Second, because the Bitcoin network is a peer-to-peer network, nodes are free to leave and re-join at will. The time to re-join the network is a linear function of the size of the blockchain (since the Genesis block). Right now, that is 140GB and I think it takes more than 24 hours to sync a full-node on a typical desktop PC. The blockchain will continue to grow at a rate of about 50-100GB per year with the current blocksize limitations; while the sync-time will grow in direct proportion to this, at least we know how much it will grow. If this were unrestricted, the sync time could actually fall behind an "event horizon" where it takes longer than 24 hours to process 144 blocks (24 hours worth of blocks), meaning no new node could ever join the network!
This is a pretty solid reason to not go for the block-size increase. Is there somewhere one can find proof to this. Like the work involved in syncing blocks and time a typical computer would take for that.
This way, we can know what hardware acceleration will support or be necessary to allow and increase in the block-size without these issues.