Let me think, and please correct me if you spot an error:  when a pool finds a block, payouts are based (with variations in the scoring formula, transaction fees, and pool fees) on 25 * work_i_did / work_pool_did.  Say I'm a small miner in a medium-sized pool who sees ~0.01 BTC per round, meaning I'm doing about 0.04% of the pool work, and the pool finds an average of 36 blocks a day (that is, about one out of every four blocks) and I find a block once a year on average.

The proposed reward system is to give 18.75 BTC to the finder who then gets no credit for the work the finder contributed to the pool for that block.  That leaves 6.25 BTC to split between me and the other miners.  Now, work_i_did / (work_pool_did - work_finder_did) is still very close to 0.04%, which means I could expect a payout not much over 0.0025 BTC each round.

Disregarding changes in difficulty and network hash rate, under the current system, I make 1 BTC every 100 rounds (2.78 days) or a total of 131 BTC per year, under the proposed system it would take me about 400 rounds (11,1 days) to make 1 BTC, or 32.85 BTC per year plus the 18.75 I'd get for finding my one block = 51.60 BTC a year.

Nope, I think I'll stick with the current system.