... and I don't believe in LN being an alternative technology.
So what's your alternative? 500 GB blocks every 10 minutes? If you read my paper, you'd see I'm for a
reasonable increase, not extreme.
My alternative is like 10 times decrease in block time in short term and
improving PoW to a pooling pressure free version and providing a strong enough infrastructure for sharding in long term. I would propose a lot of more improvements meanwhile including and not limited to moving all signature data to witness space, using Schnorr signatures, ...
But as far as I understand, the main controversial issue with block size increase proposals is not their need for a hard fork rather it is their centralization implications because of progress and proximity consequences
I think I've highlighted this enough times in the paper, but I'd like to point out an important point again: you
can limit superspace blocks, by saying in the protocol the block is invalid if its size exceeds, let's say 10 MB. Increase from 1 MB to 2 MB in SegWit did not really affect decentralization so far, neither would 10 MB. 128 MB blocks definitely would.
Also, it doesn't mean that every block will be 10 MB from now on, only in the "rush hour".
Where did I get 10 MB number? Nowhere, it's just an example. We can set it to whatever limit the community deems reasonable. The point is that we
can impose it. But now we ourselves will be able to decide what the block limit should be in today's circumstances, not 2010's Satoshi when he initially put 1 MB, looking at 2-3 KB blocks as they were back then.
Setting 10 MB limit on superspace blocks is just the same as putting a 11 MB limit. Still you are offering nothing more than a complicated algorithm tweak to reach to a point that is simply affordable by a hard fork.
Don't want to undermine your work but I think it is just about avoiding hard forks by mimicking SegWit approach.
Honestly, I hate SW exactly because of its tricky approach, it looks to me kinda
cobbling things in a hacker way. I love hack but not when it comes to core algorithm, as a rule of thumb we should keep core components elegant and smart.