5 ms·

"Go to Goldman Sachs and ask them how many of their "rockstar" traders learned the craft at home on their free time. Or go to a top research hospital and ask th

"Go to Goldman Sachs and ask them how many of their "rockstar" traders learned the craft at home on their free time. Or go to a top research hospital and ask them how many of their surgeons did the same. Or go to a top architecture firm and ask them how many of their architects learned skyscraper design on the weekends. They will probably all laugh at you." The difference in all these fields is that the practitioner, to learn, needs to things that cost significant amounts of money (trading/building skyscrapers), cause people to live or die (surgery) or go to jail or be executed (law), or need to do things that require significantly altering the real world to create an artifact(setting up skyscrapers). Hence the need for supervision, and systematic training with a lot of 'apprenticeship' feel. In most programming, the real rockstars (say Carmack or Linus or whoever) are self taught, and did learn on their free time, because all you need is a laptop and the internet and you are ready to go and creating a significant artifact (a programming language, a game, an operating system) needs only knowledge (easily acquired from books and the internet) and time. Not to take away from your larger (and valid) point that company training can be effective. But your analogies are a bit 'off' and that affects the power of your argument (imho, feel free to ignore).

rayiner 14y ago
The comment to which I was replying stated: "At the 'rockstar' level, all learning is self directed." Parse this out logically. I read this as: "self-directed learning is a necessary condition for achieving rockstar level skill." My point is that in other fields, institutional training can create rockstars without self-directed learning. That is to say, self-directed learning is not a necessary condition. Your point is that programming is different from those other fields, because you can become a rockstar learning on your own. I.e. that self-directed learning can be a sufficient condition for achieving rockstar level skill. But that does not contradict my point, which is that it is not a necessary one.
- plinkplonk 14y ago
  "The comment to which I was replying stated: "At the 'rockstar' level, all learning is self directed." " In context, that quote sounds more nuanced. "Obviously neither statement is (completely) true. At the "rockstar" level, all learning is self directed. And rockstar employers expect and rely on that. But for everyone else I think it's reasonable that there be some level of support for education from employers, and some level of self motivation from good employees. Neither is going to make up for a total shortcoming in the other." I suspect if you add "(In programming)" before the sentences "At the rockstar level .. " etc it would make more sense,especially if those sentences are describing present reality than some imagined future. And as for "the fact that it's easier to learn programming on your own than it is to learn other fields doesn't prove that", sure it doesn't prove anything. But then if we are to descend to that level of hairsplitting, your quote about other industries training programs doesn't really prove the point that something similar would work in software/programming, and is equally 'neither here nor there'. It may be true that you can create rockstars by company training, so you have a valid point in that we should probably explore the idea, look for supporting/opposing evidence and so on. It still remains to be shown to be true,and ajross has a valid argument that "as of today company training programs haven't produced programming rockstars (vs competent devs)", so your comparisons to surgeons or whatever don't really counter ajross's statement. ajross seems to be describing the present reality (and extropolating off that) and you seem to be saying that a different reality is possible (but unproven). Not much of a conflict that I can see. (but hey ymmv, I freely admit I could be interpreting all this incorrectly) And just fyi, I responded to (what I thought was ) excessive rhetorical flourishes like 'brain damaged thinking' in your reply. That phrase conveys no information beyond your annoyance and detracts from your main (and valid imho)argument. You can make a solid point without 'name calling' (note quotes), especially here on HN. EDIT: rayiner changed his post from when I typed this reply(which is all right, I do that all the time), so this answer may not make as much sense as it should . Now his point is that "self-directed learning is not a necessary condition (to become a programming rockstar)". It is probably true and certainly an idea worth exploring. One possible answer to that, especially in a forum called Hacker News, is "Show us". (to those fond of logic, to prove that a universally quantified argument, "All X s are blah" is wrong, you need to prove that "Not all X s are blah" and that in turn becomes "There exists a Z that not blah" and you have to show that there does indeed exist such a Z - aka an existence proof, what can I say, I'm a nerd). And with that, I bow out of this thread. Have a nice day.
  - rayiner 14y ago
    > But then if we are to descend to that level of hairsplitting, your quote about other industries training programs doesn't really prove the point that something similar would work in software/programming, and is equally 'neither here nor there'. It's not hair splitting, it's the fundamental question of whether self-directed learning is necessary to achieve rockstar-level programming skills. This seems to be commonly assumed in the programming world, which I think is a bit of cultural brain damage. Indeed, I think the proposition that programming is somehow different from those other intellectually demanding fields in trainability is the exceptional proposition that has the burden of proof.
    - ajross 14y ago
      plinkplonk already pointed it out, but I'll say it again: this fact is commonly assumed in the programming world because it is empirically true. If that's "brain damage" then you have a strange definition for that term. We observe facts ("at the rockstar level, all learning is self directed") and make hypotheses to support a theory ("self-directed learning is necessary to achieve rockstar-level programming skills"). If you want to assert a theory counter to that, then I'm sorry but it is you that needs to provide some evidence to support it. And seriously: there is none. There are no "trained" rock stars out there that I've seen. Can you find just one? Someone who through diligent classwork or apprenticeship reached a top-tier position in their field? I've never seen it.
      - rayiner 14y ago
        First, even if it's empirically true that there are no trained rockstars in the programming field (which is a point I'm not going to concede), that doesn't mean that "at the rockstar level all learning is self-directed." There is an obvious difference between "all the rockstars I can see are self-directed learners" and "self-directed learning is necessary to become a rockstar." Asserting otherwise is a basic failure of logical reasoning. My theory is that programming isn't any different in this regard from other intellectually demanding fields except culturally. I think the assertion that it is different is the one that requires proof, or at least an explanation for the mechanism by which it might be different. As for self-directed learning, I'll point to Dennis Ritchie as an example. He's arguably somewhere in the middle between self-taught and trained. On one hand he started at Bell Labs concurrent with/immediately after Harvard. On the other hand, UNIX was something of a side project. Compare him to his father, who was a scientist at Bell Labs working on switching theory, and received his training in the AT&T engineering machine. What's different about the two men? Why is Dennis's field not trainable but Alistair's field trainable? My assertion is that there is no difference except the fact that there was an established framework for training engineers but not really one for training programmers.

rayiner 14y ago

The comment to which I was replying stated: "At the 'rockstar' level, all learning is self directed." Parse this out logically. I read this as: "self-directed learning is a necessary condition for achieving rockstar level skill." My point is that in other fields, institutional training can create rockstars without self-directed learning. That is to say, self-directed learning is not a necessary condition. Your point is that programming is different from those other fields, because you can become a rockstar learning on your own. I.e. that self-directed learning can be a sufficient condition for achieving rockstar level skill. But that does not contradict my point, which is that it is not a necessary one.

plinkplonk 14y ago
"The comment to which I was replying stated: "At the 'rockstar' level, all learning is self directed." " In context, that quote sounds more nuanced. "Obviously neither statement is (completely) true. At the "rockstar" level, all learning is self directed. And rockstar employers expect and rely on that. But for everyone else I think it's reasonable that there be some level of support for education from employers, and some level of self motivation from good employees. Neither is going to make up for a total shortcoming in the other." I suspect if you add "(In programming)" before the sentences "At the rockstar level .. " etc it would make more sense,especially if those sentences are describing present reality than some imagined future. And as for "the fact that it's easier to learn programming on your own than it is to learn other fields doesn't prove that", sure it doesn't prove anything. But then if we are to descend to that level of hairsplitting, your quote about other industries training programs doesn't really prove the point that something similar would work in software/programming, and is equally 'neither here nor there'. It may be true that you can create rockstars by company training, so you have a valid point in that we should probably explore the idea, look for supporting/opposing evidence and so on. It still remains to be shown to be true,and ajross has a valid argument that "as of today company training programs haven't produced programming rockstars (vs competent devs)", so your comparisons to surgeons or whatever don't really counter ajross's statement. ajross seems to be describing the present reality (and extropolating off that) and you seem to be saying that a different reality is possible (but unproven). Not much of a conflict that I can see. (but hey ymmv, I freely admit I could be interpreting all this incorrectly) And just fyi, I responded to (what I thought was ) excessive rhetorical flourishes like 'brain damaged thinking' in your reply. That phrase conveys no information beyond your annoyance and detracts from your main (and valid imho)argument. You can make a solid point without 'name calling' (note quotes), especially here on HN. EDIT: rayiner changed his post from when I typed this reply(which is all right, I do that all the time), so this answer may not make as much sense as it should . Now his point is that "self-directed learning is not a necessary condition (to become a programming rockstar)". It is probably true and certainly an idea worth exploring. One possible answer to that, especially in a forum called Hacker News, is "Show us". (to those fond of logic, to prove that a universally quantified argument, "All X s are blah" is wrong, you need to prove that "Not all X s are blah" and that in turn becomes "There exists a Z that not blah" and you have to show that there does indeed exist such a Z - aka an existence proof, what can I say, I'm a nerd). And with that, I bow out of this thread. Have a nice day.
- rayiner 14y ago
  > But then if we are to descend to that level of hairsplitting, your quote about other industries training programs doesn't really prove the point that something similar would work in software/programming, and is equally 'neither here nor there'. It's not hair splitting, it's the fundamental question of whether self-directed learning is necessary to achieve rockstar-level programming skills. This seems to be commonly assumed in the programming world, which I think is a bit of cultural brain damage. Indeed, I think the proposition that programming is somehow different from those other intellectually demanding fields in trainability is the exceptional proposition that has the burden of proof.
  - ajross 14y ago
    plinkplonk already pointed it out, but I'll say it again: this fact is commonly assumed in the programming world because it is empirically true. If that's "brain damage" then you have a strange definition for that term. We observe facts ("at the rockstar level, all learning is self directed") and make hypotheses to support a theory ("self-directed learning is necessary to achieve rockstar-level programming skills"). If you want to assert a theory counter to that, then I'm sorry but it is you that needs to provide some evidence to support it. And seriously: there is none. There are no "trained" rock stars out there that I've seen. Can you find just one? Someone who through diligent classwork or apprenticeship reached a top-tier position in their field? I've never seen it.
    - rayiner 14y ago
      First, even if it's empirically true that there are no trained rockstars in the programming field (which is a point I'm not going to concede), that doesn't mean that "at the rockstar level all learning is self-directed." There is an obvious difference between "all the rockstars I can see are self-directed learners" and "self-directed learning is necessary to become a rockstar." Asserting otherwise is a basic failure of logical reasoning. My theory is that programming isn't any different in this regard from other intellectually demanding fields except culturally. I think the assertion that it is different is the one that requires proof, or at least an explanation for the mechanism by which it might be different. As for self-directed learning, I'll point to Dennis Ritchie as an example. He's arguably somewhere in the middle between self-taught and trained. On one hand he started at Bell Labs concurrent with/immediately after Harvard. On the other hand, UNIX was something of a side project. Compare him to his father, who was a scientist at Bell Labs working on switching theory, and received his training in the AT&T engineering machine. What's different about the two men? Why is Dennis's field not trainable but Alistair's field trainable? My assertion is that there is no difference except the fact that there was an established framework for training engineers but not really one for training programmers.