Increase throughput. There are many possible designs. Some soft forks, some hard forks. I like Lopp's goldiblocks proposal. James O'Bierne talks about $1000 self-sovereign nodes being a perfectly reasonable goal. 1MB blocks are not the answer. That was just a number Satoshi picked with no rationale. Why should we be stuck with it forever?
Discussion
Lopp sent me this presentation too in another discussion but I don’t understand how this is a robust solution:
nostr:nevent1qqstrmtk0xtd5jymyln660jl032ejs9k5yrv059mkk42dsf6y65yevqp0t4hz