pull down to refresh
198 sats \ 4 replies \ @rijndael 15 Oct \ parent \ on: MEMPOOL SPACE SAGA (My Take) bitcoin
I think you are confused about fee estimation.
The core fee estimator works one way (looking at recently confirmed blocks). The mempool-space estimator works differently (looks at what is in its mempool). There are other fee estimation algorithms out there too.
The core algorithm tends to not work well when there are rapid changes in the mempool, because it looks at what has already been confirmed. For steady activity, thats fine but hasnt worked as well as the mempool-space approach for the last two years.
A good fee estimator will help users pick the right fee to get confirmed in the time they prefer. In other words, if you pay for next block, you usually get it and are not grossly overpaying. The core fee estimator isnt better just because its in core.
Also, fee estimators aren’t consensus or policy. They’re a tool.
This:
Doesnt make sense
I agree with 90%+ of what you've written. I however feel that "spam filters" are actually better for fee estimation. And since you feel the need to use mempool's node, you're not capturing other node's snapshot of their memory pools.
Core isn't the only implementation that uses fee estimation from the recently mined blocks. Knots also does the same thing.
Using recently mined blocks, imho, sorta includes the perspectives of other nodes'(spam filterers included) snapshot, and how long the transactions took before they got to the mined blocks, from their memory pools.
You are however 100% correct that this isn't consensus yet. My only point is, using a narrow POV for fee estimation that the whole network uses, without using what the whole network perceives, can rub people differently, especially when there's scenarios for conflicts of interest cases, as is the current case with the backlash received by mempool-space. People might think they're intentionally allowing their node to relay anything just to skew things for everyone else. And then there's issues about them sophisticated UI trickery to attract people to these blobs of data(I am not even going to fixate on that).
Mempool space is not a net-negative, this is just constructive criticism.
reply
Whoa... I think I finally understood what you're getting at.
So, with the mempool.space approach, it is effectively a universal inclusion rate, and as such, it is intrinsically a high-water mark that is pressured up by all fee market activity.
However, it sounds like you're also saying that since some block producers, like Ocean+Datum or Braiins+Stratum v2, can purposely assert alternate block inclusion preferences/"filters," these choices put downward pressure on "aligned" fee markets.
So what you're seeing is that there are effectively two parallel fee markets at play? One for all-comers, more or less as described by mempool.space; and another for bitcoin spends with some time to spare, where backward-looking timechain-sourced fee rates will effectively include a discount rate over the last N blocks proportional to how many "aligned" miners versus "profit-maximizing" miners are servicing the mempool.
Is that what you're getting at? If that's what you're saying, I think I you kind of just blew my mind, wow!
reply
That's exactly what I am saying. Plus a shit ton of people use mempool-space's fee estimation as what the network's fee estimation is irl(example: timechaincalendar.com, timechainstats.com)
I love these projects to the core, cos they're very beneficial in general. It's just that these nuances can an would be exploited by malicious people.
reply
Knots is a fork of more. So it uses the same fee estimation as core.
What you ideally want to make a good fee estimates is to see whats in the template at as many mining pools as you can. You can’t, but the network is well-connected and the overwhelming majority of hashrate generally sorts by fee rate (plus some tx acceleration) so having a well-connected node and sorting by fee rate does a really good job.
Other “noderunner preferences” doesnt change things unless they are getting transactions to miners that you don’t see.
reply