The dynamic is not one in which when miners win, users lose. The issue is that users want more transactions per second and therefore want bigger blocks to be created, however, when miners create larger blocks, they have a hard(er) time propagating them to the entire network and thereby run the risk of their block being orphaned (someone else's block will be accepted first, and the miner's block and work done on that block will be wasted).

I'm fairly certain that the problem is that users with catastrophically slow or spotty internet connections are the ones that stand to lose in that situation, and that > 50% of Bitcoin's miners are in China, which partially explains the rabid past support for a refusal to increase the max blocksize.