Wednesday, October 07, 2015

Atrios programming in some extra misses

Normally Eschaton is great, but a few things have been off. Duncan Black's repeated statements that self-driving cars won't happen (or will never work) is one good example, especially as they've been improving leaps and bounds in the last 2-3 years. He says it doesn't matter anyway until it becomes policy relevant, but cruise control+++ could already get people to drive instead of flying or taking rail.

I'm not sure whether the long-run effects are positive. I'm leaning in that direction as cities get more livable with less space taken up by parking and personal cars, but who knows. It'll definitely happen in Duncan's lifetime, and I'll bet he'll eat his words in about 5 years.

Another issue:  while I agree with him that Larry Lessig is going about things wrong, Atrios consistently pooh-poohs the influence of money in federal politics (and I haven't seen a lot of concern from him about money on non-federal politics). Based on personal experience I beg to differ about the role of money in politics, at any level. I have trouble seeing why the federal level would be different from state or local, and I'd be glad to get a solution at a federal level even if it doesn't solve everything. Everything or nothing isn't a good way to go about doing politics or policy.

We've got an over 50% chance of reversing Citizens United if we elect a Democrat in 2016, so I think a lot can be done about this issue.

In his defense, Atrios was right in not caring about Congressional earmarks, and I was wrong to oppose them. Spending on water projects has been a complete mess at the federal level since earmarks were eliminated.

Finally just a weird get-off-my-lawn moment where Atrios announces urban farms aren't farms, they're "commercial gardens" because....that's what he's said the words mean. As for being small, yes they're small, and intensive production can do a lot with small spaces.

Per usual, I'll write nothing about the vast majority of time that he's right, so I can concentrate on complaining.


John said...

If you are referring to legislation that reverses the Citizens United SCOTUS Inc ruling, we will need more than a Democratic president.

We will also need a flip in the majority in both houses of congress with 60 solid seats in the senate.

John Puma

Kevin O'Neill said...

Re: money in politics. I'm really not sure how anyone can believe that large dollar donations have no effect. First, the amount of money needed makes most federal politicians into full-time fundraisers. The influence of these donations may not be as impactful as some would believe, but they do have an effect.

Second, its money that legitimizes or delegitimizes most potential candidates in the eyes of the two political parties. I've seen firsthand good candidates that were interested in running give it a pass because the party was unwilling to support them unless they put $50k to $100k of their own money in upfront for a congressional race.

Third, the effect is even more pronounced at the state and local level where you don't have the same scrutiny by the media. State capitals are far more corrupt than Washington DC. It's no accident that the Koch's and their allies aimed ALEC at the state legislatures.

Anyone that believes money doesn't influence politics needs to spend a few months actually working in politics.

JamieB said...

Autonomous features in vehicles may have come on leaps and bounds recently and no doubt we'll see increasing automation but there are still some really big barriers to overcome before we see large numbers of fully autonomous vehicles on the roads so it's far from being a given in my view.

I'd expect to see autonomy in certain parts of the drive cycle, on motorways for example or in dedicated low speed urban vehicles, but full autonomy across the whole drive cycle and on all roads nationally? I'm going to remain sceptical for a while yet.

Also don't forget that the major car manufacturers sunk billions into fuel cells without being able to commercialise the technology.

Brian said...

John - legislation can't overrule Citizens United, although it can make improvements not ruled out by CU, like improved disclosure. The only thing that can overrule the case is a constitutional amendment (which I support but will never happen) or the Supreme Court reversing itself, which would happen if one of the 5 conservative justices is replaced by a reasonable one.

There are some variations/complications as to how the reversal would happen, but the quickest way would be a blue state passing a test law to challenge it.

JamieB - you're right that it could be a while before there's full autonomy everywhere, but I think self-driving will have a big impact before that, no later than either of the following happening:

1. higher highway speed limits, possibly dedicated lanes, for self-driving cars.

2. autonomous cars being sufficiently available and driveable within urban areas that people start owning personal cars less often than they would otherwise.

There's going to be a huge push for access to self-driving cars over the next decade as more of the Baby Boomers start aging out of their ability to drive themselves comfortably.

Brian said...

Meant to say "in a decade" not "over the next decade"

JohnMashey said...

1) Nobody I know thinks there will be a sudden shift to all-autonomous, instead features come in at the top and work there way down, such as lane-keeping.

2) Google has (at least) 3 kinds of cars:
a) Highway cars, seen around here for years.
I got a half-hour ride in one Summer 2014.
Driver got onto freeway, into lane, set a target speed. Car said "ready" and took over. It edged away from big trucks as it passed. Its radar could see past a big truck in front and notice a vehicle ahead of it braking, that was not actually visible.

b) Town cars, many more sensors and algorithms. Stop signs, lights, cross-traffic, pedestrians, bicycles. (Cats an small dogs take chances).
Have been around for years, but less than a).

c) 2-person pod car, 25mph, design headed towards no steering wheel or break pedals. Think Waze connected to the controls. Tested at Moffett Field for a few years, but I saw one in traffic at corner of El Camino Real and San Antonio 2 months ago.

a) As Brian noted, the elderly, as accident rates go up.
b) Teenagers
c) The blind (I've heard a blind guy talk about what a difference this would make to his life.)
d) "Take my kid to school and come back."

The real elephant in the room: insurance
If data builds up that these things are safer, especially for certain demographics, insurance companies may notice...

Donald Gisselbeck said...

So what happens when (not if) there's a bug, a cyber attack or some such that crashes a few thousand autonomous cars at once?

JohnMashey said...

The same things that happen non-autonomous cars, which really aren't cars, but computer centers with wheels. Anyone really worried about it should insist on a car with no electronics connected to brakes, drive train, gas pedal, steering wheel or engine. Entertainment system OK if not otherwise connected.

JamieB said...

I'm not suggesting that people are suggesting that there'll be a sudden shift*. And I definitely see the huge benefits of autonomous, electrified vehicles from the point of view of emissions (both GHG and AQ) and safety.

I'm suggesting that when talking about autonomous vehicles most commentators seem to be extremely confident that the very substantial barriers that exist between where we are now and full deployment will definitely be overcome.

If we get to full automation then most of the barriers melt away. It's the bit in between where it'll be a very messy collection of conventional vehicles and partly autonomous vehicles with different proprietary systems whose source code would be illegal to inspect (assuming current laws continue).

Human factors are one the biggest challenge before we get to full automation. When 80% of your driving is automated, what will your attention be like for the remaining 20%? What happens when you hire a car from an unfamiliar manufacturer? ("On the Toyota Infineon I have at home I have the Urbandrive option as well as the EZsnooze Motorway Friend but this is a Ford Spectrum so that means I'll have the Freeway Cruzemaster 5000 only so I need to recall the urban driving that I last did 5 years ago...")

I can't help but wonder whether autonomous vehicles as they're being described now might just become the colonies on the moon of the 21st century - we got a nice shiny space station to LEO and didn't go the whole way.

*I would argue that in urban centres there is a case for a pretty much overnight switch to low speed autonomous vehicles if they can be made to work. This avoids the messy mix problem.

Blogger profile said...

"The only thing that can overrule the case is a constitutional amendment (which I support but will never happen) "


Go to

Several states are already signed up to get a state majority which demands by the constitution that an amendment be voted for. Everyone said it wouldn't happen. It's happening because "unreasonable" people ignored them. Everyone said that Republican's would never vote for it (See John at the top there), but Red states are just as up for this change, and even Republican senators (never mind a large number, if not a majority, of Republican voters). It has cross-the-board support. What has stopped it and is making it go slowly is the "CERTAINTY" that it cannot happen.

"Unreasonable" people ignoring this BS and trying anyway, ignoring "received common knowledge", are getting it done.

Go to and volunteer to get your state on the list.

Blogger profile said...

" So what happens when (not if) there's a bug, a cyber attack or some such that crashes a few thousand autonomous cars at once? "

The same thing that happens when an alien invader comes in and takes over our society to make us slaves in their tribidum mines.

Remember, it's when, not if, it happens!

JohnMashey said...

"most commentators" is the problem.

It's like climate science: some people read about it in the newspaper, others study the IPCC reports, join AGU, go to conferences and talk to researchers.

Self-driving cars suddenly seems a brand-new topic. NOT.

In bicycle range for me is CARS, which has been doing this a while, although actually started long before that.
Note that Stanford Law School and other departments are involved.

See DARPA Grand Challenge.
2004 none finished, and videos were pretty funny
2005 Stanford won, CMU #2
2007 Urban, CMU #1, Stanford #2

See Sebastian Thrun.
He went off to Google to do cars.
He gave a great talk for us at Hot Chips (2008), including videos of the DARPA challenges.

When he left, (ex-CMU) Chris Urmson took over. See comments, like:
"Partially automating a car can reduce certain accident risks, but can also create new safety challenges not easily solved by current technology. Urmson, one of Rajkumar's former colleagues at Carnegie Mellon, said he worries that drivers could muff the handoff when an automated system suddenly demands they start making decisions about where to steer."

We have a current exhibit at museum.

Ignore commentators.

Brian said...

John Mashey has an excellent point regarding insurance. Partly-automated cars will be cheaper to insure, and the more automated, the cheaper. This trend will be self-reinforcing and pushed by the market without a need for a government policy change. It will also help move automation features down from high-end to cheaper cars.

John Brookes said...

Self driving cars are a certainty. In 30 years, kids won't believe you when you tell them that people used to drive cars, and they'll be amazed when you tell them the annual death toll from road accidents.

Deliveries will be an early application. And don't count on getting any work as a taxi driver.

After some time I don't even think people will actually garage their own cars. You'll just call up a car, and it will turn up. And it will be great for the elderly, who will no longer use their independence when they get too old to drive safely.