Statistical significance [LWN.net]

Statistical significance

Posted Jul 30, 2009 20:23 UTC (Thu) by man_ls (guest, #15091)
In reply to: Active discrimination, or feeling oppressed? by Baylink
Parent article: OSCON keynote: Standing out in the crowd

Your previous statement (FOSS communities are statistical noise compared to the larger scientific and engineering community) made little sense. Now what you say has no meaning at all.

Catastrophic events are not usually measured statistically: comparing them to the whole population is meaningless. Otherwise we would dismiss earthquakes as "statistical noise", since they happen very rarely at any point on the Earth crust. Or, say, nobody would have cared about the Tiananmen killings or the WTC attacks since the probability to die in any such events is "statistically insignificant" across the whole population. Nevertheless it may be interesting to put the whole thing in perspective, such as comparing it to the much worse death toll of traffic accidents, but a comparison cannot hide the horror of a massacre.

One death threat is one too much and should be repelled by any human being, be it in the whole US or in a small town, but especially from within Debian developers given what Debian stands for. What is the use of pointing out its "statistical insignificance"? I can only think that you are trying to hide the fact that they are horrific events, and that is sad.

to post comments

Statistical significance

Posted Jul 30, 2009 20:33 UTC (Thu) by Baylink (guest, #755) [Link] (5 responses)

Sure, cause that's a perfectly reasonable derivation to draw from everything I've said here.

You're makin stuff up now; quit it.

The point remains, though, that however you characterize a death threat, and howver you characterize "everyday" sexism, they have little to nothing to do with one another, and things which might mitigate the latter will have pretty much no impact on the former at all -- it's not a discomfiting behavior pattern based on a difference in perception, it's a *crime*.

Even the threat, in case people have missed that, is a crime -- a felony in some jurisdictions, depending on how it's phrased and delivered.

Here, though, it's just a red herring. Trying to tie it to other behaviors you consider sexist merely makes you look like you have no other rhetorical points left to make.

Statistical significance

Posted Jul 30, 2009 21:17 UTC (Thu) by man_ls (guest, #15091) [Link] (4 responses)

I am sorry if I misunderstood you but I honestly don't see how.

Anyway, you are right that I don't have many rhetorical points to make. The fundamental issue is this: there are very few women in free software projects and that is a problem. You may disagree about the reasons or about the remedies but if you don't see the problem then we don't have a common ground.

After this acceptance, we may start discussing whether death threats specifically made to women participating in Debian (according to the DPL: "harassing women for supposedly destroying the free software movement") are sexist or not. Once we determine that they are, we may try to link them to other sexist behaviors because, well, all of them are sexist -- that is the common link. But until you see the problem there is certainly no point in all of this.

Statistical significance

Posted Aug 2, 2009 22:33 UTC (Sun) by Baylink (guest, #755) [Link] (3 responses)

> there are very few women in free software projects and that is a problem

For the projects?

Or for the women?

If the former, show your work; I'll stipulate the latter as a problem (to the amazement of some of my detractors here, probably).

Statistical significance

Posted Aug 3, 2009 20:22 UTC (Mon) by nix (subscriber, #2304) [Link] (2 responses)

Excluding half the human race from any endeavour is a problem.

If you only listen to purely economic arguments, look at the leap in GDP
of most nations when it became common for women to have jobs.

It appears that we are, intentionally or not, excluding women from our
endeavour. Thus...

Statistical significance

Posted Aug 12, 2009 23:09 UTC (Wed) by linuxrocks123 (subscriber, #34648) [Link] (1 responses)

It's not half the human race. It's "the women who would participate in OSS development but aren't going to now because we do things they don't like". That is a fairly small number and, given current CS enrollment statistics, amounts to /maybe/ 10% of the current OSS developer base.

So, if we spent some enormous time and effort eradicating whatever these women consider "sexism" from all OSS projects everywhere, we'd get a 10% boost in developer numbers. We'd have to subtract from this all the men we lost because we either kicked them out for being "sexist", or they left because OSS development wasn't fun for them anymore.

I put "sexist" in quotes because it doesn't really matter whether the behavior is sexist or not based on any legal or reasonable definition of the term; all that matters is what these women consider sexist, which, importantly is likely to be a much more expansive definition. It might be impossible to reject a patch from a few female contributors -- for any reason -- without being called "sexist" by them. Since some patches need to be rejected, regardless of the gender of the contributor, we'd need to create some official definition of sexism and then have ways to try accusations. This administrative overhead would be an additional cost to be deducted from that 10% figure.

I don't doubt that there's some actual sexism in OSS. There are assholes everywhere. But, from a practical standpoint, sexism in OSS is, always has been, and always will be a complete non-issue. We're talking about volunteer groups, not companies with a legal obligation to pretend to care about this particular type of assholery, and these groups of volunteers have all naturally created behavioral standards that by and large work well for them. They'll expel the worst assholes without prodding, but most people learn when they are children that it is usually fairly easy to ignore assholes, so they'll let "productive assholes" stay on and just mentally killfile them. This is exactly the right response. Outside of groups like Debian that /enjoy/ doing everything with a tour-de-force, no one will ever care if a few whiners claim they're being discriminated against even though they're being invited to contribute with open arms -- just like everyone else -- and no one ever should.

Statistical significance

Posted Aug 12, 2009 23:27 UTC (Wed) by nix (subscriber, #2304) [Link]

We have repeated evidence of death threats being received. Thinking that
this is problematic is not 'whining'.