Do you consider murdering a thousand people you don't like to be better or worse than letting ten thousand randomly-selected people die because you can't be arsed to do anything about it?
It ought to be better. None of the factors of either option (murder, don't like, allow to die, randomly-selected, death due to apathy) are worth more than one human life. Thus, it is a simple question of scale. All the possible consequences - such as 'now people will be afraid if I don't like them', 'well, I can't be held socially or legally responsible for their deaths' - just do not outweigh 9,000 human lives.
That said, if I ever encountered this situation in real life, I would be immediately convinced that I had made a mistake in my reasoning, and would spend as much time as I possibly could looking for the alternative where nobody dies.
To break up the awkward silence at the start of a recent Overcoming Bias meetup, I asked everyone present to tell their rationalist origin story - a key event or fact that played a role in their first beginning to aspire to rationality. This worked surprisingly well (and I would recommend it for future meetups).
I think I've already told enough of my own origin story on Overcoming Bias: how I was digging in my parents' yard as a kid and found a tarnished silver amulet inscribed with Bayes's Theorem, and how I wore it to bed that night and dreamed of a woman in white, holding an ancient leather-bound book called Judgment Under Uncertainty: Heuristics and Biases (eds. D. Kahneman, P. Slovic, and A. Tversky, 1982)... but there's no need to go into that again.
So, seriously... how did you originally go down that road?
Added: For some odd reason, many of the commenters here seem to have had a single experience in common - namely, at some point, encountering Overcoming Bias... But I'm especially interested in what it takes to get the transition started - crossing the first divide. This would be very valuable knowledge if it can be generalized. If that did happen at OB, please try to specify what was the crucial "Aha!" insight (down to the specific post if possible).