First of all, I agree one of the stipulations should be that the account is active over the last N days/ weeks/ months.

From there, tie the distribution algorithm to the number of "unique upvotes" that an account receives from accounts above X rep that are still active (over the last Y months), multiplied by the weighted account SP (with a cap at Z SP).

Better yet, give unique votes that came from longer ago more weight (12 months ago could be weighted at 1.0, 11 months at 0.95, 10 at 0.9, etc.), so that it's harder to game it by creating new accounts to upvote oneself and also rewards people who've shown dedication to the platform.

Speaking of which, ENGAGEMENT with the platform should also be rewarded, so # of posts for the account could also be factored into the formula.

What does this all measure in the end?

The highest scoring accounts will be those that have had the most engagement with the platform, who've had the most unique accounts upvote them, and have stuck it out the longest with the platform. It's hard to game because of the "time since vote" weighting (an alternative solution is to only count votes from accounts that are N days old as a "unique vote") and the required rep (let's say something like 45, or so) to qualify as a "unique vote".

It should be obvious if an account gets hundreds of upvotes from hundreds of newly created accounts with 45+ reps (presumably from a few whale upvotes) that someone is trying to game it -- these could be flagged (either automatically or manually) and counted out of the distribution algorithm.