I agree that 800 is too much, and appropriate for banning.
Since you can "unvote" any particular karma vote you've made, wouldn't it be easy enough to implement limits on downvotes of a particular person in day, week, month, year?
You reach your max, and the next time you try, you are prevented, and you get a message saying "It is a bannable offense to karma bomb other users". That could be a rollover and a triggered message sent to your account.
(Note that the limits could be parameterized in increasingly complicated ways (scaled to karma of "victim", perhaps). The point is not "the perfect set of limits", but to find something better than the current limits. The problem can be ameliorated, not annihilated in all hypothetical cases. Life is full of tradeoffs. )
Problem limited and offenders who try to game the system are warned (I think the second is important too).
That should take care of all but the most committed douchebags without any required intervention from you.
As one of the Powers That Be Who Does Something Useful Around Here, I'd hope that your needs in your chosen useful duties would have pull with the feature development queue.
(EDIT: Maybe easier to run a nightly scan notifying people when they have gone over their limits. )
I've gotten sufficient evidence from support that voiceofra has been doing retributive downvoting. I've banned them without prior notice because I'm not giving them more chances to downvote.
I'm thinking of something like not letting anyone give more than 5 downvotes/week for content which is more than a month old. The numbers and the time period are tentative-- this isn't my ideal rule. This is probably technically possible. However, my impression is that highly specific rules like that are an invitation to gaming the rules.
I would rather just make spiteful down-voting impossible (or maybe make it expensive) rather than trying to find out who's doing it. Admittedly, putting up barriers to downvoting for past comments doesn't solve the problem of people who down-vote everything, but at least people who downvote current material are easier to notice.
Any thoughts about technical solutions to excessive down-voting of past material?