If the blocks become larger it implies that there is increased adoption and use of Bitcoin. A high volume of low fee transactions over the long run would actually pay out more for the miners compared to a high volume of low fee transactions. The idea that if we increase the blocksize over time as the technology allows as it improves would lead to the network collapsing is not supported by the facts.
If we had a sixty four megabyte blocksize limit for instance with the same fees per transaction payed out today it would already exceed the present block reward. It is not impossible to have blocks this large, in a decade from now it might even seem trivial. It would even take BIP101 eight years to reach this point, and the newly proposed Bitcoin Classic more then twelve years. Most of the other blocksize limit increase proposals are even more conservative then this, which is fine. My point being is that a high volume of low value transactions can pay for the networks security and this would not lead to the network collapsing as you claim, especially as the limit is slowly increased as our technology improves.
As I explained
HERE, as long as the blocks are not full, the fee income will always be a small percentage of the block reward, and it shrinks every 4 years. A reduced block reward + fee will make future miners only be able to mine a fraction of a bitcoin per each block (even fiat wise that means lots of money), greatly reduce the mining incentive, especially when most of the coins were mined
25-50 BTC per block fee is a much better incentive, and maybe the best incentive for the miners to continuously expand their operation, since it keeps every miner feels like an early adopter. But that is impossible to reach if we don't have block size limit
I think that you are wrong in this conception, It has been shown that we can have a fee market without a blocksize limit.