It's not possible to have an uncapped block size. As blocks get larger they take longer to propagate through the network. Longer propagation means more forks, which means more work is wasted on forks which do not become canonical, which lowers the effective hash rate of the network and therefore the threshold for a 51% attack.
2MB is far too small, but it would be a gross mistake to overcorrect by removing the cap entirely.
> On an actually uncapped Bitcoin instance, Blocks (Block size) and transaction fees will grow exponentially
Transaction fees would drop to zero. The fee is a bid in an auction for scarce block space. If block space is not scarce then the winning bid is always zero.
2MB is far too small, but it would be a gross mistake to overcorrect by removing the cap entirely.
See, for example: https://www.gsd.inesc-id.pt/~ler/docencia/rcs1314/papers/P2P...
> On an actually uncapped Bitcoin instance, Blocks (Block size) and transaction fees will grow exponentially
Transaction fees would drop to zero. The fee is a bid in an auction for scarce block space. If block space is not scarce then the winning bid is always zero.