I can see how "to align with the weaker party" can be a sensible goal for a society. However, in this case it directly conflicts with the presumption of innocence principle, which is an even higher goal for a just society.
So no, the burden of proof lies in whoever is making the accusations, whatever the (un)balanced of power between parties is.
What we should do is (1) not judge any party until there are enough facts on the table (of which we don't have many right now); (2) strive for an impartial account of the events; (3) judge only on and after evidence has been collected. If you want to support a cause, pushing for people to follow this recipe would work much better than pushing for "the weaker party is right until the stronger party proves themselves".
Because sexism in tech is a fact, this occasion is an opportunity to discuss it. Debating whether Horvath is right or wrong is not only futile, but is also potentially harmful to her, and that harm is far greater than GitHub's. Presumption of innocence is only a principle in legal disputes, and this, currently, isn't one, nor are we the jury. There will be no great injustice done if the assumption turns out to be false, but there is a lot of harm being inflicted on Horvath right now by people equating her position with that of a rich company and suspending any form of judgment. This does not only harm her personally, but the entire industry by implying that sexism in tech is disputable and pending "further data", which it isn't.
Given that sexism in tech is real and serious, that GitHub has a bit of a spotty background on that issue, and that GitHub does not deny any of the allegations, this is a perfectly good time to assume something bad has happened and discuss measures to correct the situation in the industry. If it turns out GitHub has fallen victim to the very real injustice of sexism in tech, I won't shed a tear for them, but will be excited to see them face the challenge and lead the way forward.
The presumption of innocence doesn't mean that anyone must either think or pretend to think that someone is innocent, any more than freedom of speech means that facebook can't take your youtube link down.
This is a good illustration of the point in the grandparent comment. This is simply not true, and I can't even think of any fallacy/ belief/ blindspot that could be adopted by a somewhat rational listener to make it sound convincing. (just to address the recalcitrant credulous among you: consider whether there is discrimination against Down Syndrome sufferers amongst tenured faculty on math departments)
In topics where people have deep vested personal interest or emotional investment, the quality of arguments that they would view as convincing goes down (often way down). The tragedy here is that the discourse about complicated topics which are important is overrun by people who have either (or, most often both) vested interests in a particular outcome, or little knowledge about the topic.
Saying "period period" at the end, I think, indicates that on some level, the parent commenter knows that s/he is presenting a weak argument. The poster's emotion overpowers this, however.
QED
The reason I put "period" is because many people aren't aware of the current body of research in gender studies and sociology, and modern definitions of sexism (a simple introduction can be found here[1], esp. unintentional sexism: "The tendency to use intent, rather than result, to measure whether something was offensive and inappropriate, and therefore sexist, is tied into male privilege and the way that it enables sexist practices to be seen as normal."). If women are largely absent from a group wielding any social power, then by definition sexism exists (or sex-based discrimination, if you prefer), unless you can prove that being a woman is a relevant debilitating disability, which I don't even think anyone conjectures. Because the data suggests[2] that diversity in Silicon Valley is severely lacking, this is a serious problem in Silicon Valley.
(Any rationalization such as "but what if women don't want to be in tech" is irrelevant. The assumption is that all groups want to take part in powerful institutions, and if they don't, it must be because somewhere along the line society encouraged them not to want that, thus effectively removing them from power.)
[1]: http://finallyfeminism101.wordpress.com/2007/10/19/sexism-de...
[2]: http://money.cnn.com/2013/03/17/technology/diversity-silicon...
Another case in point, it was a counterexample, not a comparison.
In your case, I now believe you are primarily just a sloppy/ bad thinker (who is also heavily invested in a certain position), but my original point about emotion clouding judgement still stands.
And, again, I'm sorry for my sloppy thinking, but I believe that "consider whether there is discrimination against Down Syndrome sufferers amongst tenured faculty on math departments" is not a counterexample because pertinent disability most certainly doesn't fall under any definition of sexism/racism and other unfair discrimination (same goes for quadriplegics in the NBA). Pitting disability against sexism as a counterexample appeared to me to be a brow-raising comparison. This fallacy is called reductio ad absurdum or a straw man.
If you'd like, I can come up with a better counterpoint (or so it would seem to my feeble mind, which I actually had to exercise for this): white players in the NBA. However, that, too, would not fall under the modern definition for racism, as racism (like sexism) is "prejudice + power", and I don't think anyone suggests white men lack power or influence in American sports.
And if I was sloppy by not qualifying the term "a well defined group of people" to exclude disability, forgive me, I thought it was patently clear from the context, namely a discussion on women. And please don't try to counter with discrimination against those who don't know programming or village idiots, because the definition excludes that kind of "discrimination", too, and I hope that's clear enough without me quoting vast amounts of research verbatim.
3,000 us army deaths in Iraq, and of those, 2,938 were men, 62 were women.
You are correct, but only if the term 'discrimation' includes 'discrimination based on skill levels.' If not, then you are assuming that power-relevant skill is distributed evenly, which is baldly false in many real-wrld situations.
And you are absolutely right. Sex discrimination is systemic, probably starting in kindergarten and even at home. However, it's hard to fix education if young girls don't have lots of positive role-models to look up to, and you can't make those role models unless education is fixed. So fighting sexism has to be systemic as well, and at all levels at once.
In the case of GitHub specifically, the issue is even clearer. This is a company that for years had a blog with little else than their drinking escapades. Perhaps that's changed, but the burden of proof is on them.
The only thing that is up for dispute is whether this particular case is an example of sexism in tech or not. We can't be 100% sure, but at this time it's safer to assume that the weaker party is right, if only because the stronger party can handle a quick-to-judge public much more easily than the weaker party. I'm not saying it's boycott time, but it's best to at least open the discussion with that assumption in mind. Again, the burden of proof is on GitHub.