transaction fee will never be enough to compensate with the current size...
The equation is pretty simple:
revenue = Pb * Hp * (BR + Tx)
Where Pb is the price of a BTC (in fiat);
Hp is the fraction of total hash power a miner has;
BR is the block reward; and
Tx is the transaction fees in an average block.
Obviously, to stay in business, a miner must keep revenue > costs.
All of the variables in that equation can change. The most certain one is BR, and that certainty depends on the protocol not changing.
As BR decreases, some of the other variables must increase for a miner to remain in business. And they all can change:
Bitcoin price (Pb) can go up as we see in the exchanges, but for this to happen, Bitcoin adopters can't be scared away, and this proposed protocol change would scare them, in my opinion.
Percentage of hash power (Hp) can change as well, when less-profitable minors drop out. For instance, when BR is halved next, typical revenue will drop by almost half, causing about half of the minors to retire, which will keep revenue constant for the remaining minors.
Tx is not significant now, but in 100 years or so, it will become more important. At this point miners will include as many transactions (with fees) as possible in blocks. Tx depends on 2 factors: the fee per transaction and the number of transactions. The plan is that Bitcoin will be used much more in the future, and thus the number of transactions will increase.
the mining business at some point will ask for block reward that cover cost and this will be the turning point to increase supply. and since they are the concensus they will have the power to dictate it. (most miner don't keep most BTC mine, but sell to pay cost so 20% drop in price for xM coins/Year supply will be acceptable for them)
That would be counter-productive. If the BTC supply were increased, we would have inflation, which would decrease the value of the BTC. If it gets to that point, then Bitcoin is finished.