Review - Carmageddon

:date: 2024-03-29 21:29 :tags:

I get my "news" by reading last quarter's copies of The Economist. They recommended this book and its author, Daniel Knowles, was their Africa correspondent for 3 years, so I thought I'd check it out. The premise of this book is that cars are bad. That message is preaching to the choir for me, someone who has been fighting the evils of cars for over thirty years. While my life's work to save humanity from an infestation of parasitic cars has generally been a failure, it does look like some people are slowly catching on that the problem is actually a problem.

The idea that cars might be more enslaving than liberating is now starting to find some footing for a slightly broader audience. Popular and righteous YouTuber CityNerd has a good video on the topic which normal people are encouraged to watch. Nerdier people are encouraged to read the paper he cites: Car Harm: A Global review of automobility's harm to people and the environment (Miner,Smith,Jani,McNeill,Gathorne-Hardy) If you're really interested in the topic, you can do all that and read this book.

After talking about how everybody appreciates the exhilaration of driving fast while unmolested by other idiot drivers, the book offers this description of The Problem.

I feel like this is a bit muddled, but really, where do you start? It's like having the premise that genocide is bad  —  where do you even start? With the obvious or the subtle? The book goes on to haphazardly hammer away at various aspects of car culture's awfulness but that doesn't make any of the arguments wrong.

On page 11 he comes up with some research that puts a dollar amount on commuting: apparently $40k is the pay cut people will take to not have to do it. (Is this the US? The average 32min each way commute?) For me personally it's way higher! He mentions the car friendly awfulness of Houston which is no lie. If I'm reading page 88 right, it looks like he's saying (car produced?) air pollution kills as many people per year as a C19 pandemic. The book claims (p182) that in 2021 790k electric bicycles were imported to the US while 690k electric cars were purchased. That's kind of interesting if true. The author asserts (p164) that "right on red" is a dangerous American habit, but no statistics are given. I wonder, how is it much worse than the British roundabout he (probably rightly) lauds for being a relatively safe feature (e.g. T-bone crashes are rare in Britain). Anyway, the book is duly peppered with factoids like this.

On page 86 he brings up something I have no doubt is true. "'Buses are way cheaper'... But officials do not like them, for precisely that reason: Unlike other big construction projects, they do not generate big contracts with the possibilities for kickbacks." Now, I can vouch for this one myself because when my job was to build expensive computer installations, I had many discussions with people where I would tell them, "Hey, maybe you don't need a whole new cluster of 500 computers; perhaps we could optimize your software to be more efficient." And they get an uncomfortable look and say, "Heh heh, ya, that's, uh, interesting, but look, our grant puts us in charge of $X_million for 'computers' so let's build that cluster, ok?" Or, they say, "Hey we need a $X_million computing installation." And I say, "Oh, great, what do you need it to do?" And they say, "Cost $X_million to increase our budget." The idea that traffic engineering departments could cure a lot of their problems by not pissing away money on massive projects that will have perverse effects thanks to induced demand is not what they want to hear. And, yes, induced demand is covered, just as it is in the 2008 book Traffic which I reviewed in 2017.

Chapter 10 basically covered Shoup just as I did six years ago. Essentially "free parking" is not free. He did have some fine examples of the madness of this form of bright red communism: 0.5 parking spots are required for every resident of a nursing home and a full space is required for every 500sqft of cannibis retail. Same with bars. Really the number one thing that can be done to make transportation less evil/car-centric is to introduce a crazy invention called the free market and price parking properly according to land value. By doing away with the totalitarian communist parking space requirements whose costs are passed along to society, a lot of the problem would go away naturally.

In Holland in the early 1970s their enviable shift to reasonable bicycle infrastructure apparently came from a campaign called "Stop The Child Murder". Sadly that's not something Americans are too concerned about if the country's stance on universal healthcare for children is anything to go by. Sorry, but I'm going to shame Americans at every opportunity until that's fixed.

Chapter 7 is devoted to the problems of electric vehicles. He offends Musk fan-boys by bringing up inconvenient facts like particulates are not improved with electric cars (indeed worse because of heavier vehicles), same with (tire) noise, if coal powers your grid they might release more CO2 than hybrids, the Congolese are getting raped, the grid really can't sustain everyone using it for transportation, etc, etc. Look, I can't see electric cars as any kind of panacea either but I think the author understands they have some potential to help somehow with something. Whatever. The whole electric car thing does seem like a distraction, but I'll leave that for other people to worry about because I don't really care.

What I do care about is the topic of chapter 8. This chapter is a poorly thought out critique of autonomous cars. The main argument seems to be based on name-calling, with the concept of computers handling the driving being mocked as "bionic duckweed" (which can be understood to mean "ridiculous boondoggle"). Now, I do not defend the stupidity and failure of imagination that runs the autonomous car business. There are serious problems in that industry. However, I think the author is pretty lame to pick on self-driving cars in 2023. Back in 2016, when I said there were serious problems with how things were being envisioned, well, that was a bit bolder. And 100% correct in hindsight. So with my better track record, let me set the author straight on some things.

He says with respect to autonomous cars, "Roads will supposedly become safer and more efficient." Supposedly? Bro, have you been on a public road? When that guy eating his Taco Bell and watching porn on his telephone is also trying to drive an enormous F150 what the fuck could possibly be less safe? For some reason a million deaths by human drivers is ok, but one death by a computer control is unacceptable. Well, since I'm personally at the front of the line for getting slaughtered in the street, and I can do basic math, I'm not too keen on that thinking. The author acts like he's the first person to think of the brilliant idea that everyone should use public transportation. People like me have been there and done that. Dude, not happening. Go tell all the junkies of the world to stop injecting heroin  —  you'll have better luck. The only way the problem is less problematic is to put car drivers in a cage where they can't hurt anybody. I think there is no fundamental technical reason why computer control could not be that cage. And rather than telling these junkies that they have to kick their habit, I feel like you'll have more luck telling them they can have 100x more of the texting in a single occupancy car that they're addicted to. But, hey, that's just my opinion.

He mentions that it is ridiculous to factor in autonomous technology in any kind of infrastructure decision making and long term planning. Why? How could this be sensible? He scoffs at the idea of a transformative technology springing into existence yet then contradicts himself by acknowledging that the internal combustion engine did just that. On p107 he admits SpaceX is no "bionic duckweed". I am certain that cheap sensors and parallel processing combined with astonishingly accurate neural networks are also no trifling minor detail. These are a big deal and how we take advantage of them is up to us.

He seems to not understand how AI works in vehicle robotics applications. He is concerned that it has to "rely on what scientists call 'machine learning'". He goes on to show his ignorance by imagining this as "they have to rely on the computer training itself, by making mistakes and being corrected. Trial and error." Ok, first off, how do you think human drivers train themselves? I'm afraid it's trial and error. But that shows a profound misunderstanding of machine learning as applied to vehicle robotics. There are many things a vision system can't do as well as humans. But does he realize there are many, many, many things that automated systems can do with AI voodoo that outperform humans by orders of magnitude? 24/7 without ever being drunk.

On p105 he brings up the Uber crash. (Which I covered in detail years ago here and here and here.) For fuck's sake man, this is a beautiful perfect example of why autonomous cars are essential: you can give a driver all the help in the world, pay them a salary to pay attention and they will still fail at their job. To blame a test system is bullshit. The designers knew it was a test system and put a failsafe human in the loop. The surprise wasn't that the system being tested failed when it encountered unfamiliar conditions; the surprise was that the human failed because humans suck at driving safety. While transportation safety depends on continuous human attention there is no transportation safety!

What he really seems annoyed with is using magical robo-taxi dreams to not confront reality properly. To an extent that's reasonable, but I also disagree with him because I would claim that some of that robotic technology is relevant to the discussion. He says, "...when self-driving car advocates argue that their rise will mean that technologies such as trains will be made redundant... they are almost certainly wrong." Let's just walk through this with a little more substance and nuance. Like The Economist would. Let's ask ourselves what are trains for? Why trains? What is a train? Why would you use a train? Well, there are two reasons. The first one is that you don't need to steer a train  —  it is guided by tracks. (Let's not even get into the weird, weird, weird situation where trains need "drivers" who can not really even do anything meaningful to control the train. This is usually a political/union thing and not a technical decision.) So the tracks mean you don't have to steer your vehicle  —  great. You know what other kinds of vehicles do not need steering? All of them! Today no vehicles need to be steered by a human. If you have a route you'd like to run like a train, you can make pretty much any vehicle run it. If I personally singlehandedly can make a dumb speedboat run a route within the width of a railway corridor, it is a solved problem. Right away, people miss the point and jump to conflate this technical marvel with not mowing down things/people that should not be run over. That is a completely separate problem. If you just can't get that problem out of your head, take it as completely solved the exact same way trains solved it in the early 19th century by clearing the tracks and maintaining a right of way.

Ok, so now we can see that you can put down a pair of steel rails or you can now put down two strips of cheap concrete pavers (or nothing, just grade it) and your vehicle will not stray from your planned path. What's the real difference then? The profound advantage steel rails have over other systems is that the rolling resistance is much lower. Sometimes this is the ideal parameter to optimize as when you are maximizing the amount of mass you need to move. Are you moving as much ore as you possibly can from the Australian interior to ports and processing centers? Steel rails is probably the way to go. But aren't you always optimizing for moving mass? No! Steel wheels (that match steel rails) are heavy. They take a lot of energy to accelerate up to speed. If your vehicle makes frequent stops, the case for steel rails is not so strong. Another situation where steel wheels may not be optimal is when the load demands are quite variable. For example, sometimes you need to move a full train and sometimes it's mostly empty. What kind of transportation systems make frequent stops and often have high variability in their loading? Well, the exact kind of public transportation networks that maybe should be thinking through all of the alternatives when making difficult-to-reverse decisions that will deeply impact the future. Oh and another consideration for high density transportation planning: rubber tires are incredibly loud and obnoxious under serious momentum loading but train wheels are worse! This is why you can hear a train at night 5km away. To scoff at vehicle automation as nonsensical just because Elon Musk has indeed been silly with foolish prognostications of his companies' miraculous products is its own kind of nonsense. Baby. Bathwater.

(Check out this fine video about train rolling resistance subtleties if you would like to understand the trade-offs better.)

It's frustrating that (p110) the author wants to lump "autonomous driving" with "stupid driving". I get it. Having everyone sit in their own autonomous car would probably create induced demand where the computer could economize the road usage (higher speeds, closer following, idealized planning, less dipshittery and crashes, etc). But remember that people trapped in computer controlled traffic doing whatever they want is way more righteous and correct than having them suffer in their cars while insisting they pilot those cars in a way that does not murder me. They can all burn in hell. The whole planet can burn in hell. My priority is not being murdered in the short term, thanks. But ya, definitely give some thought to the more holistic stupidity that has gripped our transportation Zeitgeist. Autonomous car tech, IMO provides a way to make a clean break from the old way and for pricing in the externalities properly. If someone wants a planet destroying ride in a private car because randos are gross, fine, but they should pay 10x for that. Correctly designed autonomous systems can provide that flexibility in theory. In practice? Who knows. The whole world is insane when it comes to transportation and I don't see that magically turning around any time soon.

Whew! Let's take a deep breath and relax now since that's the end of my rant about autonomous car technology not being properly given a chance to sensibly ameliorate the disaster of cars.

On to Chapter 11... Wow, that Volkswagen emissions thing was a real shitshow. Car companies are really just awful in pretty much every way. This book didn't even cover the auto industry's shitty record with safety cover-ups. The fact that politicians hold up (and bail out) auto makers as paragons of society and providers of "good" jobs is a ludicrous misallocation of societal good will.

On to Chapter 12... More fun trivia: apparently a Sherman tank (5.84 x 2.62 x 2.74m) was about the same size as a modern F150 (the longest that I know about is the SuperCrew at 6.17m). This chapter talked about the absolute madness of CAFE standards and how selling efficient cars means you can also then sell polluting ones in a global sense. Tesla apparently is cleverly subsidized by this somehow. (p153) "Before 2022, Tesla was making more from selling the right to pollute to other car firms than it was in profit from selling its own cars." But that's hardly a big win for the environment. It is pointed out that a clean car is good, but "if you buy a Tesla, or another electric car, you are not in fact taking a whole car's emissions off the road. Rather you are just giving somebody else a license to pollute a little more. As long as the CAFE standards remain as they are, individuals buying electric cars cannot do much to improve the overall level of climate change emissions, because the car industry will continue to aim [for] the overall target."

I found this quote on page 178 funny and wrong: "Safety concerns are why cyclists in cities in the United States, and indeed much of Britain, tend to be Lycra-clad young men who thrive on the adrenaline rush of weaving through traffic to get around." Let me stop you right there, Dan, and correct this. I live where I can watch cyclists all day long on one of the world's most magnificent bicycle highways  —  indeed I am doing so as I type this  —  and I can tell you that the Lycra tends to be correlated with age. There is a reason that MAMIL, a Middle Aged Man In Lycra, is a thing.

But it is far more important to note a bigger misconception in that quote. Even truculent fuckers like me  —  who have been very close to pulling assholes right through the safety glass of their stinking automotive cocoon to beat them to death with a U-lock  —  do not thrive on constantly being the victim of attempted murder. We actually fucking hate it. What you are really seeing when that "crazy" guy passes your car on a bike mixing it up in traffic like he has a death wish is the extreme end of the bell curve for defiance. He is Spartacus! That guy would rather be dead than a cowardly slave to everyone else's world destroying cars. How many bicycle mounted thrill seekers do I see running with cars where there is a perfectly sensible alternate route to safely ride a bicycle? Zero. It never happens. So that attitude is wrong. And (as mentioned on p180-181) if that crazy cyclist can survive, his life expectancy is going to be statistically way higher than that pathetic fat coward stuck in traffic in his monster truck emotional support vehicle.

And that's all I have to say about this book. Cars are shit. Cars are shit. Cars are shit. If that's kind of a shocking idea for you, maybe you need to read this book. But if you know me personally, well, you probably know that cars are shit and exactly why. I hope this book finds a new audience and does some good.