[rhacker]

Wouldn't this make sense?

https://medium.com/@hoffa/400-000-github-repositories-1-bill...

So the number of people that prefer spaces far outweighs the tabbers.

Now, take the pool of people that use spaces and tabs and let them compete for a 100k job. 9 times out of 10 even if randomly hired it will be a space person.

Also, as a by the way, on the medium post. The reason Java has so much tabs is specifically due to the default Eclipse formatting that prefers a mix Tab/Space approach. Which is absolutely horrifying in my personal opinion. I don't have any numbers but Eclipse probably holds 80%, followed by Netbeans and/or IntelliJ. (both of which default to spaces)

[NobodyNada]

> Take the pool of people that use spaces and tabs and let them compete for a 100k job. 9 times out of 10 even if randomly hired it will be a space person.

But you'll get the same result when they compete for a job that pays poorly. If all other factors were equal, the distribution of developers who prefer tabs should match that of developers who prefer spaces.

By your argument, people named Steve would make less money, because they're outnumbered by people not named Steve.

[hencq]

This could still be true though. There are more likely to be outliers in the large group than in the small group. However, there is a lower bound on the number (nobody makes below $0), but no upper bound. So the distribution is likely to be skewed.

So yes, people named Steve would make less money (though this argument might not work as well for people named Jeff, Warren or Bill I guess).

[Grustaf]

I am not sure exactly what your argument is, but it doesn’t sound correct. The differences will be larger in the smaller group, but the expected value for the average pay in a group is independent of its size. That’s why you can use a sample to find out something about a population.

[rhacker]

Very true, but something is still askew - the same will happen for women, minorities, etc.. Any time a large group and a small group, technically the small group tends to get less pay! What's going on? (I totally concede to your point, but now I'm left confused and wondering!)

Edit - ahh, hencq has a reasonable explanation for that.

[coldtea]

Well, women and minorities are totally different case, not some general case of "the small group tends to get less pay".

For one, women are more than men (slightly, around 51%, but more. And if not more, equal). So it's not true that "the small group tends to get less pay" in this case. Women are less only in some jobs (like IT). There are more women than men in others (e.g. in education it's over 60% women). But even in those jobs women might be paid less than men.

Second, women and minorities being paid less can be explained by many factors, from discrimination, to women (either naturally, or by nurture) not pursuing promotions/raises aggressively enough, to exploitation (e.g. since some minorities are in bad condition socially, or might be afraid of deportation etc, one can exploit that to pay them less, "take it or leave it"). These practices of course can not be the case with tabs/spaces.

Third, there are countries where the majority gets less pay. E.g. in South Africa the whites are paid more than blacks, same for other countries with a smaller more privileged citizenry and a wider base of rural/indigenous/colored/etc population.

[coldtea]

>So the number of people that prefer spaces far outweighs the tabbers. Now, take the pool of people that use spaces and tabs and let them compete for a 100k job. 9 times out of 10 even if randomly hired it will be a space person.

Err, is the distribution is only affected by the number in each category (and not e.g. tab/space preference), then you'd be able to sample the same results for 30k jobs. 9 times out of ten it will be a space person there too.

Thus, if the 28,657 survey respondents are also randomly distributed, the ratio of poor vs good paid would be the same as real life, as would be the ratio of tabs vs spaces.

In other words, one wouldn't expect a skew in one category in favor of the other (if the choice wasn't a factor). Of course correlation != causation, but correlation can point to causation, or to a third, unknown factor that correlates with both attributes.