20MB will last for a couple of years, then we'll meet the same problem ad infinitum...
1) Not ad infinitum. Once we reach mainstream adoption growth will slow to roughly the global economic growth rate.
2) All that really needs to happen is exchanges have to make a few infrastructure changes to allow instant hard fork arbitrage. They have to make it so that in the event of a hard fork it will be really easy to trade old BTC for new BTC, and vice versa. To me that's a perfect solution. Hard forks should not be a major issue. If they are, then whatever is making that so is what needs to change. Bitcoin needs to be able to adapt easily, on a dime, as the market sees fit. Note: This doesn't imply changes will be less conservative than they have been; the market price will reflect the valuing of conservatism and the precautionary principle, only favoring changes when absolutely needed, and only favoring sudden radical changes in the event of a dire emergency. It's basically the entire wisdom of the market ready to make the best decision that reflects all the luminaries out there.
You've heard of prediction markets and how amazingly accurate they are? Well with Fork Arbitrage (FA) exchanges function as prediction markets for which fork will succeed. If you have insight, you stand to make a lot of money. If you have no insight, sit tight and whichever fork wins your money is worth the same as it was before. Even in the unlikely event of, say, a 70/30 split between old and new this remains true, because you automatically own an equal share of both ledgers.
As we discussed last time it requires a bit more than just exchanges. Wallets and other tools (merchant systems) would need to support it too (or people would need to keep all their coins on exchanges, which is unrealistic and undesirable).
In theory this could work but infrastructure-wise we are
far from it being realistic.