If you don't want to apply a label to the algorithm, how about "Foo Bar's decision to run Algorithm X against Dataset Y resulted in consequences that predictably increased prejudice, so it would have been better if he hadn't done it"?
If that's okay, note that this is pretty close to what parfe said -- he's clearly upset at the fact that pg chooses to run an algorithm on Hacker News, rather than at the algorithm itself.
Let's assume that it's a bad design decision. Your logic behind it being prejudiced is still faulty, as it ascribes a fallacious cause.
The algorithm merely shuffles page rankings based on two criteria: upvotes and comments, both of which are user-supplied. The algorithm does not support prejudice, it simply behaves in the fashion it is programmed to. You're completely ignoring all of the mundane stories that go down due to the same factors, and cherry picking the ones about sexism.
Time for dinner. heh.
Don't blame the algorithm for the fallacies of the people using it. The only fault in your scenario is the PEBCAK.
All known halting program/input pairs are inherently a little bit prejudiced.
Either way, it absolutely does not apply here. HN's algorithms sinking sexism-related stories is a side effect. The fact of the matter is it will sink any story that has an imbalance in the upvote-to-comment ratio, whether it's about sexism, Haskell or the Church of the SubGenius.
This is the same logic as saying that encryption is evil because it can be used by terrorists to hide information. Encryption always does the same thing, it's fundamentally neutral, regardless of what's being encrypted. The same way HN's algorithms are fundamentally neutral, regardless of what happens to trigger their procedures.