Hacker News Reader

Comment by cmenge | Hacker News Reader

It's not that programming skill (whatever that is) is bimodal, it's the effectiveness of the engineer in a typical development scenario. That stems at least partly from the fact that programming is hardly ever a one-person endeavor, contrary to running.

Since scaling a team increases friction and overhead, a small number of highly skilled engineers can be absurdly more effective than a set of teams, which also makes them more valuable, but not necessarily as extreme as in sports. Add preferential attachment to the equation (good engineers try to work with other good engineers / hire only top talent), and the U-curve starts to make sense IMHO.

View on HN

It also doesn't count for how low skill (inherent, trained, etc.) can have a negative impact on a project by incurring vast amounts of technical debt that will have to be repaid that outweigh any positive contributions made.

I haven't really seen any evidence that this is true.

The talent myth as Adrian describes it is one of stereotyping: there's no data to accurately measure skill in programming so we simplify things in our heads and believe you're either talented or not. This myth appears to directly support your claim: untalented individuals are a cost burden on the enterprise. If this were true there would be far fewer enterprises to match the number of actually talented programmers: that is to say you and I are not likely one of them.

The reality I've experienced suggests that deliberate process correlates strongly to quality. In the absence of true talent we can still write great software by leveraging processes and tools to wrangle complexity and manage defects. We see this in plenty of engineering disciplines. Why is software so special?

>The reality I've experienced suggests that deliberate process correlates strongly to quality. In the absence of true talent we can still write great software by leveraging processes and tools to wrangle complexity and manage defects. We see this in plenty of engineering disciplines. Why is software so special?

Because of the entry level.

Think of this joke: What do you call the guy who graduates from medical school as the bottom of his class? Doctor.

The point is that even though there is a worst, they still have to have gotten so far. Even the worst engineer had to do good enough to become an engineer, which is a far greater test of skill than what I've seen in some interviews, especially when those hiring are not technically experienced.

I think many who call this a myth happen to have rarely come across these people in technical roles. At worst they may have interviewed a few, but they were quickly cast aside. They've never met a person who managed to get into a senior position as a cargo cult programmer who can't explain even the basics of the frameworks they are using.

The point Adrian was trying to drive home is that for any quantifiable skill we can measure we find that there is a bell-curve across candidates. The over-whelming majority of people are average in every human endeavor that we can accurately measure in this way. The theory goes that software development is not different.

If we assume that the bell-curve is likely to exist then it's probable that only a handful of people exist in this current generation who are at the far-end of the curve. If the success or failure of an organization involved in developing software is dependent on the talent of its people alone then I'd rather speculate on futures.

If there is such a thing as the Mozart-of-programming you can't expect to hire a whole team of Mozarts. Any policy to only hire the best available programmers is upholding a culture of egotism and hubris. It's just not probable.

I've only seen this belief propagated by people who are more concerned with being better than everyone else. It's convenient that their finger only seems to point outwards. I don't like working with people whose sole talent is the ability to recognize the lack thereof in everyone else but themselves.

Reply on news.ycombinator.com