Jonathan Toomin on SegWit

[arnoudk]

From: https://www.reddit.com/r/Bitcoin/commen ... ?context=3

why is all this fighting going on?

Because SegWit is not a good short- or medium-term solution to the scaling problem. SegWit has a lot of good stuff in it, but the element that allows for immediate increases in capacity -- the 0.25x discounting function for witness data -- is not a good thing.
SegWit provides 1.375 to 1.75 MB of capacity in the near future (for 50% to 100% wallet adoption). It also comes with a 4 MB adversarial condition. This means that miners and network engineers have to design their systems to be able to handle 4 MB blocks without bogging down, but we only get to actually use about 40% of that capacity.
This 4x adversarial case will make it very difficult to increase the blocksize limit in the future. With a 1 MB base block size, it's 4 MB, but with a 2 MB base block size, the adversarial case would be 8 MB.
On the other hand, with a simple block size increase hard fork, we get 2 MB of capacity with a 2 MB adversarial case.
The 4x adversarial case is a direct result of the 0.25x discounting function that the current version of SegWit uses. The way this works is that SegWit says that one byte of data in the base block (what we currently think of as the block) is equal to one byte, but that one byte of data in the witness block is 0.25 bytes. This is an accounting trick, not compression: each byte in the witness block still has to be sent over the network as one byte, and it still takes up one byte of disk space; the only place where we are counting it as 0.25 bytes is in the accounting for the block size limit.
I don't like this accounting trick. I think it has harmful effects on the network's ability to scale. I also see it as a sort of tax subsidy for Lightning Network and sidechains. I don't like that subsidy. I would rather see those technologies compete on their own merits, not because we changed the fee rules to be biased towards them.