Thoughts about capping rewards in an incentivized way

I was thinking about the capping of rewards on posts, to make more "room" for smaller posts to get something from the reward pool and prevent the piling of whales on the same post. 

I think it was dantheman writing that the first whale gets something like 80% in curation reward, so whales don't have the same incentive to upvote when there's already some heavy upvoting, yet the piling up continues up to the thousands.

In theory this is fixable by curation (whale downvoting to reduce payouts) but in practice it is problematic (reputation, less incentives for downvoting compared to upvoting and getting curation rewards etc). 

Now caps are a bit "communistic", so instead of saying "a max cap of 1000 or 5000 should be introduced", I was thinking alternatives. 

One of these alternatives involve a voluntary self-capping, through incentives:

Example: The platform offers a maximum cap of "unlimited", "9999", "4999", "999", "499", "199" that the author wants to cap himself.

These go like these:

If authors are aiming at (unlimited) / 10000+, they get a multiplier of 0.35x on their earnings

If they are aiming at 9999 max, they get a multiplier of 0.5x on their earnings 

If they are aiming at 4999 max, they get a multipler of 0.6x on their earnings

If they are aiming at 999 max, they get a multiplier of 0.75x on their earnings

If they are aiming at 499 max, they get a multiplier of 0.85x on their earnings

If they are aiming at 199 max, they get a multiplier of 1x on their earnings 

(the numbers and limits can be adjusted - they are just an example)

So the lower the cap that one selects, the higher the multiplier of one's earnings: The different multipliers of the range provide the motive to impose a cap on yourself in order to increase your own multiplier. The lower cap you use, the higher the multiplier you are getting. This gives a sense of control on the process.

Now, say an author writes a post and are are aiming at unlimited payouts. They get a multiplier of 0.35x on their earnings. So if they really hit 10.000, they get 3.500. But if they hit 100.000, they keep 35.000. It might sound a ripoff but if there were caps involved at, say, 1000 or 5000, then that money beyond 1000 or 5000 would be lost anyway. Remember: This is a cap-alternative system, so it will by necessity involve reward reduction in huge rewards.

Likewise, if an author declares a post in the 999 max category, and he gets upvotes for 1500$, then he's maxing-out at 999 - he doesn't get to make more than 999 because that's his desired maximum. (To hit 999 you'd need 999/0.75 = 1332$ worth of upvotes.) 

In a way a mechanism is created where users try to "self-penalize" themselves the least possible (maximizing their multiplier) while also aiming at what they consider realistic. 

This could also be marketed in a positive way by doing the maths the opposite way (it has the effect that instead of appearing self-penalizing, it appears self-rewarding for not being greedy).

Spin-offs of the above would be to go with a various reduction curves (from very steep to less steep) after one hits a limit...

I haven't really figured any drawbacks besides (1) difficulty in determining the default value for new users (some celebrities would have a problem) and (2) introducing extra complexity. I'm not in favor of solutions that require more than 10-20 seconds to explain in terms of user interface. 

Anyway, just throwing ideas on the table. They might be useful to those designing the game-theory components of the system.