Nostr Web Client

"Subsat summer" shows that the filters work. Here's a question for you: if the 1sat filter doesn't work, why is there a huge wall at the 1 sat mark? (This chart is for blocks 910047-911047 and is from x.com/mononautical) "No filters" would produce something like a normal distribution.

There is clearly something "special" about 1 sat per byte, as demonstrated conclusively by simple blockchain analysis. I think it is widely used in preference to the smaller values because the smaller values are *harder* to use, as you have to bypass the filters to use them.

Reply to this note

Please Login to reply.

Discussion

ihsotas 3mo ago

I think the filter controversy is talking past each other mostly. Utxo is not the preferred place to put data which I believe is why op return was created in the first place. People who want to put data into the chain currently use the utxo set because it is cheaper and can do so with greater size than op return…this is not good. We can’t make them stop but we can offer a less bad option. Perhaps people would be willing to pay a bit more and put their trash in the trash can instead of just tossing their garbage into a sensitive area. Time will tell.

I think the maxim “don’t let perfect be the enemy of good” applies here. I also think the conspiracy theories and calling core devs shitcoiners is perhaps the worst outcome of all this.

Super Testnet 3mo ago

Most of the people who put spam in utxos use a protocol called Stamps. The Stamps protocol was created out of an explicit desire to put the data in the utxo set. See, for example, this quote from the cofounder: "The point of Stamps is to offer a method of encoding on-chain art that is highly resilient to tampering and goes a step further than Ordinals by being unprunable." source: https://thedefiant.io/news/nfts-and-web3/bitcoin-stamps-seek-to-improve-on-ordinals

Enlarging op_return to accommodate these people and make their data prunable seems pointless, as they have explicitly said they don't want their data to *be* prunable. If the point is to accommodate other, "less popular" protocols or one-off spammers, then the cure seems far worse than the disease: apart from Stamps, hardly anything uses output stuffing, so even if those other unpopular protocols *do* switch to op_return (which I doubt many will do), you'll get a very small benefit in that one respect at the expense of making low-effort spamming far easier for the masses.

Jack D 3mo ago

I don't buy the normal distribution guess. Fees are not bid randomly, they are bid intelligently. Transactions are not broadcasted with continuous demand. And when a block is found, a range is removed from the mempool all at once.

Jack D 3mo ago

Also, you are showing a chart where a whole ton of txns which should have been filtered would have not been mined if non-mining filters had a significant effect. Most people had minfees set.

Super Testnet 3mo ago

I do not think those are good grounds for doubting my hypothesis. I agree that fee bids are not posted randomly, in fact I think the stats clearly show that such bids favor 1sat per byte and higher.

If your hypothesis is that this effect is due to removing a range of bids all at once, then I think it is non-explanatory: if the filters have no effect, then fee bids ought to show up with similar frequently in the nearest values on either side of 1 sat per byte, because the range removed by each block is just as likely to leave transactions unmined on one side of that value as on the other.

But in fact fee bids do *not* show up with similar frequency in the nearest values on either side of 1 sat per byte. They favor 1 sat and higher. A better explanation is that something makes it harder for users to *set* feerates lower than 1 sat per byte; and in fact, most wallets don't *let* you do that, because they were designed with the filters in mind.

Jack D 3mo ago

You have to doubt a hypothesis if you cannot rule out the null hypothesis. I've presented the null hypothesis in reasons for why the distribution might not be normal. I was not stating a hypothesis. I presented reasons for doubting yours. I don't know if the data can can be so easily gathered from on-chain data.

Jack D 3mo ago

Would you agree that the popular fee relay filter of at least 1 sat running on most nodes did not prevent those sub 1 transactions from being mined?

Super Testnet 3mo ago

I agree that the filter did not prevent the ones that got mined from being mined

Jack D 3mo ago

I think that is the crux of most of disagreement. It's on the definition of what "working" means.

Super Testnet 3mo ago

If the other side thinks filters only "work" if they stop *all* spam then their definition is absurd. No filter in any context I'm aware of stops 100% of pollutants. Mempool filters have a clear statistically significant limiting effect. Removing them means they no longer think that effect is worth their cost. And that's where I disagree with them.

Jack D 3mo ago

The chart you posted looks like a big failure for the minfee filter. 90% of nodes running the filter, and that many got through? It looks like they did very little if anything. Claiming "statistically significant" doesn't make something so. I've yet to see any rigorous statistical analysis from either side, and I don't think the evidence can be gathered from chaindata alone.

nicnym 3mo ago

The reason most are at 1sat is the default app like Nunchuk or sparrow doesn’t allow less than 1 sats.

This is “the filter” it is a social policy that works reduce undesired behavior.