On Mechanical Turk, academic prestige, and what supply curves leave out
Third in the weekly memo series to Michael Vassar. The questions this week are mostly about labor markets — why companies hire whom they do, why some industries reward skill and others don’t, and why so much “common knowledge” about money and class is decades behind reality. Why don’t more businesses use Mechanical Turk to do various stupid, automated tasks? Mechanical Turk labor is ridiculously cheap by first-world standards (something like $1/hour as opposed to $10/hour for a base-rate first world worker), and doesn’t require large piles of paperwork — Social Security forms, immigration forms, tax forms, etc. It’s likely that most businesses don’t know that Mechanical Turk exists, which would obviously prevent them from using it, but then, you would expect it to grow extremely quickly, and it doesn’t seem to. It seems at least plausible that most businesses hire people for the sake of having a lot of employees and not for the sake of accomplishing more work; Paul Graham has told tales of startups, companies which are supposed to be young and innovative, hiring to fill in the gaps in some predetermined “org chart”. In which case, you wouldn’t expect to see a strong correlation between how many employees a company has and how much work they get done. If the main purpose of hiring isn’t to accomplish actual work, then why do middle-class people have so much pressure on them to spend more time in the office? Maybe spending time in the office is just some random thing that came to be used as a means of determining status, which then led to a death spiral, like education and credentialism. What are the super secret powers of HYPS professors? Within economics, at least, lots of professors get appointed to high-ranking government positions (FRBOG, Treasury Department, World Bank, IMF, etc.). Prestige within academia doesn’t seem to correlate highly in a lot of instances with prestige outside of academia (e.g., Ed Witten, who is the leader of the whole cult of string theory but whom nobody outside of physics has ever heard of). Wow — I just looked it up, and it turns out that “Edward Witten” only gets twice as many Google hits as “Eliezer Yudkowsky”. I’m not sure if this means that Witten is really unfamous, or if Eliezer is really starting to get actual fame, or what. One of the fairly obvious ones is that professors, more so at HYPS, have lots of perks to make up for the crappy nominal salary; you can never be fired, you get to do outside consulting, you get some amount of status with everyone, newspapers may listen to you, the university will pay for conferences and research equipment and students’ expenses, and so on. Silly anecdote: in one of my classes, after the lecture was over, I did some quick math and determined that the professor should be getting paid $250K/yr for what we pay in tuition (it was a small class). I then asked him if he made that much last year, expecting this to be an obviously silly rhetorical question, as no university has ever paid ordinary professors that much. He then replied that he did , which greatly surprised me, as RPI isn’t even that prestigious a place for getting other side jobs. Are there any lines of work outside of academia which are dominated by the very highly skilled? It seems that the structure of most job applications essentially prohibits this. So far as I know (being largely an outsider), the way it works is, at companies from McDonald’s to Google, you send in a list of your past accomplishments to the HR people or whoever. They then read it to determine if you have some chance of being competent or qualified. If you do, they invite you to the company for a day, where they test you, talk to you, see that you’re not crazy or whatever. Hence, you get selection enough to differentiate unskilled from skilled during the first stage, but the second stage differentiates “skilled and people there like you” from “skilled and people there don’t like you”, not “skilled” from “highly skilled”. Academia is different because the application process is different; there are few enough serious candidates each year such that they can form committees and review a person’s past work, references, etc. in detail. Therefore, you really do get very skilled physicists, mathematicians, and so on as professors. I think entertainment is, to a very high degree, dominated by people who really are actually good at what they do (in addition to having good social skills). This seems to come to a high degree from entertainers being contractors more than most workers are, which makes the labor market more fluid. This goes doubly for entrepreneurs, as Paul Graham pointed out. How long behind the actual truth is “common knowledge”? Depending on context, I think, several years to several decades. The whole “you can live on welfare” thing is a decade and a half out of date. The “Ivy League schools are expensive” thing is a decade or two out of date. The “America is one of the few nations in the world where you can criticize the government” thing is decades out of date. It seems like these stick around longer than average because they’re deliberately pushed as propaganda by different political groups: the first by Republicans, the second by people who say “going to an elite college isn’t worth it”, and the third by everyone including the educational system. “If you go to college, you can get a good job and start a middle class family” is several decades out of date, and it seems like it’s something that would have various interest groups pushing it — but, to be honest, I can’t think of who . Nobody ever ran an ad or published a book on how you can start your own middle class family. Maybe it’s just the next generation imitating the previous generation, which makes some sense, as the gap between generations is now about three to four decades, which is how long that particular bit of info is out of date. The thing that Paul Graham mentioned about art (“As usual the popular image is several decades behind reality. Now the misunderstood artist is not a chain-smoking drunk who pours his soul into big, messy canvases that philistines see and say ‘that’s not art’ because it isn’t a picture of anything. The philistines have now been trained that anything hung on a wall is art. Now the misunderstood artist is a coffee-drinking vegan cartoonist whose work they see and say ‘that’s not art’ because it looks like stuff they’ve seen in the Sunday paper”, Marginal ) is at least five decades out of date; presumably, in this case, the artists themselves are pushing it as propaganda, to make them seem like the oppressed underdogs ( Less Wrong: Why support the underdog? ). This happens in investing, but on a shorter timescale — “everyone knows” in 2007 that housing is a great investment, which is seven years out of date; “everyone knows” in 1999 that tech stocks are a great investment, which is five or six years out of date. How accurate is the standard supply/demand curve model of the free market? The first obvious adjustment, I think, is “stickiness”. When people can compare prices easily, competition pushes everyone down to the lowest level at which you can make a profit. When people can’t, or don’t want to for whatever reason, they tend to get an idea of “how much X should cost”, and stick with it, even if it’s available elsewhere for cheap. At one extreme is gasoline — since everyone posts their prices in big capital letters, it’s easy to see which one costs less, and so prices in any given area will differ by 5% or 10% at most. At the other extreme is stuff made for rich people. Since rich people don’t really care about cost, you would expect ridiculous profit margins on goods that are obviously luxuries, like “bottles” at clubs that are marked up 5x or 10x. Something which is even more obvious, and which I didn’t even think about because it seemed so obvious, is that the standard supply/demand model ignores the cost of production. Nobody is going to sell X for $9 when it costs $10 to make, no matter how much demand there is (unless they can exploit the demand in some other way by cross-selling or something). The car market is reasonably liquid, although not as much as gasoline or food (and there’s a lot of haggling with loans and trade-ins which they can stiff you on), but nobody sells new cars for under $10K, because you can’t make a car for under $10K. Markets for used goods, as long as they’re reasonably liquid, should follow the supply/demand model much more closely. Nobody will sell a textbook new for less than $30 or whatever, because that’s how much it costs to make. But if you’re selling it used , and you don’t need it anymore, you will take any price — you will sell it for $5 or $10, because $5 or $10 is better than nothing. And because there’s little demand for old calculus books as compared to new calculus books, you probably will sell it for $5 ( or even less ). Same thing with used cars. Crazy idea: people spend, not based on what they can actually afford, but based on what they are “supposed” to be able to afford. Hence, as middle-class living got more and more expensive, the gap grew between what people could spend and what people thought they could spend, and debt exploded everywhere. How could Renaissance possibly justify giving entry-level employees such low salaries ? Assuming they make $5B/year and they have a few hundred people, they could pay each employee an extra $1M/year, down to the janitors, and not even blink. The usual way for finance people to do it is to hire a disproportionate number of people at this level, and then fire most of them within a few years, which precludes paying them a lot, as they’ll then expect to get that much or more at their next jobs. It seems odd, though, that Renaissance would do that; it runs against what I perceive as the culture of the place, which is insular, secretive and a bit cult-like, not a two-year stopover for analysts on their way to greatness. Maybe it’s a device to prevent stupid people from applying. It seems reasonable that lots of people would look at this and think, “wow, people in finance really don’t get paid all that much compared to lawyers/doctors/middle managers, so I’ll just go to law school/med school/business school and do that instead”. Meanwhile, people who really care about finance or are smart enough to ask questions realize that they’re probably going to get paid $1M or more within five years and apply anyway. I wonder if I should apply at Renaissance after I finish undergrad — it seems plausible enough that I have the necessary credentials. I’d have great connections, a good salary, probably a great salary after a few years of experience, and it seems like a good work environment. Of all of the world-analysis topics one could write a book on, which would sell the best? One obvious heuristic is to write about things which affect people’s lives directly. E.g., a book on the history of Europe would be interesting to read, but literally orders of magnitude fewer people care about the history of Europe than about the subprime mortgage crisis. Maybe you could ghostwrite it and then get someone with a PhD or other high-level credentials to list themselves as the author. This is only a winning strategy if you care more about getting the ideas out there in the public view than the personal credibility and connections you’d get from writing a best-selling book. To avoid excessive competition, I think you would do better if you wrote about something that wasn’t already in the popular press. E.g., everybody and their dog will be writing about the economic collapse, the Obama presidency, subprime mortgages, and so forth for the next few years. It seems like a good field would be long-term secular trends — e.g., the creep of education and credentialism, or various demographic shifts — as these are harder for the press to run lots of stories on (why write about AIDS this month when it’s exactly the same as last month?). I really do think you’re allowed to use numbers and percentages, although not statistical analysis or algebra. Even if people don’t really know what they mean, they make it sound more “official” or something. Doubleplusgood if they’re numbers people can relate to: few people understand what a “billion dollars” means, but many more will understand what “$400 a month” or “$12,000 a year” means. I wonder what percentage of Americans have the math skills to construct a budget. It seems like a large percentage of people don’t; they just pay all the utilities and stuff automatically over the Internet or whatever, and then buy random shit with whatever money is left over until there isn’t any left. My dad is better than average in this respect, in that he runs a small business and knows that you have to make more than you spend if you want to be successful, although my mom is not. It seems like a fairly good idea for a book is to give people a basic understanding of the world that will allow them to be worth $10M by age 50. Things like: create a budget, save a large percentage of your income, don’t go into debt, don’t blow money on stupid stuff, learn basic investing (like tradeoffs of risk vs. reward), and that kind of basic common sense. It seems like someone must have tried this already, though — maybe the group of people who don’t understand stuff like this also don’t read books. For the record, it is a matter of perpetual irritation to me that, even if I had great interpersonal skills, a Harvard degree, and the diligence to work a hundred hours a week — in addition to the intelligence and rationality skills I already have — I could only make twice as much as an investment banker as the average semi-literate American who doesn’t know what the square root of 144 is. How can we make money off the incredibly massive arbitrage between the First World and the Third World ( Hacker News thread )? A fairly simple way to do it would just be to buy a whole block of housing in Poland or wherever, and then move there and live off the interest of relatively modest dollar-denominated savings. I wonder why more people don’t do this sort of thing for their retirement; presumably because they haven’t been told to. Will the law just become more and more complicated until nobody can really understand it? People have a large incentive to add things to the law, to benefit them or their organization or some favored cause. However, nobody really has any incentive to remove stuff from the law, which leads to things like LoonyLaws.com . The portions of the law which are actually enforced are limited by how much police officers and attorneys can commit to memory, so I can predict that, as time goes on, laws on paper will mean less and less (a lot of them don’t mean much even now).