's Law.

The current hype around self-driving cars, and its eager consumption by the technically unsavvy population, recalls the periodic generational buzz over artificial intelligence. Each time, a sudden burgeoning of utopian and dystopian pipe dreams quickly fades as the complexity of the task becomes apparent once again.

The notion of computer-controlled cars tooling seamlessly bumper to bumper down the highway can be easily simulated in steady state mode, but the neat artificial scenario quickly falls apart when transients occur: if a motor breaks down, or a car wants to merge into the flow or change lanes, or somebody's destination suddenly changes and necessitates a series of quick trans-flow maneuvers. This scenario is directly analogus to fluid flow in perturbed time-varying medium, where turbulence and trans-sonic shocks rapidly break up the flow coherence. (Oops, I feel like I just belched in church.)

And yes, no-stop intersections are achievable--in computer simulations--when the traffic density is low enough, everybody does exactly what they are instructed, and/or instantaneous vehicle halts and speed resumptions are permitted. Right.

Bottom line: Enjoy the YouTube videos, but give those GoogleBugs plenty of room. They still need gobs of pampering.

Is your title Some Impacts of Driverless Cars a sly reference to the recent impact of that Google driverless car on a VTA bus?

Allen Akin: you raise an interesting twist. But I do believe that web-powered, critical-mass services like Uber are successfully minimizing the BC component. When I call for a Uber, the map usually shows me 4+ cars available within 3 miles. Then I am typically taking a 25 mile trip to SFO, where I'm sure the Uber driver need only wait a few minutes for a pickup.

I also read recently that approximately one-third of cars on SF streets are looking for parking! (it was a serious analysis, not a snarky comment.) And of course there is a snowball effect: those cars hunting for parking are creating congestion making it even harder for all the other cars hunting or trying to get across town. Shared cars, autonomous or not, could radically reduce this.

BW wrote:

"Going to eventually kill another whole class of low end jobs that are something of a "job of last resort" for (mostly) desperate men that can't find other work.

"Well, crap, if I can't find a manufacturing job, and i can't get a gig swinging a hammer, I guess I can always go drive a (truck, cab, uber, etc..)"

This sounds extremely condescending. Believe it or not, swinging a hammer is healthier than imprisonment in a cubicle with your vertebrae fusing together as you stare at a laptop screen manipulating funds instead of building anything. Or coming up with yet another silly "app" to perpetuate the smartphone addiction epidemic that is slowly killing our society.

We have a compromised free market system wherein people can choose to be pragmatic or else suffer the consequences. I believe that a purer free market based on meritocracy will help people find better jobs.

There is no shame in blue collar jobs. Some of them are more complex than you think.

rez: try again.

Or even better: try getting an entry level job swinging a hammer. Go ahead, please. The wages have been cut in half.

Nothing wrong with a good, strong trade job. Particularly, an organized trade. It's just they've been attacked and are being gutted as well. Good try though, as opposed to addressing anything related to the topic: driverless transportation, and in this case, the effect it will have on certain sectors of employment.

@pickpocket: There's no free lunch. For a given set of origins and destinations AB and CD, the BC connection IS the minimum -- you can't do any better.* The only way to reduce the total number of vehicle miles travelled is to reduce the amount of sharing: Trade more vehicles and more parking places and more driver waiting time to get fewer total miles and less passenger waiting time. That's what's happening when you see a lot of Uber drivers in a small area.

You could cherry-pick the routes that have the shortest empty (BC) legs. That's what's happening when a driver takes you to the airport and then hangs around there waiting for another fare. This is not a bad thing; it's a case of using shared cars for the routes they handle best and leaving the others to some different mode of transportation. It wouldn't make sense to use shared cars for a San Francisco to Palo Alto commute; the empty leg would be huge. For that kind of route it's buses and trains FTW. The WSJ article implies this, though I wish it had been more explicit about it.

From what I've read, cruising for parking places is a real problem in some areas. However, if you don't have parking places for them, shared cars have to cruise while waiting for their next fare. The more shared cars you have (to minimize miles travelled and passenger waiting time), the more parking places you need to prevent cruising. If you make the cars wait in centralized garages, then you add more empty miles getting to and from the garages. Still no free lunch.

I think the bottom line is that shared cars (driverless or not) have some great advantages, but the math shows that carelessly replacing private cars with them would actually increase traffic congestion. They're not a substitute for mass transportation.

* You can do worse, though, and that's what you'd expect in practice because of the parking issue.

I'm with BW - the average working man has seen his world wiped out, due to nafta, etc..
Used to be a man could pick up a 2nd job driving etc, but not any more.

In 1991 I was getting around Zhuhai City or Shenzhen (I forget which for this tale) with someone who was Cantonese speaking. We used a popular van service to get around. The vans with a few rows of seats had a two person crew, typically a couple whose business it was. One would drive, the other stand in the doorway. The one in the doorway would ask, when flagged down, where you were going. If your destination was convenient to the other people/destinations already on the bus they would quote a fare. If you accepted, you paid and got on. It was OK to get turned down by a van - you weren't convenient. So flag down another, there were always a few around. The vans we see, did not have a fixed route and their travels might look like Brownian Movement (random) from above. Their system self adjusts to demand.

We see that the process demands a good knowledge of the city and a good short term memory for queuing the people in destination order and so on. Whether the business model still operates in China I don't know. But it is only an extension to the current software map/routing packages many use including cab drivers. And a natural for self driving vehicles. While some of us would not like to use a cab with such a package, small bus or passenger van sharing is acceptable to most people. It would not be surprising if Tesla and Google had something like it in mind operating through an app on our cell phones to pass on our present locations and where we want to go. The app could calculate the fare for our acceptance.

The question of the impact of self driving vehicles on employment that was raised should be a plain, understandable example of a general problem that is discussed in articles, but not part of political discourse at all though it affects so many millions of people. Perhaps that is due to our Pay-To-Play political system. In general it's better to say that job descriptions are going away before jobs are going away. And change is accelerating while human life and health spans are increasing.

Substantial parts of every generation become quite rigid in their thirties, it's likely genetic, and unable to assimilate a changed world around them - a natural source of a kind of conservatism. That large part of the population is being crowded harder and harder.

Many jobs cannot be automated. An everyday one is child care. We presumably wouldn't dare to bring up kids with robots because it would short circuit their development. They would tend to become sociopaths and robot-like behind the facade themselves. If humans don't have others to interact with early we likely don't even fully develop consciousness. We are making trouble for ourselves that way already. On the other hand, as Japan is doing, elder care help is quite possible.

Mercedes announced that they would employ more human factory workers where their flexibility is of advantage. In the US it's Main Street that is more entrepreneurial, innovative, and net new jobs creating than big corporations. Making it easier to get capital and streamlining their legal environment should be a part of any jobs program. Main Street has been defined as 500 or fewer employees.

"We presumably wouldn't dare to bring up kids with robots because it would short circuit their development."

What about TV or, more recently, tablet computers? Both qualify as robots. TV even explicitly calls its content programs.

Let's remember there is no such thing as a self-driving car.

Currently, and most probably for many decades there will be
the requirement that a driver sit in the driver seat and pay
attention and be ready to take the wheel at any time.

That is really not much of a win unless you are shuttling more
people with you, like a cab/Uber driver.

The problem with Uber is the pay is so low it is not really a
serious job. At least that is what I've heard. The costs of a car
insurance, gas, maintenance are pretty high. Or maybe Uber
drivers just say that to dissuade competition?

Google is rightfully nervous about having a driver ready to take the wheel at any time. The driver can't read the car's mind to understand whether it's acting correctly. Just imagine driving with your spouse having a duplicate set of controls. As it is, I always see that right foot stomping on the imaginary brake pedal. Maybe purchasers will need a 40-hour course behind the wheel to get used to the situation.

"Some of my friends remind me that road expansions initially reduce congestion but soon attract more drivers and congestion returns."

ome of my friends remind me that building housing initially reduces the jobs-housing ratio but it soon attracts more people and the imbalance returns.

These must be derivatives of a general societal law, in the manner of Gresham's Law. Would you know who it is named for?

Thanks for the response @Allen Akin I still think there are so many forces that make car sharing a net reduction in congestion. And contrary to @VR's statement, it's probably easier than ever for an average guy to pick up a decent 2nd job driving.

- I still think services like Uber have reached critical mass such that BC distance and wait time is minimized.

- your example of a commuter from SF to PA...they will pick up a new fare very quickly and I expect the routing algorithms do (or could) optimize selection so driver goes either to another pickup-rich location or toward their home base.

- consider that SF is in the happy situation where a similar (or slightly greater) number of commuters leave the city each a.m. as come into it. So between 6 and 9 a.m. there are two large populations going each way. I don't buy your vision of a mass of Uber drivers waiting for extended periods or doing empty return trips.

- last week I encoutered my first Uber driver who was simultaneously a Lyft driver! This is awesome...free markets will further commoditize car-service and level the demand and optimize routings.

- and we haven't even touched on Uber-pool (multiple people carried simultaneously on similar routes.) No question this reduces car-miles.

When it comes to driverless cars, we don't really have a good idea of how they will be used. I know that in certain urban areas people will walk or use public transport to the local mega grocery store and use a taxi, uber or Cityshare (a system similar to zipcar) their groceries home and leave the car at the curb. In cities where bike share systems are popular, we see that many bikes are picked up at one location and dropped off at another on such a regular basis that the business model is that the bikes need to be put on a pickup from the popular drop off places and returned to the popular pick up places overnight so that the one way riders are able to reuse them the next day.

Taking these concepts to the logical rationale that similar uses of driverless cars would mean that yes the cars have to be returned, but the beauty of a driverless car is that the cars could be taken back to the "depot" for overnight cleaning, servicing, charging and then returned to the pickup locations for use the following day. The beauty of being able to return them to the pickup locations is that to a large extent that can be done overnight when the roads are empty.

However, the longer a car remains at a popular drop off location the more time it is not in use. The same can be said with bike share. Zip cars tend to be returned to their original pickup location. City Share cars work on a different model and they tend to be more of a one way ride as they work on a time factor rather than a mileage factor with the cost of the ride increasing on 15 minute increments.

If we ask whether the popularity of driverless cars becomes a share program versus personal ownership, we should, I think, look at the models of rideshare and bikeshare models, or even car rentals, rather than just relying on uber style usage.

re: jobs - yes there is a LOT of cross-over by drivers between services. Despite the 'loss leader' approach of paying drivers a bit higher wage than will be offered in the future (by both services at this time, typical tech buying marketshare) it is pretty hard to make a living driving now. Sure anyone can get an Uber gig now, but the point is - driverless cars changes all that and removes a complete sector of service jobs.


"Going to eventually kill another whole (sector) of low end jobs that are something of a 'job of last resort' for (mostly) desperate men that can't find other work."

I worked on Automated Guided Vehicles ( AGV ) for a time. This discussion has to face the fact that proper sensor to servomechanism has a finite " reaction time " to assess a situation and respond to it properly. Before any AGV was allowed to be used, you had strict rules from OSHA about how long it would take to stop if any sensor was activated; kick sensors ringed the vehicle, while a special " forward bumper " would deform and stop the vehicle.
These Google Vehicles must have the equivalent of a supercomputer on board and up to 1 BILLION lines of code to operate independently on regular roadways. These robots ( yes, they still are autonomous robots ) may need years ( and some AI ) to handle any situation that the external ( and internal ) sensors will supply. The " bugs " could be lethal at proposed highway speeds. The occasional crash or stop means the program or computer still has problems.

This situation was supposed to be resolved for jets using the new GPS guidance systems. One company, ( B ) said that a certain new jet " could takeoff and land using it's autopilot and GPS " and the flight crew wouldn't even handle the controls ". You see what has happened with that boast...

This trial and error program debugging using the customer may work fine for Bill Gates but I don't want it on our nation's roadways. I'd rather see debugging done on a test track ( like we did on our AGVs ), especially when thinking of allowing higher speeds for a vehicle....Now if Google wants to make a vehicle that can operate at NASCAR speeds...

A great article. I'm currently looking at the legal ramifications of driverless cars, and who could be considered as an expert witness in litigation cases (database at www.witness.net). Let's wait for the impact!