Old physicists fade away

A commenter writes, replying to my previous post on Eric and the Quantum Experts:

>Eric, you may still have a chance to revolutionize physics, since decoherence by itself may not completely solve the problem.

Alas, I am probably too old now. There is a way outside chance I could do it, yes, but,…hmm…how to explain this…

There’s an observed pattern in math and physics that most people do their best work very young. The brighter you are, the longer you get before you’re useless for anything but teaching, but it’s rare to see real breakthroughs from people past their early thirties. Only a very few exceptionally talented workers get to be creatively productive over their entire lifetimes; in physics that means being at the Einstein or Hawking level. (A few people slightly less talented seem to get a second wind – as synthesizers, rather than innovators – in their sixties.)

It’s almost as though creativity in these fields is an isotope with a half-life that varies by field and rises with individual level of IQ or native talent or something. Nobody understands this very well, it’s all unquantified folk knowledge.

The half-life of programming talent seems to be longer than for physics talent, which in turn seems to be longer than for pure math – but still, I used to worry that I’d become a useless lump as a programmer after forty. This does not seem to have occurred; in fact, I’m more productive now than I was at twenty-five (and I was pretty damn good then). It’s an interesting question whether the half-life for programming is longer than I thought or whether I’m in the tiny lucky minority of supertalents that jump off the exponential decay curve entirely. I don’t know the answer, and don’t even have a guess I’m confident about.

Unfortunately, because the half-life of physics creativity seems to be shorter than for programming, my success in the field I’m in does not predict that I’d still be able to do original physics.

I just turned 51. That means, in order to believe that I could do really strong and original physics work now, I’d have to start with a justified belief that I’m as talented as Hawking or Einstein. This is almost certainly not the case. I would say “certainly”, except that my general track record of creativity and insight is far enough off the mean to raise just the tiniest smidgen of realistic doubt about this. And I was, after all, ahead of the physics literature on something conceptually important at least once – even a lot of physicists never manage that.

Rationally, though, it’s not enough of a doubt for me to gamble on, at this point. I like what I do, and I’m good at it, and it has made me as famous as any sane person would want to be. I don’t have any great need to go off and try to conquer physics as well.

Though I will admit, semi-relatedly, that I feel a continuing temptation to try to write a disruptive, field-upending outsider book on the application of analytical philosophy. The tools I used to spot the hole in the Schrödinger’s Cat story are way underutilized.

Published
Categorized as Science

80 comments

  1. Funny–I’m mountain time, so I saw the ‘posted at 6:56 PM’ and thought, ‘I seem to have viewed this post before it was created.’

  2. Eric,
    I’d like to take issue with “unquantified folk knowledge”
    I think that is confusing some empirical outcomes with an incomplete set of causes. Being (smart, old or well respected) enough in a field gets you the option of teaching. An economically and socially acceptable alternative to research. Some take that option, thinning the talent pool. Also, as you are more successful in a profession the expectations change. You get asked to referee journals, review grant proposals-these take away time and focus. Reducing the time you have for your own research.

    I offer a brief list of counter examples individuals:
    Philo Farnsworth (of TV fame but did a lot of other interesting things like Fusion later in life)
    Ben Franklin (all his natural philosophy/physics was done after he retired form printing)
    James Thomson (Stem cell researcher. One of the first to isolate stem cells from embryos. Just recently has induced stem cell formation from adult cell)

    I am making these points not play devils advocate but to challenge this idea that only the young are creative or imaginative. When you are young and inexperienced creativity, imagination and ambition are all you have. With age and experience come options.

  3. Many fields show an accumulation function. Biology, for example.

    Its not surprising to me that the “more you know, the better you become” in programming, if not actual computer science.

    That you are surprised is … surprising.

    James Thompson… (the other one, not a stem cell researcher, but doing interesting work in bioinformatics.)

  4. You guys keep citing brilliant individuals as counterexamples to the folk theory. This doesn’t help. What we’d need is some kind of metric for creativity and a broad study of how it age-varies over entire fields.

    On the other hand, I recognize that I may be too influenced by the folklore of pure-math and physics types. It’s reasonable that half-life statistics might not apply to fields where accumulation of experience is a large factor, and that programming might be one of those.

    It’s even possible that the younger peak of expected creativity in pure math than physics may simply be because accumulation of experience is a larger factor in the latter.

  5. ESR,

    but you don’t have to do everything yourself, you could just team up with a team of researchers who could use some help from a programmer who has a good understanding of the domain.

    And I’m sure there are some. Story. I know a bloke who won a PhD scholarship at Stanford, so he must have been good. And he quit halfway through and is now teaching math in some boarding school, because all he did was writing C++ programs for number-crunching the results of particle physics experiments and he found it unbearably boring.

    And there must be a better way to do that, surely that’s a huge waste of time, one could find or write a good framework in something like Bigloo, Haskell, Mathematica or OCaml or NumPy or whatever, where each experiment is just a plugin and even half of that plugin can be code-generated from templates or whatever. There must be a better way and I think the only reason they didn’t find it that they were physicists, they saw programming as a necessary, boring chore to dump on the most junior member of the team, and nobody in team was enthusiastic enough about it to dig deeper into it and come up with a good solution. They could have used a good programmer on the team, as it was clearly a huge waste of time. You could find such a team and participate, couldn’t you?

  6. Shenpen: this is something that I can’t think about too much lest I become depressed. When I was doing my PhD (in pure maths), I saw so many students and researchers, bright but with no understanding of programming, wasting their time writing code the way it was done in the seventies: Fortran, manual version control, no automated testing, the works. Horrible. Apart from anything, it’s bad science: if your code is neither formally proved correct nor adequately tested, how can you have any confidence in results obtained from it?

    Eric: this isn’t folklore in the sense that mathematicians use the term, because one can cite a specific piece of the literature as its source: G H Hardy’s A Mathematician’s Apology. And remember that Hardy had spent much of his career working with Ramanujan, who’s one of the field’s most spectacular early shiners: for this reason alone, I’m suspicious of his claim without the sort of proper scientific study you ask for. My own observations were that some mathematicians produce a few results and then settle into teaching or admin (and I can’t really blame them: research is damned hard and depressing), some continue with research and produce decent results until they retire, and those who are going to really shine display their genius early, but don’t dim appreciably as time goes on. Also, I think it’s getting harder to be the kind of child mathematical prodigy you have in mind: there’s just so much more mathematics to know now before you reach the cutting edge, particularly in fields like algebraic geometry.

  7. >Eric: this isn’t folklore in the sense that mathematicians use the term, because one can cite a specific piece of the literature as its source: G H Hardy’s A Mathematician’s Apology.

    I don’t disbelieve you about the source (I never read it) but I can report that this belief was general among the theory-math people I knew at University of Pennsylvania in the 1970s; that’s where I acquired it. Are you sure Hardy wasn’t reporting a general folk belief of his time?

    >Also, I think it’s getting harder to be the kind of child mathematical prodigy you have in mind:

    I assure you I did not have child prodigies in mind, unless you think childhood ends in the early thirties.

    The distinction is clear to me because I was in fact a child prodigy — well, a teenage one, anyway. In 1975 I had a paper accepted by the American Mathematical Society and was invited to present it at the AMS conference in San Antonio. I had proven a closed form for the Nth term of the generalized Fibonacci series, and I presented it to a national audience of mathematicians. It wasn’t a particularly deep result, but there were some pretty connections to combinatorics and partition theory. I was 17, and the only high-school student in the AMS’s institutional memory to pass their filters.

  8. esr Says:
    >The distinction is clear to me because I was in fact a child prodigy — well, a teenage one, anyway. In 1975 I had a paper accepted >by the American Mathematical Society and was invited to present it at the AMS conference in San Antonio. I had proven a closed >form for the Nth term of the generalized Fibonacci series, and I presented it to a national audience of mathematicians. It wasn’t a >particularly deep result, but there were some pretty connections to combinatorics and partition theory. I was 17, and the only high->school student in the AMS’s institutional memory to pass their filters.

    That’s very cool.
    By the way, it looks like there are many famous hackers who had great mathematical abilities.
    Richard Stallman and Seth Breidbart come to mind.

  9. I think it comes down to immersion. When you’re young you can completely focus your life on something, and original work in math and physics in particular seems to require this. As mentioned earlier, as you age, you become responsible for others (children, grad students, etc.), and work to enable their success in addition to your own. This thins the ‘talent pool’

    Perelman talks about this in the famous New Yorker article, Manifold Destiny. (It’s got it’s own wiki page!)
    http://www.newyorker.com/archive/2006/08/28/060828fa_fact2

  10. > Though I will admit, semi-relatedly, that I feel a continuing temptation to try to write a disruptive, field-upending outsider book on the application of analytical philosophy. The tools I used to spot the hole in the Schrödinger’s Cat story are way underutilized.

    Since my comment prompted this post, I’ll weigh in briefly here: I think one area where these tools are way underutilized is quantum gravity, which I mentioned in my original comment. This may be changing; several recent books by experts in the field (notably Smolin’s *The Trouble With Physics*, which I read not long ago, and Woit’s *Not Even Wrong*, which I haven’t read but plan to) basically make the case that the emperor has no clothes. (They’re talking about string theory, but that’s the main contender right now for a theory of quantum gravity.) AFAIK, nobody is running with Penrose’s idea of a connection between quantum state-vector reduction and quantum gravity. Maybe Ed Witten was right when he said the trouble is that we’re trying to do twenty-first century physics with twentieth-century tools.

  11. There it is. Wasn’t Perelman. Mikhail Gromov.

    “To do great work, you have to have a pure mind. You can think only about the mathematics. Everything else is human weakness. Accepting prizes is showing weakness.”

    I assert that as people age fewer of them are willing to live monastically.

    Don Knuth is.
    http://www-cs-faculty.stanford.edu/~knuth/retd.html

  12. I wonder, though, especially with the physicists if the problem isn’t consummation of talent rather than loss of it. That is to say, I suspect the greatest breakthroughs are the result of synthesizing a unique viewpoint/perspective/thought-pattern to the field, rather than some extended labor.

    Thus, the reason the greatest works of physics, or math (or music?) tend to happen young isn’t that the young are more talented but that any paradigm-shifting work you have in you is likely to come out pretty quickly after you become proficient with the current work in your field.

  13. I guess once I heard you saying “I used to be a mathematician” in one of your earlier posts.I was really puzzled.Have you ever produced a notable theorem,conjecture,…anything?
    You didn’t even get your under-grad math degree,how’s it that you call yourself an ex-mathematician?When you say “I used to be…” it clearly shows your lack of understanding of what actually the meaning of a mathematician is.One can’t be a mathematician at one time and suddenly stops being one some seconds later .It’s actually the way one has grasped mathematics and the way he thinks about mathematical problems not just being able to REPEAT Erdos’s words.
    [Just in one case your “I used to be math-guru” sentence is correct: you were a math professor(using the word in a different sense ),but keeping in mind that you don’t possess a degree that’s unlikely]

  14. > Thus, the reason the greatest works of physics, or math (or music?) tend to happen young isn’t that the young are more talented but that any paradigm-shifting work you have in you is likely to come out pretty quickly after you become proficient with the current work in your field.

    That, and any paradigm shift or major breakthrough tends to generate a lot of follow-up work, which can easily absorb the rest of your career.

  15. >how’s it that you call yourself an ex-mathematician?

    Because I used to be focused on understanding and producing mathematics, Now I’m not.

    I can still think like a mathematician and understand moderately difficult mathematics. That may be the distinction you’re interested in.

    I don’t know whether you’d consider my one publicly-presented theorem to be notable or not. The AMS seemed to think so in 1975. They didn’t know I was 17 when they invited me to present.

  16. Buck:

    One can be a mathematician without being paid for it, or even getting a degree in it. A mathematician is someone who does mathematics, just as an astronomer is someone who does astronomy. In astronomy amateurs still contribute to the field.

    It’s only recently that mathematicians and scientists got paid for it. Even Einstein worked at the patent office.

    Don’t confuse credentials with reality.

  17. >By the way, it looks like there are many famous hackers who had great mathematical abilities. Richard Stallman and Seth Breidbart come to mind.

    Indeed. They’re good examples for this discussion because I’ve known both of them for many years and thus have a pretty detailed model of what they’re good at. Met them both originally through SF fandom in the 1970s.

    I don’t think mathematical talent is necessary to be a top-grade programmer, however. The ability to generate and follow long chains of logical reasoning, yes — but the most math most programmers ever need is what’s nowadays called “discrete math”: boolean algebra, combinatorics, modular arithmetic, a bit of graph and group theory. Calculus and genuine higher math are almost never useful outside a few subspecialties.

  18. Thus, the reason the greatest works of physics, or math (or music?) tend to happen young isn’t that the young are more talented but that any paradigm-shifting work you have in you is likely to come out pretty quickly after you become proficient with the current work in your field.

    A cognitive model proposed by Arne Dietrich suggests that Eric is exactly right. However, something bothers me about this research in perhaps similar fashion to the way that the Schroedinger’s Cat narrative bothered Eric. I suspect, but cannot prove, that lifestyle circumstances of Western civilization play a confounding role in the formulation in this theory. Until a few years ago, we believed that all brain neuronal growth ceased by age 25. Recent research suggests that lack of growth of new neurons is indicative of clinical depression. (Does that mean most of us are stealth depressed for all our working lives, like Jack Nicholson in About Schmidt? I wouldn’t rule it out.)

    But it doesn’t matter what I think, anyway. I’ve been similarly bothered about the narrative concerning anthropogenic global warming. The reason why I take such a firm stance in favor of that narrative anyway is the First Rule of Compiler Bugs.

  19. >A cognitive model proposed by Arne Dietrich suggests that Eric is exactly right.

    The last sentence in Dietrich’s abstract is:

    …paradigm-busting ideas occur overwhelmingly to people in
    their 20’s and early 30’s, as indication that a nimble prefrontal
    cortex and thus chronological age is a critical factor.

    I don’t think the folk theory I absorbed could be put any more exactly or succinctly than that.

    I’m reading Dietrich’s paper now, and the resonance between his model of insight generation and my personal experience of generating creative insights is very, very powerful. This guy is onto something; take that from someone who has actually done what he calls “paradigm-busting” thinking (though his Kuhnian abuse of the word “paradigm” is the single thing I most dislike about the paper).

    I think I buy his general theory, which makes it more interesting that I am a clear exception to his thesis. I had my most important paradigm-busting insights between ages 40 and 43.

  20. >Don’t confuse credentials with reality.

    I don’t have a degree in computer science, either. Never had a single semester-hour of it. Obviously my entire career in the field is a mirage and never happened.

  21. I think I buy his general theory, which makes it more interesting that I am a clear exception to his thesis. I had my most important paradigm-busting insights between ages 40 and 43.

    That’s exactly why I don’t buy it. I think his model of cognition is sound, but as mentioned above, I don’t think that there’s enough research out there to support such a clear-cut correspondence between age and cognitive flexibility in non-geriatric adults; and believe there are confounding factors involved such that the capacity for paradigm-busting doesn’t correspond as strongly with physiological age. Our society rewards us materially for being less likely to paradigm-bust; playing by the rules and earning a paycheck so you can keep a roof over your head and feed and college-educate the kids is more important.

    I’m interested in the relation between so-called “paradigm-busting” and humor; both require similar forms of juxtapositional thinking. Plenty of people maintain razor-sharp wits well into old age; Weird Al’s music gets funnier — and more ingenious — as he gets older.

    We’ve become so neotenous now that it can take twenty or thirty years to figure out what the hell to do with your life. Do people who tarry almost invariably miss their chance at greatness?

  22. I’ve been wondering if there is a negative correlation between the ‘age when great scientific ideas can occur’ and getting into a long-term relationship with another person. Being totally immersed in a subject is, in some ways, a very lonely business, even if you are working in a group. Most people can only take so much of it, before wanting to spend some real time with another person.

    On a related note: does anyone know of a highly successful research sabbatical (a leave from teaching and administration) taken by a senior professor? I work in the biosciences and all of the sabbaticals I’ve seen have accomplished remarkably little. Maybe they kept said professors from burning out under their administrative workloads, but the scientific output was nonexistant.

  23. >I had my most important paradigm-busting insights between ages 40 and 43.

    Ah, but while Dietrich demonstrates physicists peak in their early 30s, he suggests the optimal age for historians is early 50s. Applied sociology would presumably be somewhere in-between…?

  24. >Do people who tarry almost invariably miss their chance at greatness?

    Sadly, I think the answer is yes. I said in my original post that I didn’t have a theory about whether the half-life of talent in my field is longer or I’m an exceptional supertalent, nor even a confident guess. Dietrich, by providing statistical evidence derived from a well-thought-out experimental design and grounding both in a cognitive model that describes my experience well, has changed my mind. I now have a confident guess that I’m one of his curve-breaking supertalents, like Sanger or Bardeen.

    Or that I was at just past 40, anyway; it’s all too possible I’ve lost what he calls “cortical flexibility” in the intervening decade.

  25. >Ah, but while Dietrich demonstrates physicists peak in their early 30s, he suggests the optimal age for historians is early 50s. Applied sociology would presumably be somewhere in-between…?

    Very interesting point. Shrewd argument against the supposition that I’m a supertalent.

    On the other hand, though my operating methods were like those of a cultural anthropologist or sociologist, my big insight was rather like a physics one in several important ways. The most important, perhaps, were that (a) it involved successfully and unambiguously identifying false-to-fact premises of a pervasive world-view (not just facts that are wrong but framing ideas that generated bad predictions), and (b) it produced a theory that was generative rather than merely descriptive.

    New theory in history or applied sociology doesn’t often have even one of these properties, and the combination of both is vanishingly rare. Dietrich himself argues that historians can peak later because their sort of breakthrough tends to be more accumulative and synthesizing, and not so much to involve the abrupt unlearning of old premises.

    This perfectly connects with my experience that learning and theory-construction are easy, questioning the premises you don’t know you have is hard, and all my most important insights came from identifying and dynamiting a smuggled premise. As I did, for example, both in analyzing the Schrödinger’s Cat experiment and in understanding how and why Brooks’s Law breaks down.

  26. >I’ve been wondering if there is a negative correlation between the ‘age when great scientific ideas can occur’ and getting into a long-term relationship with another person.

    I entered my long-term relationship with my wife Cathy in the 1970s. We began living together in 1985. My breakthrough insights happened in the late 1990s. I’m a data point against your conjecture.

    It is not clear that this disposes of it, however, as I seem to be an outlier in lots of ways.

  27. Jeff,

    “Recent research suggests that lack of growth of new neurons is indicative of clinical depression. (Does that mean most of us are stealth depressed for all our working lives, like Jack Nicholson in About Schmidt? I wouldn’t rule it out.)”

    Sounds about right. A DJ friend of mine told me that most people stop discovering new kinds of music around the age of 25 and then listen to the same music as long as they live. And that’s what my parents did. Not sure it means the stoping of the growth of neurons but it just doesn’t sound healthy, and may be related to neuron growth. I didn’t bother to discover any new music between 25 and 30 (current age) either, I’m happy with my existing Rammstein compilation, which is a bit worrisome, it wouldn’t be good to get ossified in anything, but…

  28. >A DJ friend of mine told me that most people stop discovering new kinds of music around the age of 25 and then listen to the same music as long as they live.

    Interesting. By that criterion I’m still growing new neurons at 51. I’ve learned to understand and like a couple of new genres in the last couple of years – notably rap-metal, which I picked up a taste for while learning to swordfight.

    I thought this was an unusual but trivial fact about my life. Maybe it’s not trivial.

    (By the way, I like Ramstein too, but for a reason different from most of their fans. I find them side-splittingly hilarious. almost as funny as PDQ Bach.)

  29. (By the way, I like Ramstein too, but for a reason different from most of their fans. I find them side-splittingly hilarious. almost as funny as PDQ Bach.)

    Then you would love KOMPRESSOR.

  30. > I’ve been wondering if there is a negative correlation between the ‘age when great scientific ideas can occur’ and getting into a long-term relationship with another person.

    It makes it harder to maintain the necessary focus, sure. But never forget that Euler died at the age of 76 with a grandchild on one knee and a half-completed proof on the other.

  31. > I don’t have a degree in computer science, either. Never had a single semester-hour of it. Obviously my entire
    > career in the field is a mirage and never happened.

    Of course, your career is largely programming and marketing-like activities, not CS.

  32. Shenpen: I think that a lot of this is an artifact of how people live their lives. As they get older, they go out to gigs less, graduate to radio stations with less of a youth focus, and generally have more important things in their lives than keeping up with whatever music’s new. We’ll see if pervasive peer-to-peer and sites like Last.fm fix that in the coming years, I guess…

  33. My grandmother on my father’s side considered it very important, at the year’s end, to catalog any new types of music she’d listened to enough to buy in the last year. She died at (we think) 90-something. It’s hard to tell – she forged birth certificates to avoid mandatory retirement age as an elementary school teacher at least twice.

    I’ve kept that habit – I’ve also found that my more productive years in coming up with new products are the years where I glommed into different music types.

    In the games publishing industry, there are practitioners who are innovators (I’m one of them), synthesists (Steve Jackson) and recyclers (people who put out the same game with different chrome and a couple of new mechanics).

    Steve Jackson is better at combining existing mechanics into new forms than anyone I’ve ever met.
    I’m better at coming up with a new mechanic or new way of modeling something in a gameable way than anyone I’ve ever met.

    Both of us find banging out Yet Another Variation of something we’ve already done to be…dull. Steve is happiest when he’s playing mix and match with mechanics to make a nifty way to reward a specific behavior.

    (Game design is a peculiar mix of mathematical modeling, obsession on a topic, and behavioral science.)

  34. “I find them side-splittingly hilarious.”

    Interesting. I don’t know why most other people like it, but I personally find it basically a kind of music that rouses the deeply buried BDSM Dom in me, balancing on that thin edge where it’s serious enough to enjoy it but satirical and mocking enough to don’t take them _really_ seriously as I figure that would be really sick and perverted.

  35. > A DJ friend of mine told me that most people stop discovering new kinds of music around the age of 25 and then listen to the same music as long as they live.

    I’m past 30 and I’ve gotten into a few new genres recently, including some opera (which I would have thought highly improbable when I was 25) and some pretty esoteric electronic music.

    I do find it generally to be true that people don’t go out of their way in finding new music after about 25, including some who advertise themselves at every opportunity as being into ‘alternative’ culture but are irreversibly stuck with e.g. 70s punk and can’t see much past that.

    I wonder how the current state of the pop music industry is going to affect this. I think a huge part of the stuff that is in the charts sucks to an extent it didn’t use to, and I don’t think this is just me getting old. I’ve recently heard a lot of people in their late teens or early 20s spontaneously say ‘why can’t music be like this today’ when they hear chart pop from before their time.

  36. > Steve Jackson is better at combining existing mechanics into new forms than anyone I’ve ever met.

    But he is a royal pain in the ass to deal with, and his brother (Ken) is worse.

  37. >Of course, your career is largely programming and marketing-like activities, not CS.

    Um, no. One of my books has been cited as being among the ten most influential in the literature on large-scale software engineering, along with work like Brook’s The Mythical Man Month and Parnas’s foundational work on modular decomposition; even Wikipedia, with its everything-but-the-kitchen-sink approach to such lists, puts it in the top fourteen. Another one of mine (The Art of Unix Programming) is, according to my publisher, in routine use as a computer-science textbook. I’ve been a guest CS lecturer at enough Ivy League universities to lose count and an invited speaker at the Institute for Advanced Study.

    I don’t normally call myself a computer scientist, because I prefer “hacker”. But people who are indisputably computer scientists don’t seem to be in any doubt that I am one – they treat me as a peer. If infrequency of publication is what bothers you, do you plan to exclude Donald Knuth from your categorization? My output pattern is much like his: sporadic, but when I publish something it tends to make big waves.

    I wouldn’t normally make this much of a point of my stature in the field, except that we’re discussing issues about age and creativity where I am possibly a significant data point. An accurate assessment is therefore profoundly relevant.

    (Note: I am not claiming to be Knuth’s equal. He is one of the handful of CS people that I still look upon with awe. I treasure the memory of the day I got email from him. impishly reporting a bug in INTERCAL and suggesting a fix. I felt like the hand of God had reached down and touched my code.)

  38. My hypothesis for what you observed: gF decays faster than gC. Physics, math, and computer science are simply the most gF-loaded fields out there.

  39. >>> I don’t have a degree in computer science, either. Never had a single semester-hour of it. Obviously
    >>> my entire career in the field is a mirage and never happened.
    >>
    >> Of course, your career is largely programming and marketing-like activities, not CS.
    >
    > Um, no

    Actually, yes. Show us your research and resultant papers, not your books (none of which I am aware are the outcome of scientific pursuit.)

    Show us your peer-reviewed research activities.

    Show us a formal proof for even one of your programs.

    Joel Spolsky once wrote:

    The difference between a tolerable programmer and a great programmer is not how many programming languages they know, and it’s not whether they prefer Python or Java. It’s whether they can communicate their ideas.

    While opinions differ on your greatness, by this measure (and perhaps others) you are a great programmer. And thats a fine thing.

    But I have yet to see any *computer science* from you.

    Yes, Knuth is good, and yes, he’s a computer scientist. Other examples include Moon, Weinrab, Stallman, Dijkstra, Robert Floyd, James Gosling, Paul Graham, Griswold, Richie, Thompson and others.

    Neither you (nor I) are on the list (yet).

    I’m not saying you’re not a programmer/hacker.

  40. esr Says:
    >I’ve known both of them for many years and thus have a pretty detailed model of what they’re good at.

    Eric, can you please elaborate on this one a bit?
    I became very interested in this after I read Seth Breidbart’s post on UseNet (in which Breidbart called Stallman “The world’s best programmer”), but I couldn’t find any other information except for some info in the “Free As In Freedom” book (which is not much).

  41. >But he is a royal pain in the ass to deal with, and his brother (Ken) is worse.

    So I’ve been told before. SJ and I are very old friends and we get along well, but I’ve been careful never to enter a business relationship with him. I doubt either of us would enjoy it.

  42. >But I have yet to see any *computer science* from you […] Neither you (nor I) are on the list (yet).

    You are entitled to that opinion, but the people who I know for certain do not share it share it include the likes of Alan Kay, Fred Brooks, Ken Thompson, Paul Graham, and Doug McIlroy, and I think their opinion counts for more than yours. I appear to have passed the peer-review test for being considered a “computer scientist” despite having no peer-reviewed publications.

    I admit this is odd and category-busting. Well, “odd and category-busting” could be my middle name. Live with it.

  43. >They are fluid and crystallized intelligence, respectively.

    Right. And the folk theory I observed is that pure math is more critically gF-dependent that physics, which is in turn more so than programming. Dietrich does not specifically confirm this, but his results are completely consistent with it.

  44. >Eric, can you please elaborate on this one a bit?

    I’m not sure. What do you want to know about them?

    (Oh, and Seth is probably wrong. I’m not all that far south of Stallman’s level of ability, and probably comparable to Seth’s – but all three of us are way, *way* behind either Linus Torvalds or Ken Thompson, so much so that the differences between us three might as well be zero. In my humble opinion.)

  45. > I’ve been told before. SJ and I are very old friends and we get along well, but I’ve been careful never to enter a
    > business relationship with him. I doubt either of us would enjoy it.

    Oh, I”ve never been in business with him (or Ken), either. I have served as a financial ‘backstop’ for someone I respect when he was unfairy and unwillingly pitted against Ken’s rapacious manner and method.

    Steve and I still speak to each other, at least.

    > You are entitled to that opinion, but the people who I know for certain do not share it share it include the likes
    > of Alan Kay, Fred Brooks, Ken Thompson, Paul Graham, and Doug McIlroy, and I think their opinion counts for
    > more than yours.

    You are no doubt aware of the saying, “pics or it didn’t happen”. I assert that you still have yet to provide any evidence of being (or having been) a “computer scientist”.

    > “odd and category-busting” could be my middle name.

    Perhaps an interesting way to live your time on earth, but it doesn’t even rhyme with Steven.

    eocbr…

  46. esr Says:
    >I’m not sure. What do you want to know about them?

    For example, what are their (Stallman, Torvalds, Breidbart) particular strengths in programming and also is the speed of coding a good way to measure a programmer’s skill)?

    Also, what do you think of Stephen Wolfram as a programmer?

  47. >For example, what are their (Stallman, Torvalds, Breidbart) particular strengths in programming and also is the speed of coding a good way to measure a programmer’s skill)?
    >
    >Also, what do you think of Stephen Wolfram as a programmer?

    I don’t think I can answer all these questions in detail, but I’ll take a swipe at some of them. I’ll start by saying I know basically nothing about Wolfram as a programmer beyond the bare fact that Mathematica seems to be a very good piece of work. I’ve only met him once: he struck me as both almost impossibly arrogant and quite possibly bright enough to justify the arrogance.

    As for Breitbart, Stallman, Torvalds, and incidentally myself…first, my basis of fact. I’ve collaborated extensively with RMS and know him well face to face. I’ve collaborated some with Linus, read a lot of his code, and know him face-to-face not as well. I’ve never collaborated at all with Seth; my model of Seth’s capabilities is simply based on having interacted with him a lot and swapped war stories.

    Ten years ago I noted that there are two distinct kinds of master-level programming talent. RMS and I have one; Linus has the other and I think Seth is like Linus. RMS’s talent (and mine) is for novelty; we can see around corners in design space and come up with approaches that nobody imagined before but which look obvious in retrospect. Linus’s and Seth’s talents seem to be at complexity control; they’re not as innovative by nature, but they’re better at seeing the least-effort path from A to B and avoiding traps on the way. (When I say “less innovative” I don’t mean “not innovative”; I have no doubt either of these guys generates more novelty in any given year than most programmers will in their entire lifetimes. Still, RMS is better yet at that specific thing, as I seem to be too.)

    The interesting thing is that while you might guess that innovator mastery like RMS’s and mine is more important and more powerful, I don’t actually think this is true. Linus is, in my judgment, much more capable than either of us and the talent he and Seth have is rarer. And Ken Thompson…he clearly has both talents at a level so high that I am slack-jawed in awe thinking about it.

    (I will qualify this by noting that when I presented a version of this discussion that didn’t include Seth in my first paper, several reviewers suggested that I probably undervalue innovator mastery because I have it and thus take it for granted.)

    Is the speed of coding a good way to measure a programmer’s skill? I don’t know; I don’t even have a well-formed opinion. I’d have to think about that one for a while.

  48. esr Says:
    >What do you intend to do with the answers?

    Nothing except for reading them a couple of times, of course. What else can I do with them?
    I was just curious how great programmers evaluate each other (your comment about “two distinct kinds of master-level programming talent” cleared this up a bit, this is probably analogous to “2 cultures within mathematics”).
    http://www.dpmms.cam.ac.uk/~wtg10/2cultures.pdf

    And since you have a blog (where you actively respond to the comments) and you know other great guys… So I thought it was a good idea to ask you about this kind of stuff.

  49. >(your comment about “two distinct kinds of master-level programming talent” cleared this up a bit, this is probably analogous to “2 cultures within mathematics”)

    Yes, I think it is. I’d even speculate these two divides may be driven by the same difference in neural foundation. On reading the essay, I immediately recognized myself as a theorist rather than a problem-solver in Gowers’s terms; I think this may be connected to the fact that I’m more a design wizard than a master of complexity control.

  50. This is similar to my dichotomy in game design as well, save that no game is likely to be as complex as the source code for an operating system. Torvalds is a synthesist, Stallman is an innovator.

    I’ve worked at Steve Jackson Games as the print buyer, met his brother Ken at the company Christmas party, and managed to leave with the friendship with Steve intact. He and I both recognize each other as talented peers with very different approaches; sometimes I pick his brains, sometimes he picks mine. Our product niches are so wildly different that we don’t really compete for market space.

    I still marvel that Steve Jackson and Steve Cole managed to get a GURPS license signed for Prime Directive. Both of them have “lines I will not cross” in deal writing that are orthogonal to each other.

  51. > Right. And the folk theory I observed is that pure math is more critically gF-dependent that physics, which is in turn more so than programming.
    Were you thinking of that when you wrote this in the Jargon File? It seems like you are talking about gC vs. gF here:
    Although high general intelligence is common among hackers, it is not the sine qua non one might expect. Another trait is probably even more important: the ability to mentally absorb, retain, and reference large amounts of `meaningless’ detail, trusting to later experience to give it context and meaning. A person of merely average analytical intelligence who has this trait can become an effective hacker, but a creative genius who lacks it will swiftly find himself outdistanced by people who routinely upload the contents of thick reference manuals into their brains.

  52. Eric, I agree that all the evidence is that ingenuity fades with age, I would like to point out that, along with all the disadvantages of starting late–family, forty-hour work week, etc.–there is at least one advantage: an immersion in the real world, and hopefully the development of a good bullshit detector. Apparently your BSD was working at a young age, but I think it far more often that it becomes developed much later. (Obviously, some people never develop a BSD).

    I remember when I was a teen reading about how Bertrand Russell “refuted” the work of a mathematician (Frege, I believe) who had done some rather impressive work on the very foundations of mathematics (including, for instance, an actual definition of “number”) by asking, IIRC, if the “set of sets not members of themselves” was a member of itself. This led to an obvious paradox, and as a teen I was impressed by it. Later on, I read it, and although I admit I don’t have the mathematical knowledge of Russell, I can’t help feeling that this is just some word game along the lines of “This sentence is false.” I’m not sure why, but Russell’s paradox seems trivial, much like the “proof” that 2=1 from algebra class, obtained by surreptitiously dividing 0 by 0.

    My point: age can give wisdom.

    As far as music goes: I suspect that part of the reason modern music sucks is simply that we’ve come up against the limits of what can be done with music. For the past few centuries, we’ve been cranking out many times as much music as ever before. It may just be that the human race has run out of ideas, because there are only so many musical combinations that can be made.

  53. >Were you thinking of that when you wrote this in the Jargon File? It seems like you are talking about gC vs. gF here:

    Not particularly. As I said, this trait is not the same as high general intelligence.

  54. Apparently your [bullshit detector] was working at a young age

    It was. I know why, too; early study of General Semantics.

    >modern music sucks

    I don’t think so. As in all periods, you just have to be selective. I don’t experience any difficulty finding good music to listen to.

  55. > It may just be that the human race has run out of ideas, because there are only so many musical combinations that can be made.

    Dude. No. Just no. The awfulness of chart pop emphatically doesn’t mean that interesting, novel music isn’t being made. Check out some electronica – Boards of Canada, Aphex Twin and Four Tet are good places to start.

    > I don’t have the mathematical knowledge of Russell, I can’t help feeling that this is just some word game along the lines of “This sentence is false.”

    It’s more or less exactly that, but that doesn’t make it any less central or any less hard to work around if you care about foundations. Why is that? Well, mathematics could be (accurately, if strangely) described as a huge collection of such games, and Russell hit on a deep one.

  56. >I do find it generally to be true that people don’t go out of their way in finding new music after about 25, including some who advertise themselves at every opportunity as being into ‘alternative’ culture but are irreversibly stuck with e.g. 70s punk and can’t see much past that.

    You just described the contemporary “indie” scene. With a few exceptions, it seems that most of the bands in this genre are content to recycle Joy Division and XTC. (By the way, those happen to be two of my favorite bands of all time.)

  57. Regarding music, I think most people by the age of 25 have simply outgrown the things most musicians say. Or, as Beaumarchais put it, “Nowadays what isn’t worth saying is sung.”

  58. > The latest xkcd seems very appropriate. And true for a lot of geeks. :-)

    You do realize that Perl turned 21 yesterday, right?

    But hackers and computer scientists are normally different classes of programmer. ESR is a hacker. Nearly all 11th graders are hackers.

  59. On the subject of computer science .vs programming…

    I found this exchange to be a bit gruesome:
    https://lists.berlios.de/pipermail/gpsd-dev/2008-June/006057.html

    This despite the details of the algorithm being trivially findable via google:
    http://www.icao.int/anb/Panels/ACP/WG/M/M10/ACP-WGM-10_WP-09.doc

    yes, you (apparently) got an implementation going, but your unwillingness to search the literature is distressing.

    Of course, CRC was invented by a (real) computer scientist:
    http://en.wikipedia.org/wiki/W._Wesley_Peterson

    The publication of his 1961 book, Error-Correcting Codes, is regarded as the birth of algebraic coding theory, still the fundamental theory for error-correcting codes. Error-detecting and -correcting codes are indispensable for reliable digital communication, digital broadcasting and data storage systems. A second, revised edition of Peterson’s book, co-authored with E. J. Weldon Jr., was published in 1972 and is still widely used.

    […]

    Peterson says he had never seen a computer until after he got his PhD (in electrical engineering, from the University of Michigan)–and he has still never taken a course in computer science–but the machines seemed interesting, so he went to work for IBM in 1954. He was a visiting associate professor of electrical engineering at MIT, on leave from the University of Florida, when he began writing Error-Correcting Codes at the suggestion of a colleague. The book standardized terminology and notations for representing error-correcting codes. It established a framework of algebraic coding on the basis of modern algebra. Most important, at a time when few engineers were interested in algebra, Peterson’s book demonstrated that modern algebra offered precisely the right theory for the practical development of error-correcting codes. The original research that went into the book included a wide array of coding and decoding developments that are today involved in all computer disk drives as well as most digital communication systems. In addition, Peterson invented practical logic circuits for error detection and correction that became vital to industrial computer and communications applications.

    Peterson just happens to be a professor of mine (next semester): http://www2.hawaii.edu/~wes/

    So no, you don’t need CS courses to “do” computer science, but you do have to have a scientific bent, and you do have to be willing to do completely original work.

  60. Eric,

    The empirical correlation with age does not necessarily demand acceptance of age itself as the primary and direct causation.. I can’t cite any evidence to prove or disprove it, but it is possible that it is not age itself that causes the attenuation of creativity and tail of the bell curve skills. It may be other factors that tend to, but don’t necessarily have to, come with age.

    This is pure speculation, but a few possibilities come to mind. First, anyone wanting to excel in physics needs to be all about physics. Those who start their careers in the field have the most time and energy to devote to study, and few or no obligations that drain resources away from it. It’s difficult to do cutting edge physics in one’s spare time away from being a plumber. (A patent clerk, on the other hand, apparently has nothing but spare time, even when on the clock).

    This means that those who hit the ground running in full-time physics will either make some significant contribution in the first decade or two of their careers, or will move on to other pursuits, or even give up and rest on their tenure, having decided that physics in not their strong suit. Those who do make early contributions, by the nature of how research careers operate – being mostly a government run and subsidized pursuit – are often isolated from any external pressure to further achievements, and can easily lose the internal drive that got them there. Either way, there’s few people at advanced ages still making breakthroughs, but not necessarily because they are intellectually incapable.

    Those who choose later in life to pursue endeavors like physics will likely have to balance the time and energy they have for it against the needs of the obligations, interests, and habits they have accumulated to that point. This makes it difficult or impossible to devote the resources necessary for something that needs to be essentially all-consuming if there is to be much chance of success. Extreme talent can overcome this, and explain the few outlier cases of significant intellectual achievement later in life, but so could having the resources to not be held back by it, or even an unusual willingness (or “nothing to lose” circumstance) to sacrifice everything gained so far in life to it.

    There is also the factor that experience seems to often have a negative correlation with creativity. Once someone has accumulated enough “boxes”, staying inside at least one of them becomes easier. Its the path of least resistance to develop something one is already experienced in, as compared to doing something truly new. Young people don’t have any experience, so that path is no less resistant to them than working outside the box(es). I don’t think that is a function of age so much as a function of time, and that it can be overcome if one recognizes the issue in themselves and is determined to break out of it.

    So, it may not be biological age itself that hampers the intellectual conditions for creative work in demanding fields, but that decades of life lived hamper the external conditions necessary for such work. The effect may be the same, but the difference is that the latter could theoretically be overcome if the desire to do so and to accept the trade-offs is high enough.

    Or, it could be wishful thinking, seeing as how, if age is a direct causation, that opportunity is forever beyond me. Now in my early forties, I feel like my intellectual capacity would be up to it should I ever decide to pursue such a field, and that my experience and general knowledge may offer an advantage. However, the thought of devoting the effort required to singularly focus on such, for a decade or more, seems beyond any rational possibility. Still, it’d be nice to believe that its at least theoretically possible should I decide to make the hard choices necessary.

  61. personally, i’ve slowly come to the regretful conclusion that the “best work” in terms of Innovation was always done per-physicist rather than physicist-per time.

    that is, each physicist seems to bring their own perspectives to the field, and regardless of how late they start or how long they spend, they seem to bring the same amount of Innovation.

    ie: consider each physicist as a darwinian vehicle, as a experimental launch into the field of one viewpoint. if it fails, it fails. if it succeeds, it succeeds. each a spark launched into the dark by the genome. some fall on tinder. most don’t.

  62. >(Oh, and Seth is probably wrong. I’m not all that far south of Stallman’s level of ability, and probably comparable to Seth’s – but all three of us are way, *way* behind either Linus Torvalds or Ken Thompson, so much so that the differences between us three might as well be zero. In my humble opinion.)

    Eric! don’t you consider Peter Norton or Russ Nelson near the top?

  63. >Eric! don’t you consider Peter Norton or Russ Nelson near the top?

    Dunno. I’ve never actually seen any of their code. Which, now I think of it, is a bit odd given the amount of time I’ve known Russ.

  64. >I’ve never actually seen any of their code. Which, now I think of it, is a bit odd given the amount of time I’ve known Russ.

    Wow! How weird! I thought you and russ have had some joint hacking on gpsd.

  65. >Wow! How weird! I thought you and russ have had some joint hacking on gpsd.

    Russ wrote a Python client for a very old version of it, and hosted the project for a couple years, but AFAIK he never touched the core code. I’m sure he’ll correct me if I’m wrong about this.

    You’re quite right about the “weird” part, though. It is weird indeed that we’ve actually been co-developers on a project and I still don’t know what Russ’s coding style is like. Rumor has it his packet drivers are everywhere in point-of-sale and other sorts of turnkey systems, which certainly suggests that his code doesn’t suck :-).

  66. Apparently your [bullshit detector] was working at a young age

    It was. I know why, too; early study of General Semantics.

    This isn’t the first place I’ve heard good things about General Semantics; what (book?) would you suggest for followup on the subject?

    And for a further physics tie-in, what little I have read of it reminds me of a Richard Feynmann story:

    Nobel prize winning physicist Richard Feynman recalled an experience when he was a kid playing with a toy wagon that had a ball in it. He noticed that the ball would rush to the back of the wagon when he pulled on the cord, but when the wagon stopped the ball would roll to the front. He asked his dad why this was, and to Richard’s astonishment his dad said “nobody knows”.

    But this wasn’t the stone ages of physics, and there was already a name for what Richard was observing. So Richard’s dad explained further, saying “things have a tendancy to want to keep moving when they’re moving, or want to stand still when they’re not. This is called ‘inertia’, but nobody knows why it is.”

    Later in life Feynman remarked how wonderful it was that his dad knew the difference between knowing something, and knowing the name of something.

  67. >This isn’t the first place I’ve heard good things about General Semantics; what (book?) would you suggest for followup on the subject?

    I learned the basics from Wendell Johnson’s People in Quandaries: The Semantics of Personal Adjustment. It was a gift from my grandfather, who had been a student of Korzybski’s in the early 1940s. Later, I read Stuart Chase’s The Tyranny of Words. I have been told by reliable people, and have no reason to doubt, that S.I. Hayakawa’s Language in Thought and Action is as good as these.

  68. The Arne Dietrich article seemed pretty good, and it was published in a peer-reviewed journal. The reason the counterarguments that commentators made to Arne don't fly is because of Arne's claims such as "90% of all theoretical [physics] contributions occurred before the age of 40 and no theoretician over the age of 50 ever had an idea that was deemed worthy of the Nobel prize." If latecomers came "new" to the field at old age, and were still able to make paradigm-changing discoveries, then one would expect to find at least a few data points out there. But the curve tails off so abruptly, this seems unlikely.

    However, one problem with the study is that the Nobel committee is very reluctant to award two prizes in the same field to one person. The Arne study interprets this in support of their theory, however I think it's actually seen as sort of "bad policy" to give the same person multiple Nobels in the same field. For example, if you divided Einstein's major contributions among 5 people, they probably would've won 5 Nobel prizes, yet he was only awarded 1 Nobel. So this heavily biases their study against people who made a first Nobel prize winning discovery at a young age, and made a contribution worthy of a Nobel at an older age, that was not awarded due to the committee's preferences.

    I think the outliers definitely exist, for example you have Feynman at a later age introducing quantum computers, nanotechnology, and computer vision. But he isn't well known for these! One wonders whether there's a third confounding factor, such as people at a young age being more likely to champion their ideas and thus get credit for them. For example, when you look closely at new discoveries, it's often the case that many people independently had the same idea in that decade, but in retrospect, only one is usually assigned credit among the Czars of Knowledge for "making the discovery."

    Anyway, given Arne's model, which seems generally plausible, one wonders why these outliers exist. Did their prefrontal cortices degenerate less than other people? When you look at the outlier mathematicians, there seems to be this commonality that they all had "wild" creative ideas throughout their research life. Maybe the prefrontal cortex loses volume, but the creative brain circuits stay around if they continue to be used?

    I tried searching the literature for causes of prefrontal cortex degeneration, and related topics, and didn't find much outside of research into Alzheimer's, and some papers describing how magnetic stimulation could improve depression. This seems absurd. The decline of creative cognitive function among leading scientists seems to have flown under peoples' radar for "what is a disease." But this is irrational since it's of huge social import, it has tremendous consequences on human knowledge, and many of these ideas have world-changing economic value after the fundamental discoveries are applied. Rationally, it seems that a lot more research should be done in this area, to determine how valid Arne's model is and what the ramifications are to science.

  69. I also wonder whether the stereotypal lifestyle of physicists and mathematicians is actually leading to their creative decline. The stereotype is of having a very narrow experience in worldly affairs, and be rigidly, ascetically dedicated to advancing the subject by hook or crook. But maybe this is actually causing creative decline, and damage to the prefrontal cortex? Maybe instead of working out the incremental results of their previous important theory, a mathematician or physicist would be better served to take up a wild, challenging new hobby that involves sunlight, stimulation, the opposite gender, and whatever else wards off depression and monotony?

    To echo what Eric said earlier, maybe Eric's "trivia" about pursuing lots of new hobbies was what made him an outlier for having important, yet fairly reducible (axiomatic, logical) ideas at an old age? In a similar vein, I remember reading about "Richard Hamilton," who contributed the novel idea that led to the Poincare being solved by Perelman a couple years ago. Hamilton was described in the NYT as being into windsurfing, having lots of girlfriends, and being an atypical, "fun" mathematician, and he first published his idea for geometrization with Ricci flow at age 38!

  70. >To echo what Eric said earlier, maybe Eric’s “trivia” about pursuing lots of new hobbies was what made him an outlier for having important, yet fairly reducible (axiomatic, logical) ideas at an old age?

    Old thread, but…I think this is quite likely. I’ve had a horror of losing my mental flexibility, becoming a rigidified old coot, since I was in my 20s. A significant percentage of my life choices have been aimed at keeping my brain stretched and nimble.

    So far this seems to have worked exceptionally well, that is when I compare my flexibility and creative output to my age peers. I’m worried about the next ten years, though.

Leave a Reply to esr Cancel reply

Your email address will not be published. Required fields are marked *