Why not simply require fees == 50 - subsidy before the block size can be raised by double? And if fees drop bellow say fees < ( 50 - subsidy ) / 2 the block size gets halved to min 1Mb.
"For every complex problem, there is a solution that is simple, neat, and wrong."
Miners can just put 100 BTC TXN fees in their blocks they're paid back to the miner after all, it costs them nothing but the risk that someone will reorg the chain to steal their block. Also why "50" by fixing 50 in the system you end up creating price distortion.
It's not wrong, because it's not the entire solution. I forgot to add that you also determine fees by checking every 2016 blocks what the average fees per block were. Now miners can't manipulate the fee anymore because they don't know who will find the next block, well unless they unanimously decided to insert fees enough to raise the limit, something really hard to achieve since if just one big miner doesn't agree the rest would be paying him to raise the limit.
I had made a off the cuff suggestion which was similar to your motivation there but lacked these risks: Set the new maximum blocksize to the median of the prior sizes multiplied by half the change in difficulty for increases in difficulty and by the full change for decreases, clamped to no smaller than 1MB. The idea is two fold: it's a cheat-proof way for miners to prove that an increase in size won't break the fee market because they have to work increase computing power to increase it, and it indirectly measures the progress of silicon: if computers are twice as fast per unit energy, difficulty should be twice as high ... and boring nodes could hand 25% more transactions. "If miners think that security can be maintained by increasing blocksize, then let them prove it by putting in more hashpower".
I don't like it because it only addresses the security spiral to the bottom problem. It doesn't address concerns about censorship, mining consolidation, enfranchisement of the users, etc. very much. It's also moderately complicated to understand. It's also not realistic until mining is primarily on modern asics otherwise it will overly measure the change in technology type rather than the returns on mining.
I don't like this idea because I don't think enough fees would be collected to provide for adequate security. What if hashpower becomes so easy to increase that the relationship you set as a rule for block size limit being raised doesn't even beging to cover anything?
Seriously Mike? People will use miners for heat? Man for a smart engineer you sure say some pretty stupid things.
My home was heated for two winters with mining. It is not the most energy efficient form of heating, if you ignore the mining income. But mining income is real and in very cold climates heat-pumps are not as effective. Ground loop for non-air heat sources for them are costly to install, etc.
I said it's retarded to suggest it in Mikes scenario when there is no(or close to no) income. There are better ways to heat than to run a miner on a loss.