We are currently averaging at around 1.6 MB per block. Removing the witness discount would bring us back to 1 MB blocks. This would be a 37.5% decrease in L1 transaction throughput. Over the past week we have been seeing a ~16% reduction in block production rate. (presumably due to the US winter storm) Despite this fees are still very low. Yes, a 37.5% decrease would increase fees more, but I wouldn't go as far as saying that fees would immediately explode to levels of 2017 or April 20th 2024.

What should then be the next move after that?

After such a change L1 would probably become more of a final settlement layer and less useful for small payments. So we should focus on improving L2 scaling solutions like Lightning or Ark.

If you think a blocksize decrease or removing the segwit discount should be considered, why do you think that would be expected to lead to an improvement? And what improvements and other effects would you expect?

I know this question wasn't directed at me, but here are my thoughts anyway:

I think that the witness discount distorts incentives, since it makes inscriptions significantly cheaper compared to any other kind of transaction. Large inscriptions also lead to bigger blocks.

If the witness discount were removed, data embedding would probably shift to the less harmful OP_RETURN outputs, since this would then be the cheapest way.

I agree that the BIP-110 approach is not perfect. I would prefer a removal of the witness discount. But I am also not going to oppose BIP-110, since it would still reduce inscription spam.

Or why not have a soft-fork to remove/reduce the segwit discount?

Yes, I would really be in favor of this!

My understanding is that the bloating of the utxo set is a far greater risk than the bloating of blocks... which runes memecoins could still cause.

UTXO set bloat becomes less of a problem with Utreexo.
Also OP_RETURN outputs are not part of the UTXO set.

Let's all love Lain

I agree that UTXO set bloat is causing problems for node runners now, but in the long run this should be solved by Utreexo.

And yes, higher fees discourage spam. But inscriptions benefit from a 75% witness discount. JPEGs on the chain are significantly cheaper than regular monetary transactions. That's a misalignment of incentives if I've ever seen one.

Bip-110 has an 83 byte op_return limit right? So it doesn't stop Runes at all

Fine by me. I don't see OP_RETURNs as the problem.
Inscriptions are the problem.

No they aren't... but someone can 'pretend' than an op_return creates a 'token' therefore 'pretending' that one op_return and lots of outputs... means lots of tokens. "Runes" memecoins do this but there are unlimited arbitrary possibilities.

Yeah, they can pretend but Bitcoin nodes do not care about Runes. It's just an unspendable script with an OP_RETURN and some unimportant data to them.

Blocks are bigger due to the 'witness discount', witness data weighs less than transactional data

Correct. This witness discount is exactly what inscriptions make use of, causing blocks to balloon up.

Blocks aren't bloated due to inscriptions... maybe the UTXO set is, but not blocks themselves. Blocks are the same size.

Blocks might have the same weight but one look at a block explorer shows that blocks are not the same size in terms of bytes:

And UTXO set bloat can and does occur with 10 byte op_returns (with bip-110 doesn't fix) so imo bip-110 changes nothing.

OP_RETURNs are not part of the UTXO set.

true

What does PeerSwap do?

• Multiple optional swap types.
  • Currently BTC & L-BTC onchain swaps.
  • Additional wallet options can be implemented.

Source: https://www.peerswap.dev/

Except Peerswap uses Liquid

True.
And also BCH enabled much bigger blocks, which hurts decentralization. Meanwhile BIP-110 aims to do the opposite by reducing block bloat caused by inscriptions.
So the comparison to BCH makes no sense.

correct

It can be activated early if at least 55% of blocks signal for it in one 2016 block period.

The mining pools that use Stratum V1 yes. And that's probably a majority of the hashrate. It's this centralized unfortunately.

I believe they’re 50 Megahash each 😂

50 MH/s ?? A CPU can do better than that. It's probably 50 GH/s.

Pollard's Kangaroo is an algorithm for solving the discrete logarithm problem (or in the Bitcoin case: solving the elliptic curve discrete logarithm problem (ECDLP)) where you already know that the secret scalar you are trying to find is within a certain range.

$Q=k\ast G$
You know $Q$ (the public key)
You know $G$ (the generator point of secp256k1)
You are trying to find $k$ (the private key)

The time complexity of Pollard's Kangaroo is $O(\sqrt{n})$
The time complexity of a naive brute force search is $O(n)$

btcpuzzle.info is a real world test of brute force limits

Pollard's Kangaroo is faster than brute force

And BIP110 is by definition a soft fork.

Soft forks tighten the rules, so they are backwards compatible with the old more lax rules.
Hard forks introduce new rules that are invalid under the old rules.
Important distinction.