Friday assorted links

1. New Tate Watkins eBook, market-oriented approach to understanding Haiti.

2. “I sought one ring to rule them all.

3. George Borjas on peer review and the recent brouhaha.

4. First fatality with a self-driving carThe NYT article has more detail: “Neither autopilot nor the driver noticed the white side of the tractor-trailer against a brightly lit sky, so the brake was not applied.”

5. Tim Page’s return from a traumatic brain injury.


Some people say vanilla ice cream is somewhat outdated and lame, but isn't it the best ice cream? Sorry, but it's just how I feel.

Dropping truth bombs today.

Needs hot fudge. And some nuts. And a maraschino to feed to your 4 year old daughter.

Beta imitates Better.

Apparently it's a great place to advertise.

2 was interesting, in part because it did veer in a weird direction at the end. Prestige, ego, status are part of it, but not all. There are a few reasons to believe things, and things being true is just one of them. We believe things that are psychologically useful, things that bind us to groups, to name just two. There are also reasons to reject things, a flip side of those. Things may be false, or psychologically harmful, or may divide us from groups we love. And so of course arguments can be spun for ourselves and for others, for and against, any number of things.

So rather than focusing on prestige, I'd say try to believe things that are true. Whenever possible, of course.

She said she loved me, then she left me. Sorry, but it's just how I feel.

Certainly a good strategy, but I think the article was not about trying to convince the reader to ignore prestige but rather laying out a campaign plan to get people in general to ignore prestige. I agree that in doing so the article veered in a weird direction. If people are focusing on prestige too much then sure it would be nice if they did less of that, but I'm not convinced of the urgency or importance of this campaign, and even less convinced of its ability to effect change or improvement.

If he can find a way to make people wiser, that'd be great, and I guess it is indeed a wise thing to focus on a specific aspect of wisdom i.e. don't over-focus on prestige, but it seems like an odd campaign.

There are also weird loops between expertise and prestige. Sometimes both are good. Sometimes one undermines the other.

Neither are of any value unless you study the results.

To quote Borjas:

"As is much too common these days, when important people do something wrong, heads no longer roll. Would anyone be surprised if any day now the people involved issue a generic non-apology apology telling everyone that it’s time to move on? As someone else famously said: What difference, at this point, does it make?"

Study results in the Tetlock/Superforecasting way? Sure. But I think it is also very common in today's society to wave away experts, and consensus of experts, as nothing more than hot air. People who do that are very bad forecasters, very bad analysts of our world. Their results show it.

It comes down to who gets prestige and who is considered to have expertise. There are indeed experts and there are indeed people who deserve prestige for their abilities and skill. Let results determine that. Or as a long forgotten way of doing things would, value experience.

Unfortunately if results mattered 3/4 of the people in positions of influence and power wouldn't be there. So of course it doesn't matter.

"Neither are of any value unless you study the results."

Precisely, which is why Hanson's position is so odd. If what we're really after is "truth" or "results" then trying to shoot down prestige is going after the wrong target, mistaking the sizzle for the steak. Especially if we believe anon's point that expertise can be correlated with prestige. I would add that contra Hanson, I don't see an over-reliance on prestige being a major problem these days. Certainly Trump and Brexit supporters have been quite happy to ignore the prognostications of the prestigious, and American anti-intellectualism has been observed since Hofstadter or for that matter de Tocqueville.

Did not the prediction markets favor "Remain" up until and including voting day? And on "affiliating with the prestigious," Rene Girard was two generations ahead of Hanson. Girard called it mimetic desire: we imitate our models. The solution, however, is not to rid the world of models, which in the case of RH would eliminate RH.

Sage advice - I'm placing it on my refrigerator right next to "Buy Low, Sell High".

I don't think there's much good in white lies, but few things are strictly black and white in the cost/benefit analysis.

4. We don't know what the driver saw or didn't see. But he didn't hit the brakes at all. Maybe he wasn't paying attention, having earlier been saved from a serious crash by the Tesla being on autopilot:

The truck was doing a left turn across a highway, probably illegally if the Tesla needed to slow down to avoid hitting it.

Bloomberg link is broken, btw:

This is apparently the first death in 160 million miles of autopilot driving, which is better than normal driver safety.

Note that its from the autopilot feature, not from a self driving car. Many newer models have this feature. The autopilot merely stays in highway lames and stops the car from slow collisions. My car has this. Yet the feature is only an aid, not a replacement. In fact, the feature turns off if it doesnt feel your hands on the wheel

Note that its from the autopilot feature, not from a self driving car. Many newer models have this feature. The autopilot merely stays in highway lames and stops the car from slow collisions. My car has this. Yet the feature is only an aid, not a replacement. In fact, the feature turns off if it doesnt feel your hands on the wheel.


Oops, it was 130 million miles, not 160 million. (And the US fatality rate is 1 death per 94 million miles.) I was still impressed that they've covered that many miles since autopilot was introduced less than a year ago. I guess I didn't internalize either how many Teslas were out there or how many miles a fleet can do.

You can't really compare the two. The price of a Tesla is a lot more than the 1984 Dodge Omni. Autopilot users are also very selective about roads, conditions, and so on. They are basically padding the stats in their favor, we don't have enough information to say for sure but I suspect reasonably responsible drivers have a much better track record in the long run.

There's a black box (data recorder) on every Tesla car. I think the affirmation that the driver did not push the brakes is based on these records. Same for the affirmation that the "car" did not see the obstacle. The point is, there's lots of data to be analyzed. This is not the typical case of "we'll never know".

For the statistics. There's a strong correlation between car age and fatality rates. Tesla fleet is very young, it shouldn't be compared with the average miles/fatality ratio.

The driver could have been totally zoned out and not paying attention, instead of "was paying attention and just couldn't see because of the angle of the sun." I want autopilot to succeed, but the latter condition there is just wishful thinking of "well, a person couldn't see it either, so our autopilot isn't that bad."

You're right .The Sun is not important here since the car has radar sensors. Why the radar couldn't see the obstacle is missing on today's press releases. If radar failed, that's an important bug.

If the Tesla uses lidar, the brightness and angle of the sun can be important.

I read a, no collision avoidance system is designed to work above 70 km/h (40 mph). It works with radar in MBs, Teslas or Toyotas. At highway speeds you're on your own.

Thus, no hardware or software bug. The fail was in the meatbag that believed the car had a feature that simply does not exist.

What I've seen so far among the Tesla fans is that the Tesla saw the truck as an overhead sign, because of how shiny it was. If the car was coming up the hill, and able to see road underneath the truck (because there was no skirt below the truck, which is its own debate), it thought it was a sign well over the road.

We don't know for sure yet, but this seems like a good theory.

How well equipped with liability insurance is Tesla these days?

4. I've jumped the gun and given my opinion already, in earlier threads. I won't repeat that, but I'll add this link again, good internals on how it is all suppose to work:

IMPORTANT: the car thing was just from the autopilot feature of the car. Many new cars have this. The new Subaru ls have this. It was not from a self-driving car. Please edit your words

The Subaru has emergency self-braking, and never promises nor even suggests that you suspend responsibility. You are supposed to brake before hitting the truck. If you don't, the self-braking systems will try to save you.

They had to make it less automatic after people were posting videos of turning on the autopilot and then climbing into the backseat. Maybe Tesla could turn a blind eye to that, but not after someone posts it publicly.

I'm still optimistic about the tech. It can go almost 50% longer than a human without a fatal crash. And it's only going to improve.

I think you define the crux two ways. First, intentionally or through association "autopilot" was over-sold. Second, some poor souls are trusting these things because "they believe in the technology" in general, and not in the specific. They might be self-driving now, because they believe it will really work in 10 years.

Both Tesla and its buyers are more aggressive than the rest of the industry. Which is one reason Tesla has over 100 million miles of autopilot experience.

Come on, think about your actual argument here. A truck was across the highway and "neither the autopilot nor the driver saw it."

Do you seriously think the driver was looking?

Do you seriously think Tesla level autopilot is ready for a driver who is not looking?

These two questions put "autopilot" in the range of a cruise control, something you can use to reduce driver action but perversely still requires attentiveness. In that sense it was literally an accident waiting to happen.

I have driven tens of thousands of miles with cruise control and not crashed. Woo hoo. Cruise control makes driving safer.

As I said above, I really doubt the truck wasn't visible to the driver. He was just not paying attention. Maybe had his eyes closed. I hope so for his sake, since he likely died via decapitation.

And let's consider a counterfactual for a moment. If the driver had been attentive, as other auto-pilot or cruise control users might be, he might have pressed the brake and lived.

Now here's what is dangerous to me. If he had pressed the brake and exited auto-pilot, it would have improved your "safe miles driven." Those actually include attentive drivers taking control.

Thus those miles are less "self-driving" than many casual observers might assume.

Now here’s what is dangerous to me. If he had pressed the brake and exited auto-pilot, it would have improved your “safe miles driven.”

This is a really good point.

The lawyer for the truck driver claims that the Tesla had a DVD player that was turned on to a Harry Potter movie.

Naturally they trusted the experts - you're not a climate change denier, are you?

#2. Prestige. Without it the modern art market wouldn't exist. The prices obtained for pieces of art today is almost entirely disconnected from its underlying aesthetics.

Scott Sumner tells me I dislike contemporary art because I 'don't understand it'. It's status games all the way down.

Some people collect Toy Story dolls, some people collect Van Goghs. Some people in these groups are basically doing similar things from the investment and aesthetics perspectives, except that obviously they have different views on what exactly contributes to those things.

2. Sounds like he is declaring war on a certain segment of The Elite. I support him. We need contrarians to question the established order in all fields.

Academia is certainly prestige-obsessed and I think it gets in the way of both its teaching function and its pursuit of knowledge function.

Law is also prestige-obsessed, and it attracts a lot of prestige-strivers. Part of the problem there is that prestige is gained based on where you went to school, who you clerked for, the firm you worked at, rather than the actual outcomes you produce.

He's declaring war on regulatory capture. But like so many of today's kids, he thinks he's the first person to discover it.

Oh, he's 56 years old. Then he really should know better.

Given the basics of human psychology, I suspect the best way to attack prestige is to gain lots of it and use it to undermine everyone else's.

Gain lots of prestige, then come out as a fraud.

You'd think Alan Sokal's prank would have persuaded Duke University Press to close down Social Text. Still published, as far as I'm aware.


Borjas: This scandal is an embarrassment to the economics profession.

Tyler: I can't get excited about it.

4. I drive through Williston, Florida (the place of the accident), from time to time and it's very hilly (especially for Florida), the roads have lots of curves, and there lots of large tractor trailers carrying material mined from the quarries located in the area as well as tractor trailers carrying cows and horses. It's a dangerous place to drive in any car. For those who wish to find it on a map, Williston is about 20 miles southwest of Gainesville (the location of the Univ of Florida) and 20-30 miles northwest of horse farm country.

3. Oh my. One of the most entertaining kicking of an exquisitely manicured ass that I have read in a long time.

Here is the sad thing. As much as you might want to wash your hands from consensus, you cannot. All human knowledge works by successive consensus. We hope that there are two steps forward for every step back, but that's enough for progress. That is enough that knowledge, by consensus, improves.

If we actually did as Borjas recommends below, there would be no progress, just a viper's pit of conflicting opinion:

Let me get back to the pet peeve that motivated this long rant. Next time you hear about “professional consensus in peer-reviewed research,” do as I do: Roll your eyes. Who knows what went into the making of that particular sausage? And this warning applies ten-fold for peer-reviewed research in any politically charged subject."

Oh, bullshit. If there is conflicting opinion, then it is simply opinion and no more weight should be given to it.

This seems to be a case where someone copied someone else's research and got it published through connections and sloppiness in a prestigious journal. Either the subjects under discussion mean something and it is important to get it right, or it is meaningless status games where everything that is said can be ignored as the default, and considered from that basis.

We all consider the opinions of people we respect when considering what is important. When we find out that those who we are encouraged to respect are acting in a cavalier and self interested way, they discredit the whole enterprise.

You aren't doing Clinton anon campaigning are you? A flawed, seriously flawed person who made a large number of very serious lapses in judgement should be president because of her prestige and authority?

it is meaningless status games where everything that is said can be ignored as the default, and considered from that basis.

Welcome to academe.

Clayton Cramer has said his tangles with journal editors ca. 2000 persuaded him that (1) there is a corps of people in American history who are willing to accept academic fraud if the thesis is to their liking and (2) they may not be all that numerous, but the number is non-zero and some of them hold gatekeeper positions in the profession. He named names: the man who was editor of Journal of the Early Republic at that time.

Just today, I encountered another news story about a student club being witlessly harassed at a state university, consequent to which there were demand letters from legal counsel, consequent to which there was a perverse and recalcitrant response from college apparatchiks, consequent to which there was an injunction obtained, consequent to which the campus flak issues a a statement denying they'd ever done the thing which they'd just been enjoined from doing. You can never figure out if casual mendacity or utter incompetence is behind this sort of thing. There are quite an army of deeply mediocre people in academe.

The padding. The nonsense courses. The political patronage. The haphazard curriculum. Games games games.

When there are conflicting opinions, often one does and should carry more weight.

With zero other information, would you guess that you would agree with "peer review consensus opinion" in a case where 99% held the same opinion? What about 50% and the remaining half being equally distributed among 10 other views?

In my opinion there are no Martians on earth. Should the conflicting opinion receive equal weight?

The value of consensus opinion in a field is very different if you're a complete outsider who can't independently evaluate the opinions, or if you're a complete insider who knows the factions and blind spots involved in that consensus opinion. Sometimes, the consensus of experts is 180 degrees wrong, but it's pretty hard to know that (even if you suspect it may be true) when you don't know the field well yourself.

I mean, suppose the basic picture of the world of cosmology is fundamentally screwy because of politics and faction-building and grudges in the field. How would you, as a complete non-expert with little background beyond maybe a physics class or two in college, even begin to untangle who was right? By contrast, someone who did a PhD in cosmology might have a very clear idea what's going on.

One difference is that the minutiae of theoretical mathematical or physics disagreement that is well beyond a layman's grasp has essentially 0 impact on their life. The nerds can have that argument.

When one side consistently argues 2+2=5, you can guess which side is right, no matter your expertise on the subject.

Related note: "Michael Mann, scientist: Data ‘increasingly unnecessary’"

What do I see.. a couple of experts disagreeing with each other... what a conundrum!

Whatever shall we do now? Who's right? I don't know because I'm not credentialed enough to make a determination. But I'm fully confident they'll come to a consensus that is reached by pure rationality (#Rationia) and not political maneuvering or social signaling or anything else with a scent of irrationality.

All human knowledge was gained by smashing prior, incorrect consensus.

Anyone who is actually interested in advancing knowledge within their field should approach every consensus with skepticism. Even if you have to provisionally accept a consensus for the sake of argument and build something based on it, you should still be willing to question those underlying assumptions in the future. Building a superstructure on a foundation of flawed consensus isn't just a waste: it may harm the field by preventing further advancement until those whose prestige depends on that superstructure die off, and it may harm people generally if it is used as the basis for public policy.

To me it is funny. I see successive consensus. Some people only want to talk about one moment in the repeated cycle.

You can't have "smash" without a new consensus proving it.

Consensus of what? That to advance your career without much effort you find some interesting paper in an obscure and unconnected field, get someone you are connected with to referee and review your rewrite, then hope that the plaudits from having published in a prestigious journal gets you a plum appointment and hopefully tenure?

I suspect that there is consensus on that. And the consensus is becoming that published papers are simply a backscratching exercise with occasional real work that slips through by mistake.

And another consensus seems to be arising that it is plausible that in any field 50% or more of the published papers cannot be replicated.

Eventually the consensus will form that this is all untenable, and it will either collapse, be dismantled or end up being a subsidized mediocrity while serious work happens elsewhere. People who previous to this realization recognized the state and argued against it will be applauded.

So your argument will be dug up, and the consensus will be that you were a fool and somehow looking for some status marker that worked at the time and were uninterested in science and discovery or even understanding.

The climate deniers really did a number on people. They managed to flip it, so that if you do have learned and talented people with long history in a field who agree, that has become an indicator that they are wrong.

Or maybe there are not currently any strong counter-arguments. No, the common belief now is that strong-counter arguments are not needed, and consensus itself is sign of error.

I'm sure you can trace the money on that one.

You should regularly question consensus. But whatever has been passed on and/or commonly accepted for years, decades or centuries should at least be given a fair hearing.

Examples of social advances in the 20th century where questioning centuries of practice have been (or likely will be) are not so rare in the last few generations. I understand that a good way to make a name for yourself in academics is to do something new that might break a previous consensus and be more correct about something. But at the same time, a willingness to question consensus should not fall into the habit of assuming that you're more correct than the consensus.

Often, certain parts of the "consensus" are agreed upon, but there remains significant disagreement about a variety of details, and this holds true in both social and natural sciences at their respective levels of advancement in a diversity of areas of research.

I think the main thing is to not fear "looking dumb" by questioning consensus. It's not that difficult to say something like "Here are a few things that don't add up for me. Why don't you try to convince me? I'm simply not convinced, but none of the alternative explanations seem better to me."

2. Oh my. Someone setting up for the mother of all kicking exquisitely manicured asses!

The whole Brexit schmozzle was extraordinary entertainment. It could be improved with the rendering of garments, considering all the fine looking people doing the potential rendering.

How could this be encouraged?

2. “I sought one ring to rule them all.”

So, does the Straussian reading imply that his article is really about Brexit?

Here's why computer-driven cars will be almost infinitely safer than human-driven cars. After this one fatal accident, Tesla people thoroughly analyze the cause. Let's say it's a low sun angle. They potentially upgrade software and/or hardware, and the entire fleet gets safer. Contrast that with human-driven cars. A single human gets in a fatal accident caused by low sun level. Does the entire fleet improve? Not a bit!

After 1000+ computer-driven vehicle fatal accidents, computer-driven cars will be much, much safer. After 30,000+ fatal accidents every year, human drivers are no safer.

I would suggest the opposite. In response to this, a realization that a false sense of security was engendered in the drivers. A mixture of regulatory, liability and technical responses will end up implementing annoying techniques that while taking control of more of the vehicle acts to keep the driver engaged and involved. A mixture between the signage around intersections and the notifications you get from the IRS. Someone will reverse engineer the Tesla source code and find that these specific software routines are called 'hillaryclintonmode' and there will be a national political crisis where drivers, engineers, and enthusiasts will be demonized, the end result that Tesla goes broke.

The net result will be the importation of Cuban 1950's vehicles, restored to stock condition and driven with joy.

Litmus test failed. A technical error has become a political fantasy for you. What else can't we trust you on?

"5" ... Do not self-medicate and self-diagnose your first-ever set of sudden onset migraine headaches. While the whole essay is admirable, I am sure that Tim Page hopes that some readers of the essay take to heart the good advice near the beginning... Do not self-medicate and self-diagnose your first set of sudden onset migraine headaches. They could be a warning sign of something fatal. I had not read anything new by Tim Page in years and, as these things happen, had not wondered why. Now I know. He has always been a fascinating writer on music and on several other subjects.

3) Whatever, maybe. But even if it's already in some medical literature, if it's relevant, why not have it also explored in the economics literature. Worst case scenario, why not just add a thing that says "it has come to our attention that, in a different field of study, medical applications of this dataset reached not unsimilar conclusions in MMM YYYY. We believe that the entry of this method, datasets, and potential relevance of the findings are of further merit in the publication of this article, in bringing to the attention of ___________ subfield(s) of economics.

It wouldn't really be an admission of any wrongdoing, but people could see for themselves, presumably more so as time passes. I dunno, is it lucky for them or unlucky? It doesn't seem that strange to me that someone will publish something relevant and find that someone in another field had done something similar. I've done lots of multidisciplinary work, and it seems to me that, most often, the first order of business is to do lots of different kinds of keyword searches, including by trying to secure 5-10 minutes with people from other fields just to take about the project and see whatever keywords/terminology they might put you on to (which might/probably lead you to an existing area of similar research in another field). After all, it is a lot easier to place you work in an existing body of literature than to do something "completely new". And, it's also sort of expected that you should poke around in some different fields to make sure that you're aware of stuff like that. At least a bit, no? If it's "too similar" but worth publishing, why not reach out to the other journal and see if they can just publish the original article in the economics journal and some economics would have some exposure to different ways of treating similar things?

I guess the real question is whether or not the authors already knew of the existence of a similar literature. Hard to tell. Should the editors have known either/too?

Specialists cannot be expected to be in the 1-2 years cutting edge awareness of 100s or 1000s of different fields. I wouldn't concern myself with it that much except for a) they should at least openly point to the fact that someone's done something similar in a different field, and b) that whole thing with the editor's conflict of interest is a big deal - if it was just a big dumb poorly thought out mistake, imo she'd better be preparing the apology of a lifetime, because otherwise how does AEJ protect its reputation if the apology appears in any way BS or no heads roll?

Not a good situation for any involved, it would seem.

Yes but in this case the paper would not be Aer material. Poor editorial job and handling of the case.

Comments for this post are closed