There has to be something that can be done to look back at the chain's history, run the new algorithm, and see what the effect would have been over the last 3 or 4 months.
I bet you will discover the best path to take, but not only that, discover where abuses in the system (like botnets who all vote for the same poster) won't win more than they do already.
A lot of times, it seems that the best way to change these things, is to predict their outcomes. I wouldn't do that... The whole point of historical transactions is to be able to run new algorithms through history and see how it fairs.
That would be the best way of making your decision.
We can do that to simulate what rewards would be but it can never simulate changes in user behavior.
It can't, that's true. But what it could also do, is give also an idea of what type of reverse-engineering people might do with the new system, and consider what they might try to do to over compensate for the change to game the improved reward system.
The initial work to simulate things and notice patterns, trends, and possible "reward exploits" could be possible by using historical data and user patterns.
As time goes on, and since money and value is at stake, doing simulations is going to be mandatory in order to minimize any possible damage that could arise. We don't want to make buying/holding steem a roller coaster ride for investors everytime there is a hard fork. Or do we? :)
Stability with our changes and the value of STEEM, and minimizing problems needs to always be key. So if we can create tools to snapshot historical transactions and run them through algorithms, we should be doing that..
It would yield a lot of interesting data. I can think of about 50 steemit accounts that I'd be checking on, everything from a minnow, dolphin, whale, or bot for the new system. It would be interesting to see how the historical transactions run through the new reward calculation would work for / against them. :)
P.S. There are some bots, who have a nice hefty stake already (like 15,000 SP or more per node), so stake alone isn't a factor. I think reputation "IS" a factor that is hard to earn by bots.. Reputation should be a factor in the payout system. (But that opens a whole different can of worms -- but you never know)