If you want to elaborate on your theories about how transaction replacement will act as a cure-all which facilitates the dreams of everyone's skittles, SD bet, and goat cheese sale the world over, I'd listen.
lol
That's probably not the thread for it but I basically see the future of Bitcoin neither in mining supernodes (which would kinda centralize Bitcoin) and also not in some kind of probabilistic transaction verification (i.e. not every node has to receive/verify all transactions, as long as you can make guarantees about "enough" nodes having verified each subset). These would be two alternative solutions to the blockchain scalability problem. I believe we'll have payment services that can be completely untrusted but act as a brokers between their users. Every user of such a service would have a certain amount "locked in" for say a month. Locked in means that they agreed (without any risk) on a pair of transactions which cancel each other out unless they get replaced before the month is over. By sending the service a replacement version of the reverse transaction that does not pay the user back the full amount you can effectively pay something to the service in a secure and instant fashion without having to use the blockchain. You can also use this mechanism to settle bills with other users of the same payment service in a similar way and everybody can do a million micro-transactions without ever needing to do more than two transactions per month in the blockchain (the rest ist P2P, out of band).
Another intriguing idea is the
ultimate blockchain compression which would also solve the blockchain scalability problem.
Else I'm happy just to correct the mis-conception that I see a 'show stopper' in a modest block-size limiting the potential of the Bitcoin solution. It's just the opposite in fact...I suspect that it will be the main thing which could save it.
I apologize if I misunderstood you or came across as harsh, I just happen to see Bitcoin's main problems not on the technical level. I don't pretend to know the best solution to each and every shortcoming of the current implementation but I have yet to hear about a single one that's not being actively thought about and where I can't see at least a few convincing proposals to remedy the problem.