Pessimism of the Intellect, Optimism of the Will     Archive   Favorite posts   Twitter: @steve_hsu

Thursday, June 30, 2011

Please spit

Earlier today I signed an order for $50K worth of these things...



OG-500 DNA Tube

Overview

The OG-500 DNA Collection Kit provides an all-in-one system for the collection, stabilization, transportation and purification of DNA from saliva.

The OG-500 format is ideal for customers who require a compact format for collection via the post and those seeking a solution that is compatible with high-throughput processing systems and standard storage racks.

The new gatekeepers

The Chronicle has a nice article on what it takes to run admissions at a selective college. The requirements of communication skills (e.g., to address concerns of multiple stake holders) and ability to develop "data-informed strategies" apply to almost all high level leadership positions these days, whether in finance, business or academia. Perhaps we'll eventually see a squeezing out of High V, Low M types as leaders. (Fat chance!)

I suspect these numerate admissions deans know many things that can't be discussed in public ;-)

For how SAT scores predict college performance, see here and here.

For a description of Harvard's "holistic admissions" policies (adopted in the 1950s), which have become a model for other elite schools, see here. Then dean of admissions Wilbur Bender may not have been very quantitative, but he had excellent intuition about How the World Works. (See also here.)

'Numerical and Verbal'

A profession that once relied on anecdotes and descriptive data now runs on complex statistical analyses and market research. Knowing how to decipher enrollment outcomes is a given; knowing how to forecast the future is a must. Which students are most likely to apply, submit deposits, and matriculate? At what cost to the college? How likely will they be to graduate? Such questions echo in the modern enrollment office, which is often supported by one or more institutional researchers, as well as consulting firms that sell recruitment strategies in various flavors.

Search the job listings for top-level admissions and enrollment openings, and you will find that many colleges seek a "data-driven" leader, someone who will develop "data-informed" strategies. This past winter, for instance, Pomona College, in California, began a national search to replace Bruce J. Poch, who had stepped down after 23 years as vice president and dean of admissions. Among the qualifications listed in the job advertisement: "an ability to analyze and use data to guide decision-making and measure results."

David W. Oxtoby, Pomona's president, led the college's search committee. The modern admissions dean, he says, must have a "technical, quantitative facility," the ability to delve into the relationship between a student's SAT score and her subsequent performance in college, or why some kinds of students are more likely to enroll than others. Moreover, Pomona had decided to merge its admissions and financial-aid offices (a change many colleges have made already). So the new dean would need to speak the language of costs.

That's not to say anyone wanted to hire an accountant. Numbers, Mr. Oxtoby says, have not diminished the importance of communication skills. Pomona's search committee sought someone who could articulate the value of a liberal-arts education, and relate to faculty members. The "ideal" candidate, the job listing said, would also know how to talk to the news media; today's admissions leaders are also public-relations specialists with loud microphones.

"The job is numerical and verbal," Mr. Oxtoby says. "It's still all about relationships. You still need a sense of that person on the other end of the admissions process. If you lose that, you just become another technocrat, and you've lost the reason why you're doing this job."

Pomona interviewed a dozen candidates before hiring Seth Allen, dean of admission and financial aid at Grinnell College, in Iowa. Soon to occupy one of the premier jobs in admissions, Mr. Allen, 43, represents the next generation of enrollment chiefs. They've ascended during an era of high competition, learning how to market their colleges and massage the metrics that define success in admissions.

Although idealism may inform their work, they are clear-eyed realists. They are not introverts, for they must collaborate constantly with faculty members and other campus offices. They are diplomats who must manage competing desires: those of administrators who want to enroll more first-generation and low-income applicants, professors who want more students with high SAT scores, trustees who want to lower the tuition-discount rate. "Twenty years ago," Mr. Allen says, "there were not as many wants."

Drawn to statistics at an early age, Mr. Allen earned a bachelor's degree in economics at the Johns Hopkins University in 1990. He first worked as an admissions counselor for his alma mater, a cutting-edge laboratory in the then-burgeoning science of enrollment management. Mr. Allen learned how predictive modeling could project net tuition revenue, how many biology majors would enroll, and a hundred other outcomes.

Tuesday, June 28, 2011

Big pimpin'

You may have read recently about physics professor (Farleigh Dickinson University, NJ) David Flory's arrest in connection with a web site promoting prostitution.

What I don't understand is how sites like Seeking Arrangements (previous posts here and here) avoid this kind of attention from law enforcement. The fact that the arrangements are negotiated privately between users of the site may insulate SA, but surely the police could nab plenty of Sugar Babies and Sugar Daddies in a sting operation.



From physics to Goldman to Y Combinator

Better to die free than live as a slave, or something like that. The author describes his path from physics grad school to Goldman to a bay area startup. Via maoxian.

Why founding a three-person startup with zero revenue is better than working for Goldman Sachs

I joined Goldman Sachs in 2005, after five flailing years in a physics Ph.D. program at Berkeley.

The average salary at Goldman Sachs in 2005 was $521,000, and that’s counting each and every trader, salesperson, investment banker, secretary, mail boy, shoe shine, and window cleaner on the payroll. In 2006, it was more like $633,000.

In the summer of 2005, I took one look at my offer letter and the Goldman Sachs logo above it, another look at my sordid grad student pad, and I got on a plane to New York within the week. I packed my copy of Liar’s Poker for reference.

My job on arrival? I was a pricing quant on the Goldman Sachs corporate credit trading desk1. We traded credit-default swaps, both distressed and investment-grade credit, and in the bizarre trading experiment assigned to me, the equity part of the corporate capital structure as well.

There were other characters in this drama. The sales guys were complete tools, with a total IQ, summing over all of them, still safely in the double digits. The traders were crafty and quick-witted, but technically unsophisticated and with the attention span of an ADHD kid hopped up on meth and Jolly Ranchers. And the quants (strategists in Goldman speak)? Mostly failed scientists (like me) who had sold out to the man and suddenly found themselves, after making it through two years of graduate quantum mechanics, with a bat-wielding gorilla peering over their shoulder (that would be the trader) asking them where their risk report was.

... The sad truth is: quants were the eunuchs at the orgy. We were the ever-present British guy in every Hollywood WWII film: there to add a touch of class and exotic sophistication, but not really matter much to the plot (and maybe even conveniently take some bad guy’s bullet).

But things weren’t all bad! At its best, when the markets presented an apocalyptic Boschian landscape of damned souls torn asunder by hellish tortures, every Goldman grunt, sergeant, or general would close ranks and form a Greek phalanx of greed. Unlike almost every other bank on the street, Goldman could actually calculate its risk across desks and asset classes, out to five decimals [see footnote].

... What’s work like now? Writing code. Worrying about everything from our credit card billing to the pile of dirty dishes in the sink that will give us all diptheria some day. Writing linkbait blog posts to get us free PR (like the one you’re reading now). Schmoozing with investors, and playing the junior high school popularity contest that is startup funding. Keeping jealous tabs on other startups to see how they’re doing compared to us. Trying to put myself in the mind of our users to make something they’d want. Oh, and launching…finally, good God…launching.

You see, starting a product from an empty text buffer is very different from keeping a well-oiled money-machine running8. I’ve had apocalyptic fights with the other founders that almost ended in fisticuffs. I’m watching my four-month-old daughter grow up via Skype. These jeans I’m wearing will likely fuse with my skin at some point if I don’t take them off. I haven’t seen a paycheck or a loving woman in much too long.

You know what I regret most though, going from Goldman to this?

Not having made the switch earlier. ...


[Some footnotes]

It’s a somewhat different story on structured credit desks, where numerical modeling is perceived to be more important. Also, on algorithmic trading desks, where the quant writes the code that does the trading, and the sometimes blurry line between quant and trader basically disappear.

At the risk of getting sued, let me throw you geeks a bone and part the Goldman veil a bit. The Goldman Sachs risk system is called SecDB (securities database), and everything at Goldman that matters is run out of it. The GUI itself looks like a settings screen from DOS 3.0, but no one cares about UI cosmetics on the Street. The language itself was called SLANG (securities language) and was a Python/Perl like thing, with OOP and the ORM layer baked in. Database replication was near-instant, and pushing to production was two keystrokes. You pushed, and London and Tokyo saw the change as fast as your neighbor on the desk did (and yes, if you fucked things up, you got 4AM phone calls from some British dude telling you to fix it). Regtests ran nightly, and no one could trade a model without thorough testing (that might sound like standard practice, but you have no idea how primitive the development culture is on the Street). The whole thing was so good, I didn’t even know what an ORM really was until I started using Rails and had to wrestle with ActiveRecord. The codebase was roughly 15MM lines when I left, and growing. I suspect my retinas are still scarred by the weird color blue SecDB was by default.

If doing a startup is like rolling a boulder up a hill, then working at Goldman Sachs is like rolling it down the hill: you just have to stay out of the way of the boulder.

Sunday, June 26, 2011

Machines of Loving Grace

"The story of the rise of the machines, and why no one believes you can change the world for the better anymore."



This is episode 3 of the BBC documentary series All Watched Over by Machines of Loving Grace, featuring evolutionary theorists George Price (@19 & 34 & 42 minutes in) and William Hamilton. Also making appearances: Dian Fossey (@27 & 31 & 40 min), John von Neumann (@21 min), Richard Dawkins (@44 min), gorillas, chimps, Hutus, Tutsis, imperialists, mercenaries, and a mercenary (sociopathic?) mining company CEO (@7:30 min; good Hamilton gene stuff follows). Fossey is portrayed as a nut case; some of the details (e.g., about the death of her favorite gorilla Digit) do not match what is in her Wikipedia entry.

Bonus points to commenters who can classify the individuals by their V, M, and empathy scores ;-)

The episode describes the rise of the idea that humans are merely automata, sometimes altruistic and sometimes murderous, programmed by their genes. This perspective, it is claimed, is appealing because it absolves us of responsibility for terrible unintended consequences of our actions, such as the Rwandan genocide.

The logic of Machines of Loving Grace is sometimes suspect, but nevertheless it's a highly stimulating series.

Higher education bubble?

What are the economic returns from a college degree, net of individual ability? Does college add value, or is it mainly a signaling device (e.g., for intelligence and work ethic)?

Results from two new studies are discussed here and here in the Times.





The obvious adjustment that one would like to see is discussed in the comments by one of the researchers (Stephen Rose of Georgetown; second link above):

... smarter people do go to college therefore the results are biased to show that BA graduates make more than those with a HS diploma. There are education data sets that do have a "skill" measure along with earnings and degrees. Using the best statistical approaches reveals that the educational premium is still large after adjusting for differences in ability (the premium declines by about a third).

It would be nice to see this analysis. Does "ability" mean IQ? College attainment also filters for conscientiousness (does it build conscientiousness, or just select for it?), which may account for another chunk of the income premium. But I would guess there is still some added value left even after these adjustments. If half of the effect in the figures above remains after controlling for ability and conscientiousness, then college is a worthwhile investment in purely economic terms (e.g., a lifetime income boost of $300k vs tuition and lost income of about $100k).

Saturday, June 25, 2011

East Asian sociopaths?

Some would assert that CEOs and other people in leadership positions are often warm sociopaths. Interestingly, it is claimed that there is a huge variation between groups in the rate of sociopathy. Perhaps this is related to the under-representation of E. Asians in leadership positions in the West, despite their high educational achievements? (Instead of sociopathy other factors like aggressiveness in interpersonal relationships might play a role.)

The Sociopath Next Door: ... Though sociopathy seems to be universal and timeless, there is credible evidence that some cultures contain fewer sociopaths than do other cultures. Intriguingly, sociopathy would appear to be relatively rare in certain East Asian countries, notably Japan and China. Studies conducted in both rural and urban areas of Taiwan have found a remarkably low prevalence of antisocial personality disorder, ranging from 0.03 percent to 0.14 percent, which is not none but is impressively less than the Western world's approximate average of 4 percent, which translates to one in twenty-five people.

In this Salon interview, the author attributes this group difference to cultural factors. Her answer is rather odd considering that in the book she describes personality, and sociopathy specifically, as roughly 50 percent heritable. As usual, we're allowed to admit that things like g and personality have a genetic component, but not allowed to entertain the possibility of group differences in distributions.

Is there the same level of sociopathy in other cultures?

No. In East Asian countries, China and Japan in particular, there is substantially less sociopathic behavior observed. It seems to me that the only explanation for that would deal with overall cultural attitudes. In the East, individual winning is not considered the appropriate goal. The culture is more group-oriented. A sociopath born in such a culture might learn to behave appropriately the way one might learn table manners. They might not have a conscience, but because sociopaths need to fit in, the behavior might be tamped down a bit.

In the past I've heard it claimed that the (putative) dearth of E. Asian "geniuses" is due to group differences in personality factors like sociopathy or psychoticism, rather than IQ. I think the topic of group differences in personality (whether cultural or genetic) is under-investigated.

Superhumans

Eugene ("Track Town USA") has a long history in running (for instance, Nike was founded here) and hosts a number of high profile track meets during the spring and summer. One of the things I enjoy about these meets is that while they are here I often bump into superhumans (world class athletes) around town, like at the cafe or ice cream shop :-)

Speaking of superhumans, below is a picture of decathlete Ashton Eaton, who may soon have a claim on the title World's Greatest Athlete. The last time (2009) the USATF National Championships were here in Eugene, I predicted great things ahead for him. Eaton, then only a junior, placed second in the US championships. This year (competition completed yesterday) he won by a huge margin, with a total that is the 5th best all-time US score and 13th best in the world. Eaton has incredible times in the sprints (PRs around 10.3, 13.3, 46, IIRC) and is still just learning the technical aspects of the high jump and the throws. In the next few years he could break the world record.



Two winters ago UO had a family day at the rec center, and invited some prominent athletes to spend time with the kids. My twins got to play dodgeball with Eaton for about half an hour -- there were no other families on the court. Eaton took it easy on them :-) When they get older I'll remind them about their time with the World's Greatest Athlete!

Thursday, June 23, 2011

What is g?

Neuroscientist John Duncan (FRS, Cambridge) argues that the neurological origin of g is in the frontal and parietal lobes.

Popular science book and public lecture. At 22 minutes Duncan gives a good pragmatic discussion of g in response to an anti-g interlocutor (Howard Gardner disciple). The streaming player at the link doesn't let me skip ahead, so probably better to download the file and play on your machine if you just want to hear this part of the discussion. Best thing to do is get it via iTunes so you can listen to the whole talk at your leisure :-)

A Neural Basis for General Intelligence (Science)

Universal positive correlations between different cognitive tests motivate the concept of “general intelligence” or Spearman's g. Here the neural basis for g is investigated by means of positron emission tomography. Spatial, verbal, and perceptuo-motor tasks with high-g involvement are compared with matched low-g control tasks. In contrast to the common view that g reflects a broad sample of major cognitive functions, high-g tasks do not show diffuse recruitment of multiple brain regions. Instead they are associated with selective recruitment of lateral frontal cortex in one or both hemispheres. Despite very different task content in the three high-g–low-g contrasts, lateral frontal recruitment is markedly similar in each case. Many previous experiments have shown these same frontal regions to be recruited by a broad range of different cognitive demands. The results suggest that “general intelligence” derives from a specific frontal system important in the control of diverse forms of behavior.

The paper above is from 2000. Below is a more recent one, supplied by a commenter. The authors claim a linear model using MRI measurements of brain features (essentially, volumes and activities of particular subregions identified using fMRI) can account for 50% of the variance in g as measured in psychometric tests. Their sample size for the imaging analysis and initial fit of parameters was about 200, and an independent group of 48 was used to test the model.

Multiple Bases of Human Intelligence Revealed by Cortical Thickness and Neural Activation (Journal of Neuroscience)

Choi et al.

We hypothesized that individual differences in intelligence (Spearman’s g) are supported by multiple brain regions, and in particular that fluid (gF) and crystallized ( gC) components of intelligence are related to brain function and structure with a distinct profile of association across brain regions. In 225 healthy young adults scanned with structural and functional magnetic resonance imaging sequences, regions of interest (ROIs) were defined on the basis of a correlation between g and either brain structure or brain function. In these ROIs, gC was more strongly related to structure (cortical thickness) than function, whereas gF was more strongly related to function (blood oxygenation level-dependent signal during reasoning) than structure. We further validated this finding by generating a neurometric prediction model of intelligence quotient (IQ) that explained 50% of variance in IQ in an independent sample. The data compel a nuanced view of the neurobiology of intelligence, providing the most persuasive evidence to date for theories emphasizing multiple distributed brain regions differing in function.


From the conclusions:

... Although our model still does not approach the predictive power of psychometric tests, its high accuracy suggests that neurometric assessments of intelligence may soon become a useful complement to psychometric test. For example, brain images might be used to improve intelligence estimates for individuals whose psychometric scores systematically underestimate their IQ. We hope that future research will build on our neurometric model of intelligence, both refining it so that it generalizes to other populations and expanding it to enhance its accuracy.

Will someone please alert the ghost of Stephen J. Gould concerning dangerous reification activity? I imagine he might still be capable of issuing a fatwa (perhaps via the IMF ;-) despite the fact that this research was conducted in Korea. I'm sure these Koreans and their Yale collaborator are cooking their data just like old George Morton.

Wednesday, June 22, 2011

On empathy: psychopaths, sociopaths and aspies

Last week a startup CTO, who didn't know my background, characterized all CEOs as "warm sociopaths" :-) He is at least partly right: many business and political leaders are good at reading other people's thoughts and emotions, but lack genuine concern for their well being. On the other hand, many geeks are very bad at mind reading or emotional perception, yet adhere to a strict moral code.

Cambridge cognitive scientist Simon Baron-Cohen (his cousin is the comic Sacha) classifies different low-empathy types below. See this podcast talk and this earlier post about his book on autism and the systematizing / empathizing spectrum. His latest book is specifically about empathy.

I’m O.K., You’re a Psychopath (NYTimes): ... For Baron-Cohen, psychopaths are just one population lacking in empathy. ... Baron-Cohen calls these ... groups “Zero-Negative” because there is “nothing positive to recommend them” and they are “unequivocally bad for the sufferer and those around them.” He provides a thoughtful discussion of the usual sad tangle of bad genes and bad environments that lead to the creation of these Zero-Negative individuals.

People with autism and Asperger’s syndrome, Baron-Cohen argues, are also empathy-deficient, though he calls them “Zero-Positive.” They differ from psychopaths and the like because they possess a special gift for systemizing; they can “set aside the temporal dimension in order to see — in stark relief — the eternal repeating patterns in nature.” This capacity, he says, can lead to special abilities in domains like music, science and art. More controversially, he suggests, this systemizing impulse provides an alternative route for the development of a moral code — a strong desire to follow the rules and ensure they are applied fairly. Such individuals can thereby be moral without empathy, “through brute logic alone.”

David Brooks addresses related themes in his recent book Social Animals. I highly recommend this podcast talk. His opening monologue is actually very funny -- he notes the similarity between politicians and people with the genetic condition Williams Syndrome :-)

Wikipedia: ... Most individuals with Williams syndrome are highly verbal and overly sociable, having what has been described as a "cocktail party" type personality, and exhibit a remarkable blend of cognitive strengths and weaknesses.[3] Individuals with WS hyperfocus on the eyes of others in social engagements.

... While patients with Williams syndrome often have abnormal proficiency in verbal skills, they do not perform better on verbal tasks than average. This syndrome is characterized more by a deficiency in other areas of processing. [Glib, but often mildly retarded.]

I would guess that "neurotypicals" strike aspies the way that Williams sufferers strike the rest of us. Imagine how disturbing it must be to live in a society dominated by and structured around people so different from yourself.

Tuesday, June 21, 2011

PhD Comics: the movie

PHD Movie Trailer from PHD Comics on Vimeo.


I met Jorge Cham, the cartoonist who draws PhD Comics, a few years ago at Sci Foo. Cham was a roboticist (grad school at Stanford and ME instructor at Caltech) before becoming a full time cartoonist. I notice this movie was made in collaboration with a Caltech theatre company.





Monday, June 20, 2011

Equilibration in progress

The US salary figure for MBAs from "leading schools" seems too low to me. Is this apples to apples? Still, it's incredible what people are earning in China and India. One private equity guy I know told me they are hiring top talent in Beijing/Shanghai for USD $100k+ these days.



Deutsche Bank Securities: Summary

- For decades, outsourcing of manufacturing and later of services to Asia has defined the trajectory of global production. However, sharp increases in wages/salaries in both China and India have led some observers to question whether or not this dynamic is coming to a close, particularly in view of the renewed competitiveness of US manufacturing.

- We found that the American manufacturing worker is indeed the most competitive in the developed world given the weak dollar and decades of efficiency gains. However, our study suggests that an industry that has moved already to China is unlikely to move back to the developed world. China's main problem is that it needs to move up the value chain even as low-end products are competed away to even cheaper locations. The country's inland provinces are unlikely to protect it against the shift. Unfortunately, China's cost advantage falls in high-end sectors and it will most likely succeed in segments where its large domestic market gives it critical mass.

- India too will find that it has a falling cost advantage in high-end services due to sharp increases in salaries. However, recent experience suggests that it can hold its own as one goes down the value chain. We found that companies have discovered ways to tap a large pool of low-skill workers from the hinterland. If it gets its policy framework right and avoids appreciation of its real effective exchange rate, it could potentially do the same in manufacturing.

- The British worker was found to have improved the most in the developed world although still lagging behind his/her German and French counterparts. Meanwhile, the Japanese worker looks the most vulnerable at current exchange rates."


Here is their summary on China. I was told that Foxconn (makers of Apple products) will move to the interior of China as opposed to cheaper labor markets like Vietnam, due to ecosystem and infrastructure issues.

... The question is whether or not the ten-fold increase in nominal USD wages and the prospect of more increases in the future have made China unattractive for further outsourcing of manufacturing from the developed world. This must be gauged against the fact that Chinese workers have also become far more productive due to education/training, improved infrastructure and so on. Our estimates indicate that the average Chinese hourly wage is now around USD 2.7. Wages in the manufacturing sector are generally lower at around USD 2.4/hour. Our informal survey suggests that the average Chinese worker is at least a third as productive as an American. Thus, we should compare USD 7.2/hour to USD 25.6/hour in US manufacturing. This is still a very large gap even allowing for 15-17% annual increase and perhaps some CNY appreciation.

... The economics of outsourcing manufacturing to China, therefore, varies a lot from product to product depending on the relative importance of each input. On one hand, China now has a number of established industrial clusters with their eco-systems of supply chains, manpower and so on. These clusters are being helped by the fact that China itself is a growing market for many products. These factors give Chinese manufacturing a degree of cushion. On the other hand, real estate prices have spiraled out in China in recent years. There are also concerns about the costs and risks of maintaining long supply chains for products that are still mainly sold in developed markets.

At the risk of generalizing, our overall assessment is that rising Chinese wages in themselves have not closed the case for outsourcing manufacturing to China. An industry that has already moved to China is not likely to move back to the developed world. The real competition in these industries comes from other developing countries. This is particularly true for low-end products with wafer-thin margins. For instance, Vietnam’s wages are half of those in China. Although the quality-adjusted gap is smaller, Chinese wages are rising much faster than Vietnamese wages and therefore making it increasingly worthwhile to shift certain kinds of manufacturing south of the border. It has been argued that China can compete back by encouraging investments in the cheaper inland provinces but we found that the wage gap is not large enough to justify this for products meant for export. The average wage in the large inland province of Sichuan is only 23% lower than in the coastal province of Zhejiang (the gap is much smaller for manufacturing wages). After accounting for the additional logistical cost of moving inland and for future wage growth, we feel that an export-oriented industry that is no longer viable on the east coast will move abroad rather than inland. The inland provinces will mainly succeed in sectors aimed at the domestic market or where there is special government support.

... This is not to suggest that the Chinese cannot compete on innovation and high-value activities. After all, it is a country capable of high-tech nuclear and space programmes. Recent census data also showed that the proportion of college graduates jumped from 3.6 to 8.9 per hundred between 2000 and 2010. Nonetheless, the Chinese cost advantage declines as we go up the value chain. As already shown in the previous section, Chinese MBAs are already half as expensive as equivalent Americans. The gap may be smaller after considering relative productivity and the pace of salary increases. In our view, therefore, the future trajectory of China’s production base will have a strong bias towards activities where the domestic market gives it critical mass. According to Jun Ma, Deutsche Bank’s Chief Economist for China, rising wages will create a big domestic market for equipment and machinery and the country could even leverage this for the export market. At current exchange rates, we think that Japan could be at most vulnerable to such a development. It could even affect Germany in the long run despite its famed ability to compete on design and quality.

Friday, June 17, 2011

Summer in Eugene

One week until the USATF National Championships come to Eugene.




View from the deck :-)




New Ed. school building.



We have fancy organic slow food carts on campus.








This was actually taken back in Taipei, but it fits the summer theme. Why are my kids so much better looking than I am? :-)

Human capital mongering: M-V-S profiles

The figure below displays the math, verbal and spatial scores of gifted children tested at age 12, and their eventual college majors and career choices. This group is cohort 2 of the SMPY/SVPY study: each child scored better than 99.5 percentile on at least one of the M-V sections of the SAT.





Scores are normalized in units of SDs. The vertical axis is V, the horizontal axis is M, and the length of the arrow reflects spatial ability: pointing to the right means above the group average, to the left means below average; note the arrow for business majors should be twice as long as indicated but there was not enough space on the diagram. The spatial score is obviously correlated with the M score.

Upper right = high V, high M (e.g., physical science)
Upper left = high V, lower M (e.g., humanities, social science)
Lower left = lower V, lower M (e.g., business, law)
Lower right = lower V, high M (e.g., math, engineering, CS)

Because of the selection criteria I wouldn't be surprised if the SDs are large in this population. Many of the SMPY qualifiers could have relatively average V scores and vice versa for SVPY. So the variation between the highest and lowest scores in each ability could be larger than in the general population.

Thursday, June 16, 2011

High V, Low M

I sent the message below to a social scientist I know who (like many, perhaps understandably) is confused about Stephen J. Gould's status as an evolutionary theorist. Many Gould readers are surprised to learn that his main expertise was the paleontology of snails and that he struggled with higher mathematics. When I first encountered Gould's essays as a kid, I concluded that there was just no there there. He was all literary flourish and little depth.

Which brings me to an observation I've been meaning to write about. It is that high verbal ability (which Gould certainly had) is useful for appearing to be smart, or for winning arguments and impressing other people, but it's really high math ability that is useful for discovering things about the world -- that is, discovering truth or reasoning rigorously. The importance of math ability manifests in two distinct ways:

1. Powerful (deep) models of Nature (e.g., electrodynamics or evolutionary theory) are themselves mathematical. Most of the incredible progress in our understanding of the universe is just not available to people who do not understand math. For example, we can talk until we are blue in the face about the Uncertainty Principle, but there is no precise understanding without actual equations.

2. The statistical techniques used to analyze data obtained in a messy, complex world require mathematical ability to practice correctly. In almost all realistic circumstances hypothesis testing is intrinsically mathematical. It is quite easy to fool yourself statistically if you don't have strong math ability, but rather are simply following cookbook recipes.


High verbal ability is useful for more than just impressing others -- it typically implies a certain facility with concepts and relationships between ideas -- but high V alone is a dangerous thing. The most confused people I meet in the Academy tend to be high V, low (modest) M types.

More on the V / M split in this longitudinal study of gifted children (SMPY / SVPY -- see esp. figure 4).


Gould appeals to high V low M people who do not actually understand evolutionary theory at a mathematical level. He never made any important contribution to evolutionary theory other than as a popularizer.

Note this is distinct from his deliberate obfuscation of topics like IQ in Mismeasure of Man. He wrote some incorrect things there about factor and statistical analysis, but perhaps those distortions were intentional. See The Mismeasure of Science: Stephen Jay Gould versus Samuel George Morton on Skulls and Bias.



Paul Krugman:

Now it is not very hard to find out, if you spend a little while reading in evolution, that Gould is the John Kenneth Galbraith of his subject. That is, he is a wonderful writer who is beloved by literary intellectuals and lionized by the media because he does not use algebra or difficult jargon. Unfortunately, it appears that he avoids these sins not because he has transcended his colleagues but because he does does not seem to understand what they have to say; and his own descriptions of what the field is about - not just the answers, but even the questions - are consistently misleading. His impressive literary and historical erudition makes his work seem profound to most readers, but informed readers eventually conclude that there's no there there. (And yes, there is some resentment of his fame: in the field the unjustly famous theory of "punctuated equilibrium", in which Gould and Niles Eldredge asserted that evolution proceeds not steadily but in short bursts of rapid change, is known as "evolution by jerks").

What is rare in the evolutionary economics literature, at least as far as I can tell, is references to the theorists the practitioners themselves regard as great men - to people like George Williams, William Hamilton, or John Maynard Smith. This is serious, because if you think that Gould's ideas represent the cutting edge of evolutionary theory (as I myself did until about a year and a half ago), you have an almost completely misguided view of where the field is and even of what the issues are.


http://www.pkarchive.org/theory/evolute.html



John Maynard Smith:

"Gould occupies a rather curious position, particularly on his side of the Atlantic. Because of the excellence of his essays, he has come to be seen by non-biologists as the preeminent evolutionary theorist. In contrast, the evolutionary biologists with whom I have discussed his work tend to see him as a man whose ideas are so confused as to be hardly worth bothering with, but as one who should not be publicly criticized because he is at least on our side against the creationists. All this would not matter, were it not that he is giving non-biologists a largely false picture of the state of evolutionary theory."


John Tooby:

"Although Gould characterizes his critics as "anonymous" and "a tiny coterie," nearly every major evolutionary biologist of our era has weighed in in a vain attempt to correct the tangle of confusions that the higher profile Gould has inundated the intellectual world with. The point is not that Gould is the object of some criticism -- so properly are we all -- it is that his reputation as a credible and balanced authority about evolutionary biology is non-existent among those who are in a professional position to know...

These [major evolutionary biologists] include Ernst Mayr, John Maynard Smith, George Williams, Bill Hamilton, Richard Dawkins, E.O. Wilson, Tim Clutton-Brock, Paul Harvey, Brian Charlesworth, Jerry Coyne, Robert Trivers, John Alcock, Randy Thornhill, and many others."


http://cogweb.ucla.edu/Debate/CEP_Gould.html

Wednesday, June 15, 2011

Incentives matter, but what kind?

Monetary incentives work for simple, mechanical tasks, but not so well for cognitively loaded tasks.

Once monetary compensation is high enough that people can turn their attention to the work, additional compensation (or emphasis on compensation) can actually have negative consequences. What really elicits good performance: autonomy, mastery and purpose.


Tuesday, June 14, 2011

Foo: exuberant geeks



Making the future.



CTOs and CEOs and Founders and Inventors and Creators.



Never before have the economic and creative prospects been so good for a young person with quantitative or technical abilities. In my generation we had fewer options: defense, academia or big stodgy corporations. Today you can code or analyze data or build mathematical models for a hedge fund, a bank, a social networking startup, a web publisher, an e-commerce site, a video game company, Google, etc., etc. If you can manage a team and communicate your vision to investors and partners, all the better. The sky is the limit.




"The Times gets a billion impressions a day. How do they optimize their ad revenue? To get real-time or nearly real-time analysis of, say, how many people in Wisconsin who were on Zappos within the last hour also viewed a Style section article, we need a big hadoop deployment and in-memory database that let's the user slice through a 100 dimensional parameter space. It's business intelligence on a supercomputer. Remnant ad space is auctioned on AdSense to competing bots, with the whole thing -- cookie analysis, automated bidding, ad fetching and placement -- taking place over 500 milliseconds."

"We're empowering people in rural India by giving them access to piece-rate work over the internet."

"90% of mechanical turk work is web spam." :-)

"They tried to buy us with shares priced on a secondary market, but I wouldn't trade my execution risk against that valuation risk."

"Yes, it's a flashlight but it's LED with a lithium ion battery. 3000 lumens shined into a burglar's eyes will blind them for 20 minutes. Those fins are for heat dissipation; the tip reaches 180 degrees."

"Computer vision is next. Much easier than NLP, but it'll be hacks strung together like everything else."


A few more Foo Camp 2011 photos here. Scott Berkun shares his insights here.

Sunday, June 12, 2011

Foo 2011 photos

Opening dinner.




The scheduling scrum.



Werewolf?



Make zone.





A schedule board.



Face board.



A session on Bitcoin. There seems to have been a bit of a crash today -- the dollar exchange rate went down to $8 from $28. Was there more rhetoric from Senator Schumer?



Outside camping.



Inside camping.



Geeks and booze.



Favorite quote of the day: "Startup CEOs are warm sociopaths."

Friday, June 10, 2011

Foo Camp 2011

I'm off today to Sebastapol, CA for another Foo Camp :-)



Here is what I wrote about it last year.

Thursday, June 09, 2011

GRE and SAT validity

If you are a professor at a research university you have probably spent time on graduate admissions. How good is the GRE as an indicator of candidate quality? Is the subject score more useful than the general score? What about relative to undergraduate GPA? Similar questions apply to the SAT and undergraduate admissions.

In both cases the answer is that standardized tests have roughly as much predictive power as GPA (SAT is about as powerful as HS GPA; GRE similar to undergraduate GPA). Not bad for a brief test! When these factors are combined the overall predictive power is increased. My opinion is that standardized tests load more heavily on cognitive ability and less on conscientiousness relative to course grades, hence the non-redundant information in the two measures.



This is a meta-analysis of GRE validity by Kuncel and collaborators. Here are slides from a related talk. See also this blog discussion (social scientists) of the GRE.

I first came across Kuncel's work when I was investigating SAT validity using U Oregon undergraduate data. Unlike many of the people working on this topic, Kuncel is neither an idiot nor an ideologically motivated faker. For example, he understands issues like restriction of range, variation in course and subject difficulty, and self-selection effects. See these slides from his talk on SAT validity. Video and audio here (bonus: Thomas Espenshade; Kuncel starts at 27 minutes in).




Validities for various standardized tests:

Monday, June 06, 2011

Goodbye Taiwan

Tomorrow night I'm returning to the US. My 9 month sabbatical here went by fast!



Here's a drawing I found while cleaning out my desk. The kids can read English and a few Chinese characters, too.




Here are some photos I took at the airport in Macau on my way back from BGI.





My last official activity here at Academia Sinica will be to give a colloquium in the Institute of Physics. Below are two figures of interest, produced by one of my collaborators. No Harvard, MIT or other Boston area students were harmed ;-)


Reaction time and g -- task is to determine whether a number flashed on a screen is greater or less than 45.


Note 1280 is about 90th percentile in the population (IQ 120) whereas 1560 is about 99th percentile (IQ 135 or so).

Huh? What? But, my Sociology professor told me it's just a social construction! 8-(


Three intelligence tests:

Millionaires next door

Percentage of families with at least $1 million USD in investible assets (excludes primary residence). Data from BCG Global Wealth 2011 Report. ***

1. Singapore 15.5%
2. Switzerland 9.9%
3. Qatar 8.9%
4. Hong Kong 8.7%
5. Kuwait 8.5%
6. UAE 5%
7. United States 4.5%
8. Taiwan 3.5%
9. Israel 3.4%
10. Belgium 3.1%
11. Japan 3%
12. Bahrain 2.6%
13. Ireland 2.3%
14. Netherlands 2.3%
15. UK 2.2%

*** Wealth is defined as total assets under management (AuM) across all households. AuM includes cash deposits, money market funds, listed securities held directly or indirectly through managed investments, and onshore and offshore assets. It excludes wealth attributed to investors’ own businesses, residences, or luxury goods.

Friday, June 03, 2011

Busy day at BGI

The Mayor of Shenzhen visits BGI. CGU may become a state Key Lab :-)



Shenzhen TV station interviews wunderkind Zhao Bowen.



Dutch documentary film crew arrives.




Film crews collide -- Dutch shoot Shenzhen TV shooting Bowen.



Microphone prep for filming a CGU meeting.

Blog Archive

Labels

Web Statistics