# Half-Space Algorithms For Identifying Geniuses

In his latest New York Times op-ed column, David Brooks, the conservative liberals can most stomach, attempts to tackle the problem of “what makes a genius”. This is, of course, the kind of reasonable length topic that one can explain in a single newspaper column (it’s the New York Times, you now.) The article begins, like all great op-ed, with a strawman that would make Dorothy proud:

Some people live in romantic ages. They tend to believe that genius is the product of a divine spark. They believe that there have been, throughout the ages, certain paragons of greatness — Dante, Mozart, Einstein — whose talents far exceeded normal comprehension, who had an other-worldly access to transcendent truth, and who are best approached with reverential awe.

Having properly stuffed his straw man, Brooks then lights it afire with his main thesis:

The key factor separating geniuses from the merely accomplished is not a divine spark. It’s not I.Q., a generally bad predictor of success, even in realms like chess.Instead, it’s deliberate practice. Top performers spend more hours (many more hours) rigorously practicing their craft.

This is, of course, a miraculous discovery, worthy of a true genius! Did you know that you can identify geniuses by the use of a two dimensional plot and circling those in the upper right hand corner? I had no idea.

Increasingly I find myself running flat into this argument. Take property P of some data set. Notice that quality Q correlates with P in such a way that there is a threshold value of Q below which data elements do not give you P, but where higher Q leads to a higher chance of P. Conclude that Q cannot be the only explanation for P. Search mind for other quality QQ different from Q. Notice that quality QQ correlates with P in such a way that there is a threshold value of QQ below which data elements do not give you P, but where higher QQ leads to a higher chance of P. Conclude that having qualities Q and QQ lead to P. I call this “argument by half-spaces.” Just keep dividing your space by half-space equations and your sure to capture your prey. But why the users of this argument feel that it is sufficient to stop at two or three half spaces is a mystery to me.

And, seriously, is Brooks really so dense to believe what separates out geniuses is this extra degree of practice? Or is he just worried that he needs to separate himself out from the rest of the high IQ crowd?

1. #1 Markk
May 1, 2009

This is the Malcolm Gladwell 10000 hour argument. It has been around for a while and is fairly popular. I would believe it more except that whenever I look into one of Gladwell’s amazingly convincing theories too closely they fall apart due to bad logic or bad evidence, so I don’t trust him anymore. He always still sounds convincing on his face to me. He quotes a lot of research but when you look at it, it doesn’t say what he says. Anyway …

2. #2 Dave Bacon
May 1, 2009

Markk: try searching for “Outliers” in the search box for this blog (on the left sidebar) đź™‚

3. #3 Pig
May 1, 2009

I recommend training a support-vector machine for this endeavor.

4. #4 Geordie
May 1, 2009

Pig: SVMs are so 2008. What you really need is a good deep belief network or sparse coding approach.

5. #5 Jonathan Vos Postj
May 1, 2009

Better beware of notions like genius and inspiration; they are a sort of magic wand and should be used sparingly by anybody who wants to see things clearly. (JosĂ© Ortega y Gasset, â€śNotes on the novelâ€ť)

Does one have to be a genius to do maths?

6. #6 Pig
May 1, 2009

Geordie: Manifold Learning algorithms for the win. Imagine writing a pop sci article entitled “The Genius Manifold.” How could would that be?

7. #7 Geordie
May 2, 2009

nice

8. #8 Jon
May 2, 2009

You ‘scientists’ with your ‘algorithms’. Isn’t it clear that Mr. Brooks has a political agenda which is to introduce a metric that can be easily understood and would lead people (99+% of whom clearly aren’t geniuses) to believe that there is a method that ‘ordinary’ people can use to ‘become’ a genius (or at least to be recognized as such).

I am sure that we can all ‘agree’ that the idea that you need 10000 hours (or that there is any generalizable threshold number of hours) to transcend proficiency and become great is pretty weak. But this assertion is a smart way to move the focus onto extended training of people and put off paying highly educated and skilled people a decent salary until they’re in their 40s. (We are just waiting to hire you ‘until you cross the genius threshold’.) But in reality you don’t need a lot of training, especially in emerging disciplines, to make a significant impact. If you want a way to keep smart, hard-working (not genius) people underpaid and in training longer, just keep raising the bar. Look at NIH funding statistics – the average age of first major (R01) grant has climbed to 42. The average lifetime of funding appears to be about 15 years (this number is not easy to come up with, but 4-5 grant cycles seems right), so you end up ‘in training’ starting in graduate school, for ~20 years, to have a productive career of ~15 years.

* – the 2007 NIH report is here: http://report.nih.gov/NIH_Investment/PDF_sectionwise/NIH_Extramural_DataBook_PDF/NEDB_SPECIAL_TOPIC-AVERAGE_AGE.pdf

9. #9 David
May 2, 2009

But in reality you don’t need a lot of training, especially in emerging disciplines, to make a significant impact.

Right, you just need to get dumb lucky. And if God is on your side, you will. So who needs discipline?

10. #10 sep332
May 2, 2009

With a nifty bit of handwaving at the beginning, you manage both to framr the argument as “conservative” vs. “liberal,” and to obscure the fact that the point made is definitely more liberal than conservative. The issue here is bad math.

11. #11 Dave Bacon
May 2, 2009

Actually sep332 I didn’t mean to imply anything about this being liberal vs. conservative (you might note my description of Brooks is totally uncorrelated with the main point of the post.)

12. #12 Jonathan Vos Post
May 2, 2009

The 10,000 hour rule of thumb’s been thrashed out for over a decade on the internet. More politically provocative of Malcolm Gladwell is the notion that American pedagogy is bankrupt compared to most other nations because of its addiction to “aptitude” (he says in interviews: “whatever THAT means”) as opposed to “how hard are you willing to work?” He suggests dividing pupils into classes not by test scores but by demonstrated willingness to try harder.

I deny that my students are stupid. Ignorant, yes, but that can be cured. None of my students are stupid. But many have been trained to be very VERY lazy. And many of the rest are willing to work tediously hard without seeking any insight to allow clever work in shorter time. Nor do the class syllabi and textbooks encourage working SMARTER.

I tell my students again and again that the world is filled with people who work hard all their lives and have nothing to show for it. The two keys are that you DO need to work hard (or marry rich) on the one hand, and to work smart (which is the use of meta-cognition to allow base-level laziness).

13. Reminds me of stories I’ve heard about becoming an “expert” in anything and about how it takes a certain number of hours spend practicing that task to be considered as such…

I guess I’m an “expert” in a lot of things by now- sleeping, breathing, blinking, etc, etc, etc- perhaps even a GENIUS by this man’s definition! (I think Mr. Brooks should follow the link I’ve posted above and perhaps get a better definition of this word which he seems to toss around so lightly!)

If our new definition of “genius” is determined by the number of hours spent practicing any given task, then we’ve truly turned our back on the roots of this monumentally important, and ancient word.

Back to the textbooks for me- thanks for the break Dave- my professors at American Sentinel probably wouldn’t be too stoked to hear that I spend so much time on your site, but I definitely the break!

And for the record- you’d get my vote for “genius” rank far sooner than Mr. Brooks…

14. #14 David
May 6, 2009

be not afraid of greatness: some are born great, some achieve greatness, and some have greatness thrust upon ’em.

or: there are many ways to skin a cat.