Between P and P + H, where P is when the submission was Posted and H is some (constant?) number of hours, any downmods on the submission increment the downmod count but this has no effect whatsoever on the ranking. When we hit P + H, the downmods are made effective. An upmod on a story cancels a downmod.

Then if a story has lots of inappropriate early downmods but is actually interesting, it will get high on the front page and the mass of legitimate upmods will cancel the illegitimate downmods.

Showing the number of effective downmods (abs (- downmods upmods)) on a story might or might not be appropriate.

Apparently a lot of this has been addressed on reddit with some clever code. I think it's far cleverer to just leave the feature out!

At first this would cause a karma hit, so I tried a workaround: after my submissions peaked, I would delete them and this would present a smaller surface area for revenge attacks.

This turned out to be a bad idea because it deprived reddit of the history. So I switched to a new tactic: I simply stopped participating in the forums and started submitting things under a nom de plume.

I think that they have changed things so that revenge downmods do not affect karma, but it doesn't matter to me. I stopped one day and asked myself why I was bothering to worry about my reddit karma?

Then I asked why I was bothering with reddit?

Then I closed up the nom de plume and HN is the only such site I visit. And let me tell you, I am a far, far happier man for it :-)