6 ms·

I disagree. It is one those things that are said once, repeated enough and make enough sense to be believed, although I suspect they are not true. If you combin

I disagree. It is one those things that are said once, repeated enough and make enough sense to be believed, although I suspect they are not true. If you combine an average programmer with an average statistician, you get a profile that is not particularly attractive. If the comment that follows is: but is it not what data scientist are? the answer is that good data scientist have good (or great) programming for what the programming they need to do. My undergrad and Master's are in an interdisciplinary field, which had at the time felt needed by Universities offering those degrees, but in the end the professional profile was not of interest. An average knowledge of chemistry, engineering, physics, and other sciences does not make you an interdisciplinary scientist, but someone who can do a little bit of many things, but not at the level that is required in order to get paid for it. The Dilbert's creator has excellent drawing abilities for the comics he is drawing —- here the lesson is that you need to do excellent relative to the quality that is needed, not that if you combine average talents you get world-class results.

phkahler 6y ago
>> If you combine an average programmer with an average statistician, you get a profile that is not particularly attractive. Nobody said being average at both was enough. The parent post specifically said if you know more than average in both areas. You still need to be good, but the combination of good in multiple areas is claimed to be sort of like being extraordinary in one. I think there's some truth in that.
- mywittyname 6y ago
  I still think the parent has a strong point though. Stats and CS are correlated skill sets. Almost all CS people learn a good bit of statistics as part of their education process, and I suspect that most statisticians are passable programmers. The fact that we have a term for the combination of both skills suggests that being great in both is not unique, it just makes you a great data scientist. The crux behind the idea of being better than average at two things makes you individually extraordinary is that the two skills are not correlated. An above average singer is probably an above average player of one or more instruments, but I'm sure there aren't many who are also great ice cream makers.
  - coryrc 6y ago
    > Almost all CS people learn a good bit of statistics Oh how I wish that were true
  - gloatyegret 6y ago
    > Almost all CS people learn a good bit of statistics as part of their education > The crux behind the idea of being better than average at two things makes you individually extraordinary is that the two skills are not correlated. I don't know if "almost all CS people" come from better schools than mine, but I've never had a proper statistics course, and my degree has a "data science" stamp on it. Having good statistics skills, and not just good knowledge of some machine learning frameworks, is probably rarer than you think.
    - phkahler 6y ago
      >> but I've never had a proper statistics course, and my degree has a "data science" stamp on it. How do they justify that? ML is not IMHO the same thing, but maybe times are changing?
      - gloatyegret 6y ago
        I've given up trying to understand. The school focuses heavily on software development, and they don't seem willing to change the curriculum too drastically to make room for a good data science track. But they want to have one anyway... I wondered if my degree was an exception.
patmorgan23 6y ago
I think the idea is be the a good (better than average) programer and good statistician radther than being a amazing top 1% statistician or programmer.
devtul 6y ago
Curiously enough Scott Adams(Dilbert creator) talks a lot about what he call the "talent stack", which is what we are discussing here. If you have a combination of skills that have synergy then you can have great results.
ska 6y ago
I suspect you are looking at this the wrong way. Someone who is average at a number of things is probably a decent generalist; valuable but not particularly are. But the argument is being above average in two or more areas can make you as impactful as being top of your field in one area, even though you are nowhere near that talented in any different field. Even with your example, I would counter that a huge number of people successful in data scientist roles would not be successful (or at least, much less so) either as programmers or statisticians. It seems many of the earliest wave of data scientists (e.g. pre bootcamps etc.) have failed to thrive at at least one of the two before ending up in "data science" (caveat, agree this is poorly defined).
- borroka 6y ago
  A decent generalist that nobody may want, though. You are right that "successful in data scientist roles would not be successful (or at least, much less so) either as programmers or statisticians", assuming they do not train a bit more and become better statisticians (a subject in which world-class experts have wildly different opinions about the very foundations of the field) or programmers (a 1X, say). What I had maybe issues articulating is that the Dilbert's creator is focusing on the system of skills (and, what does "better than average" mean? Better than the population average or better than the average of people professionally working, say professional comic artists? Because we are talking about two different things) instead of the application of the skills. He had a brilliant (a posteriori) idea that did not require to be neither Caravaggio nor Dave Chapelle. The lesson is that you can have success drawing like a 6-year old (which is better than what I can do) and some deadpan humor about office work which is funny one every ten times. Which is, IMO, way below the average skills of professionals. His "lesson" is between "nothing new" and "so what": nothing new because we have seen many successful people having near-zero talents or skills (that would be a long book for me), and so what because it is not the above average skills (and I ask again: average of what?) but the more or less fortuitous choice of a profession or activity. We have vague-casting, vague-posting, and, in this case, vague-philosophizing.
  - ska 6y ago
    A decent generalist that nobody may want, though. Right - but I think that pretty clearly wasn't what he was talking about. I don't love Adam's formulation, but it is close to something real.

phkahler 6y ago

>> If you combine an average programmer with an average statistician, you get a profile that is not particularly attractive. Nobody said being average at both was enough. The parent post specifically said if you know more than average in both areas. You still need to be good, but the combination of good in multiple areas is claimed to be sort of like being extraordinary in one. I think there's some truth in that.

mywittyname 6y ago
I still think the parent has a strong point though. Stats and CS are correlated skill sets. Almost all CS people learn a good bit of statistics as part of their education process, and I suspect that most statisticians are passable programmers. The fact that we have a term for the combination of both skills suggests that being great in both is not unique, it just makes you a great data scientist. The crux behind the idea of being better than average at two things makes you individually extraordinary is that the two skills are not correlated. An above average singer is probably an above average player of one or more instruments, but I'm sure there aren't many who are also great ice cream makers.
- coryrc 6y ago
  > Almost all CS people learn a good bit of statistics Oh how I wish that were true
- gloatyegret 6y ago
  > Almost all CS people learn a good bit of statistics as part of their education > The crux behind the idea of being better than average at two things makes you individually extraordinary is that the two skills are not correlated. I don't know if "almost all CS people" come from better schools than mine, but I've never had a proper statistics course, and my degree has a "data science" stamp on it. Having good statistics skills, and not just good knowledge of some machine learning frameworks, is probably rarer than you think.
  - phkahler 6y ago
    >> but I've never had a proper statistics course, and my degree has a "data science" stamp on it. How do they justify that? ML is not IMHO the same thing, but maybe times are changing?
    - gloatyegret 6y ago
      I've given up trying to understand. The school focuses heavily on software development, and they don't seem willing to change the curriculum too drastically to make room for a good data science track. But they want to have one anyway... I wondered if my degree was an exception.

patmorgan23 6y ago

I think the idea is be the a good (better than average) programer and good statistician radther than being a amazing top 1% statistician or programmer.

devtul 6y ago

Curiously enough Scott Adams(Dilbert creator) talks a lot about what he call the "talent stack", which is what we are discussing here. If you have a combination of skills that have synergy then you can have great results.

ska 6y ago

I suspect you are looking at this the wrong way. Someone who is average at a number of things is probably a decent generalist; valuable but not particularly are. But the argument is being above average in two or more areas can make you as impactful as being top of your field in one area, even though you are nowhere near that talented in any different field. Even with your example, I would counter that a huge number of people successful in data scientist roles would not be successful (or at least, much less so) either as programmers or statisticians. It seems many of the earliest wave of data scientists (e.g. pre bootcamps etc.) have failed to thrive at at least one of the two before ending up in "data science" (caveat, agree this is poorly defined).

borroka 6y ago
A decent generalist that nobody may want, though. You are right that "successful in data scientist roles would not be successful (or at least, much less so) either as programmers or statisticians", assuming they do not train a bit more and become better statisticians (a subject in which world-class experts have wildly different opinions about the very foundations of the field) or programmers (a 1X, say). What I had maybe issues articulating is that the Dilbert's creator is focusing on the system of skills (and, what does "better than average" mean? Better than the population average or better than the average of people professionally working, say professional comic artists? Because we are talking about two different things) instead of the application of the skills. He had a brilliant (a posteriori) idea that did not require to be neither Caravaggio nor Dave Chapelle. The lesson is that you can have success drawing like a 6-year old (which is better than what I can do) and some deadpan humor about office work which is funny one every ten times. Which is, IMO, way below the average skills of professionals. His "lesson" is between "nothing new" and "so what": nothing new because we have seen many successful people having near-zero talents or skills (that would be a long book for me), and so what because it is not the above average skills (and I ask again: average of what?) but the more or less fortuitous choice of a profession or activity. We have vague-casting, vague-posting, and, in this case, vague-philosophizing.
- ska 6y ago
  A decent generalist that nobody may want, though. Right - but I think that pretty clearly wasn't what he was talking about. I don't love Adam's formulation, but it is close to something real.