[Nov 2024] Fees are low, think about Consolidating your small inputs @2.0 sat/vb

[stompix]

I've seen this before: suddenly, the size of the mempool drops quite a lot, but it's the fees "in the middle" that disappear. If it's not a bug on the site, all I can think of is that someone with many large transactions paying 10 sat/vbyte decided to double spend some of the inputs, thereby invalidating the older much larger transactions. But I have no idea if that's what happened.

Normally is a major purge on tx from a lot of nodes, the 300MB default hitting a huge chunk of tx waiting in the mempools, but as I see it now, rather than a major purge in the 10sat/b or a replacement I would put the bet on some major ordinal inscriptions being canceled by a double spending.


That means from block to 821156 to block 821187, 31 blocks would have found impossible to clear that, 63vMB have disappeared, if they would have been replaced they would have been confirmed but only 30 of them, leaving 33 plus the extra incoming tx still in the mempool and not erased.
This is more than even Binance is able to spit out, no way a mixer or another exchange would have left that much hanging in the mempool for so long.

[mikeywith]

Normally is a major purge on tx from a lot of nodes, the 300MB default hitting a huge chunk of tx waiting in the mempools, but as I see it now, rather than a major purge in the 10sat/b or a replacement I would put the bet on some major ordinal inscriptions being canceled by a double spending.

Or could it be the thing that we were discussing in the diff thread? a simplified explanation is below


Let's say there are currently 100,000 transactions on the mempool (ok there is no such thing as 'THE' mempool but just to keep things simple)

Take 4k transactions to be the average transactions per block.

Say 40k transactions are paying over > 200 sat / Vbyte , and then the remaining 60K are paying only 1 sat Vbyte.

You are a mining pool with a lot of money, you also know those 40k transactions will need 10 more blocks to clear, so you do this

create your own 40K transactions paying 100 stat / Vbyte, and wait ..

If people keep outbidding your 100 sat transaction, you are safe and making a good income, but then if say the mempool runs out of >200 sat transactions and only 8k left, so now you are only 2 blocks away from starting to lose money (because your 40k transactions are artificial and were meant to spam the blockchain only if other pools mine them you lose money).

within the 2 blocks left, you go and spend 1 input of each of your 40k transactions and pay 250 sat/ Vbyte, that way you are certain that your 40k transactions will be invalid and it won't cost you as much.

The average user doesn't know any of that, so if you spam the blockchain with 50 sat transactions, you are likely to keep new transactions coming at above 50 sats, if you run out, you purge your 50 sat transactions and lower the bar to 40 sat, of course, this remains just a "theory".

[philipma1957]

The fees are clearly getting out of hand. This is crazy, I have survived the first fee surge during the hard fork times and I can say that current fees can compare. I remember paying several grand for moving my hard fork gains. Well I believe we're going into that direction. Kinda stupid: Bitcoin price will moon but the fees will be too high to withdraw.  

Thanks monkey pic fans!  

do not blame ordinals.  this can be done with out ordinals.

and when we get to 3.125 size reward it will be 2x cost effective for the large pools to do it.

[hosseinimr93]

within the 2 blocks left, you go and spend 1 input of each of your 40k transactions and pay 250 sat/ Vbyte, that way you are certain that your 40k transactions will be invalid and it won't cost you as much.

Unless you (the mining pool) manage to find a block and include the replacement transaction, you will still lose a big amount of money.

According to BIP125, the fee paid for the replacement transaction must be greater than the total fee paid for all the transactions that are evicted from the mempool after the replacement.

Let's say I have made a transaction with 100 inputs and 100 outputs with the fee rate of 100 of sat/vbyte and now I want to replace that with a transaction with 1 input and 1 output.
Even if I use the fee rate of 250 sat/vbyte for the new transaction, almost all nodes will reject it.

[philipma1957]

within the 2 blocks left, you go and spend 1 input of each of your 40k transactions and pay 250 sat/ Vbyte, that way you are certain that your 40k transactions will be invalid and it won't cost you as much.

Unless you (the mining pool) manage to find a block and include the replacement transaction, you will still lose a big amount of money.

Take note that according to BIP125, the fee paid for the replacement transaction must be greater than the total fee paid for all the transactions that are evicted from the mempool after the replacement.

Let's say I have made a transaction with 100 input and 100 output with the fee rate 100 sat/vbyte and now I want to replace that with a transaction with 1 input and 1 output.
Even if I use the fee rate of 250 sat/vbyte for the new transaction, almost all nodes will reject it.

this strategies only work for pools with 20% or more of the hash rate

and they will become 2x cost effective when blocks go to 3.125 coins.

[mikeywith]

According to BIP125, the fee paid for the replacement transaction must be greater than the total fee paid for all the transactions that are evicted from the mempool after the replacement.

You are correct but bip 125 is not a consensus rule, i.e not following its rules doesn't make a block invalid, your node could be set up in a way that it doesn't even signal RBF and still allow transaction replaceability.

Here is a theory of why other pools would mine the second set of transactions even if the sum fee is lower on that second transaction, and just to keep things simple, we would just assume it's a 270 transaction with 100 inputs and a single output  

Total vSzie = 14810 Vbyte paying 100 sat per Vbyte would mean 3,998,700 sats , you could fit 270 of those in a 4 block i.e you make 399,870,000 sats

You are a pool with 10% the hashrate, i,e you find 1 block every 10 blocks, and you see that the second transaction spends 1 input of each transaction, a transaction size of 14810 paying 1350 sat / Vbyte or 19,993,500 , while the original transactions pay 100 sat, and you still have some transactions paying 200 sat/Vbyte that are enough only for 2 more blocks , why wouldn't you as a pool rush to include that transaction that is paying almost 7 times the other transactions?

bip 125 concept maximizes profitability on a global scale, i,e if enforced by everyone and all transactions are non-miners, if those 270 transactions were user transactions and no miners were playing games, and they were all on the same line, it would make sense, why replace a 4 BTC reward for a mere 0.2 BTC reward? reject the second transaction keep 4 BTC worth of reward floating until it's time comes, but we don't know if no miners are "playing those games".

Antpool and Foundry alone own 52% of the total hashrate, so they are most guaranteed to find every other block, in my above example if their "spam transactions" cost them 4BTC they are almost guaranteed to retrieve 2 BTC in the worst case scenario even without having to play those games (mempool runs out of higher paying transactions and those spam transactions start to be picked by others), obviously, those vague number of mine would not work -- but if they carefully evaluate the numbers they could manage to artificially raise the fee rate by injecting BTC into the system in order to extract more of it.

This remains but a theory that some people believe in, it could be valid, we don't know if there are enough real users paying anything above 10 sats, and half of those transactions could be there for spam, where x, y, z pools spend 0.2 BTC on each block to raise the total fees to 1 BTC, and if they find at least 1/4 of those blocks -- they make more money.

[philipma1957]

According to BIP125, the fee paid for the replacement transaction must be greater than the total fee paid for all the transactions that are evicted from the mempool after the replacement.

You are correct but bip 125 is not a consensus rule, i.e not following its rules doesn't make a block invalid, your node could be set up in a way that it doesn't even signal RBF and still allow transaction replaceability.

Here is a theory of why other pools would mine the second set of transactions even if the sum fee is lower on that second transaction, and just to keep things simple, we would just assume it's a 270 transaction with 100 inputs and a single output  

Total vSzie = 14810 Vbyte paying 100 sat per Vbyte would mean 3,998,700 sats , you could fit 270 of those in a single block i.e you make 399,870,000 sats

You are a pool with 10% the hashrate, i,e you find 1 block every 10 blocks, and you see that the second transaction spends 1 input of each transaction, a transaction size of 14810 paying 1350 sat / Vbyte or 19,993,500 , while the original transactions pay 100 sat, and you still have some transactions paying 200 sat/Vbyte that are enough only for 2 more blocks , why wouldn't you as a pool rush to include that transaction that is paying almost 7 times the other transactions?

bip 125 concept maximizes profitability on a global scale, i,e if enforced by everyone and all transactions are non-miners, if those 270 transactions were user transactions and no miners were playing games, and they were all on the same line, it would make sense, why replace a 4 BTC reward for a mere 0.2 BTC reward? reject the second transaction keep 4 BTC worth of reward floating until it's time comes, but we don't know if no miners are "playing those games".

Antpool and Foundry alone own 52% of the total hashrate, so they are most guaranteed to find every other block, in my above example if their "spam transactions" cost them 4BTC they are almost guaranteed to retrieve 2 BTC in the worst case scenario even without having to play those games (mempool runs out of higher paying transactions and those spam transactions start to be picked by others), obviously, those vague number of mine would not work -- but if they carefully evaluate the numbers they could manage to artificially raise the fee rate by injecting BTC into the system in order to extract more of it.

This remains but a theory that some people believe in, it could be valid, we don't know if there are enough real users paying anything above 10 sats, and half of those transactions could be there for spam, where x, y, z pools spend 0.2 BTC on each block to raise the total fees to 1 BTC, and if they find at least 1/4 of those blocks -- they make more money.

much better than a 51% attack

and it explains weird difficulty  shifts.

They don't have to use ordinals to do this.  And it becomes easier when blocks drop to 3.125 in April 2024

[mikeywith]

and it explains weird difficulty  shifts.

They don't have to use ordinals to do this.  And it becomes easier when blocks drop to 3.125 in April 2024

It becomes easier the more concentrated the hash power becomes, the 4 top pools own roughly 80% of the hashrate.

1- Antpool = Bitmain
2- Foundry = Bitman's largest partner?
3- Viabtc = Funded by Bitmain (technically owned by them)
4- F2pool=  Owned by QWang Chun a (censorship cunt)

I don't see why wouldn't these 4 guys sit together in a closed room and think of all the different ways they can increase their profit collectively, they are not stupid to do something that would risk damaging Bitcoin so bad like a 51% attack or forking the blockchain, so ya, playing with difficulty and fees are two things they can do to increase their profit without risking much.

It could be that Ordinals were the catalyst for this "attack", they realized that people are willing to pay >50 or > 100 sats to outbid Ordinals, so why not make that the new standard, why settle with 1-2 sat /Vbyte when you can fake the new normal to be 10-20 sats? or even better 40-50 sats?

LoyceV; sorry if this has gone a little off-topic, but ya, it seems like consolidating for cheap isn't going to be an option for a while.

[hosseinimr93]

You are a pool with 10% the hashrate, i,e you find 1 block every 10 blocks, and you see that the second transaction spends 1 input of each transaction, a transaction size of 14810 paying 1350 sat / Vbyte or 19,993,500 , while the original transactions pay 100 sat, and you still have some transactions paying 200 sat/Vbyte that are enough only for 2 more blocks , why wouldn't you as a pool rush to include that transaction that is paying almost 7 times the other transactions?

You are right. There is no doubt it's more profitable to include the transaction paying 1350 sat/vbyte.
But the question is will nodes relay that transaction and will other mining pools receive that? I doubt it.

There's a big probability that a mining pool includes the original transactions and I think that's a risky game.


I am not saying it's not possible to play such a game.
It's even possible that the bad mining pool play the game more dishonestly, but as I said it can be so risky.

Let's say I am running a mining pool and I benefit from high fees. My mining pool owns 25% of the total hash power and we can mine 1 block in every 4 blocks on average.
I broadcast 1000 transactions with the fee rate of 100 sat/vbyte. Each transaction includes 100 inputs and they have created some long chains of unconfirmed parents.
I have made some  transactions with 1 input and 1 output that can invalidate all those 1000 transactions. If my mining pool manage to find a block and those 1000 transactions are less than x vMB from the tip of the mempool, I will include the 1 input-1 output transactions in the blockchain and invalidate all those 1000 transactions.



A correction to your calculations:

Total vSzie = 14810 Vbyte paying 100 sat per Vbyte would mean 3,998,700 sats , you could fit 270 of those in a single block i.e you make 399,870,000 sats

67 of those transactions.
Each block can contain up 1 vMB of transactions.

[mikeywith]

But the question is will nodes relay that transaction and will other mining pools receive that? I doubt it.

Mining pools have a dozen nodes around the globe, and more specifically , they are perfectly connected to one another, it's essential to avoid orphan/stale blocks, I don't remember hearing about any orphan race for ages now.

There's a big probability that a mining pool includes the original transactions and I think that's a risky game.

It all depends on how you operate it, if the four largest pools do it together, even without invading their original transactions, they would only lose 20% of whatever the invest to spam the blockchain, so if they raise fees by anything equivalent or more than the 20% they lose -- they still make more profit.

Let's say I am running a mining pool and I benefit from high fees. My mining pool owns 25% of the total hash power and we can mine 1 block in every 4 blocks on average.
I broadcast 1000 transactions with the fee rate of 100 sat/vbyte. Each transaction includes 100 inputs and they have created some long chains of unconfirmed parents.
I have made some  transactions with 1 input and 1 output that can invalidate all those 1000 transactions. If my mining pool manage to find a block and those 1000 transactions are less than x vMB from the tip of the mempool, I will include the 1 input-1 output transactions in the blockchain and invalidate all those 1000 transactions.

That would be easier than getting other pools to mine your second transaction but it would make things a lot more obvious to anyone watching the blockchain.

I am not someone who deeply believes in this theory, to me, this is probably not something they are doing extensively just yet, but I am almost sure that it's in the pipelines, they have been probably testing this for a while and it seems like it "could" be working fine for them, looking at yesterday's transaction type distribution only about 10.5% were Ordinals, so the "normal" transactions made near 90% of them and fees were pretty high.

A correction to your calculations:

Noted and thanks, ya was doing the calc in weight units at first.

[stompix]

~

Or could it be the thing that we were discussing in the diff thread? a simplified explanation is below
~
If people keep outbidding your 100 sat transaction, you are safe and making a good income, but then if say the mempool runs out of >200 sat transactions and only 8k left, so now you are only 2 blocks away from starting to lose money (because your 40k transactions are artificial and were meant to spam the blockchain only if other pools mine them you lose money).
within the 2 blocks left, you go and spend 1 input of each of your 40k transactions and pay 250 sat/ Vbyte, that way you are certain that your 40k transactions will be invalid and it won't cost you as much.

Yeah mikey I know of it but there is a flaw with that, and the thing is that the chunk that went missing was in the 10sat/b area, something that hasn't been touched in a month at least, there was in no danger of being included in a block, and disappeared exactly when fees hit another record. So, why would a pool that is using that tactic to make tx fees surge withdraw something that at this point was in way in danger of  being confirmed and make them lose money and moreover, it didn't had any sort of effect in the last week?

It would have made perfect sense if the mempool was decreasing, that I can agree with but we're looking at the exact opposite!

The buildup in the 10-12 sat byte area from 60vMB to 130vMB happened pretty organic over two weeks, the lowest point they were buried was at 34vMB from the top, at the moment of the disappearance they were under at least 130vMB. I understand the reason behind them faking this wall of 10-12sat/vb, what's the point of erasing it right now and in such a fashion that for sure has made people curious and maybe even expose them? If they are trying to manipulate the fees and they are smart enough, shouldn't they be also be smart enough to cancel those tx over a few days and not in one single go? What was this rush for?

The sheer size of it makes me think is large ordinals inscriptions done by factory, they realized they will never have a chance like this and canceled their projects altogether.

[Gormicsta]

But the question is will nodes relay that transaction and will other mining pools receive that? I doubt it.

Mining pools have a dozen nodes around the globe, and more specifically , they are perfectly connected to one another, it's essential to avoid orphan/stale blocks, I don't remember hearing about any orphan race for ages now.

It is true that mining pools contain nodes all around the world, and that these nodes are interconnected to prevent blocks from being stale or orphaned. This is an essential component of their business since the miners would lose money if there were any orphaned or stale blocks. Orphan races are extinct thanks to the network's resilience and high degree of communication between nodes.

[mikeywith]

Yeah mikey I know of it but there is a flaw with that, and the thing is that the chunk that went missing was in the 10sat/b area, something that hasn't been touched in a month at least, there was in no danger of being included in a block, and disappeared exactly when fees hit another record. So, why would a pool that is using that tactic to make tx fees surge withdraw something that at this point was in way in danger of  being confirmed and make them lose money and moreover, it didn't had any sort of effect in the last week?

Maybe they are preparing a new wall at a higher tier?

The sheer size of it makes me think is large ordinals inscriptions done by factory, they realized they will never have a chance like this and canceled their projects altogether.

This probably makes more sense in this exact incident, it would fun to investigate this if someone stored those transactions and then traced their inputs in new transactions to see, for example, this transaction https://mempool.space/tx/acbfae1b01c1188434b69bb40765d7e00d3b9a36d6de9f05dd1bfce0b900d3e7 came right around that drop in mempool size, with over 650 inputs it could have invalidated/removed up to 650 transactions.

It is true that mining pools contain nodes all around the world, and that these nodes are interconnected to prevent blocks from being stale or orphaned.

It's thought to be that large pools are connected even outside of the BTC protocol with faster "methods" when a pool first finds a block, they want to let the other pools know as soon as possible so that other pools would start working on the next one to avoid orphaning (more like staling but whatever, seems like people prefer to use the term orphan instead) each other's blocks, it's less important to propagate the block to non-mining nodes.

[Don Pedro Dinero]

Hi folks, what a crazy thing, you can see there has been a huge dump of transactions this morning, which doesn't usually happen on a Sunday and now if you want a confirmation in the next block we are talking about over 600 sat/vB. To be honest some of the conversation in the previous posts, dense technical content I don't understand, but I wonder how far this is going to go.

[LoyceV]

over 600 sat/vB

Johoe is going to need a new color scheme if this keeps going!
This reminds me of the end of 2017, where the highest I've paid was around €25 for a transaction with one input. Segwit now costs around the same, Legacy inputs are currently above that.

I wonder how far this is going to go.

One would expect small transactions to disappear, but the Ordinal spammers seem to have deep pockets, which can only mean they're still making new victims. It doesn't make much sense.

[stompix]

over 600 sat/vB

Johoe is going to need a new color scheme if this keeps going!

Maybe this:

Seems appropriate!

The good thing in this is that the clog happened with a 50 minute block difference so it's improbable we will see that frequently but yeah, a next block tx is more than WU charges me for sending $1000 to 3 random countries Africa.

Maybe they are preparing a new wall at a higher tier?

Well, I don't see any evidence of an actual wall being built on top of that as the there is no large increase in one chunk alone but that could of course mean nothing. They cold just spread 3vMB over 20-25 and another 3vMB over 40-45, if they are smart enough to do the coding for those transactions and beforehand linking them the way we talked then hiding them in small batches would be the easy part.

Now, if they would actually do it, yeah, we would pretty much be F word!
No other option than to migrate to the sidechains, as thousands are doing right now:
https://bitinfocharts.com/comparison/transactions-doge-dogecoin.html#3m
https://bitinfocharts.com/comparison/litecoin-transactions.html#3m

I find it both amusing and ironic being forced by jpgs to use a meme coin!

[mikeywith]

but the Ordinal spammers seem to have deep pockets, which can only mean they're still making new victims. It doesn't make much sense.

Let's run some numbers.

17-12-2023

In the following:
Ordinals refer to both Brc-20 and non-Brc20
Normal refers to "normal/non-Ordiinals" bitcoin transactions.



Transaction count total (550k):

Ordinals = 335,000 transactions
Normal  = 220,000 transactions

Transaction % (100%):

Ordinals = 60.36%
Normal  =39.63%

Fee count (504 BTC / $21,168,000 ):

Ordinals = 188 BTC / $7,896,000
Normal = 316 BTC / $13,272,000

Fee % (100%)

Ordinals = 37%
Normal = 63%


The average Ordinal transaction paid 0.00056 BTC = 23.52$
The average normal transaction paid 0.0014 BTC = 58.8$


So normal transactions paid 150% more than Ordinals.

It seems like everyone has deep pockets.

if they are smart enough to do the coding for those transactions and beforehand linking them the way we talked then hiding them in small batches would be the easy part.

With more than 20M profit a day at stake compared to the old 1M or less, they can surely afford the proper code to manipulate these transactions, let's pretend its Foundry and Antpool alone doing this, this brings them 10M a day instead of the average 0.5M they used to make on fees, Antpool keeps all fees on their PPLNS (which they call 0% fee ) and own a lot of their own mining gears, and foundry is the same, so it's safe to assume that they would gain at least half of whatever they manage to make on fees, make that 5M a day, if they do this for 10 days every month that's 50M a month, that's a lot of money, they won't even need to play with 50M to spam the blockchain, one-tenth of that is enough.

I would also guess that many of these would be naive users who overpay their fees, and wallets that overestimate fees also add to this mess.

[philipma1957]

but the Ordinal spammers seem to have deep pockets, which can only mean they're still making new victims. It doesn't make much sense.

Let's run some numbers.

17-12-2023

In the following:
Ordinals refer to both Brc-20 and non-Brc20
Normal refers to "normal/non-Ordiinals" bitcoin transactions.



Transaction count total (550k):

Ordinals = 335,000 transactions
Normal  = 220,000 transactions

Transaction % (100%):

Ordinals = 60.36%
Normal  =39.63%

Fee count (504 BTC / $21,168,000 ):

Ordinals = 188 BTC / $7,896,000
Normal = 316 BTC / $13,272,000

Fee % (100%)

Ordinals = 37%
Normal = 63%


The average Ordinal transaction paid 0.00056 BTC = 23.52$
The average normal transaction paid 0.0014 BTC = 58.8$


So normal transactions paid 150% more than Ordinals.

It seems like everyone has deep pockets.

if they are smart enough to do the coding for those transactions and beforehand linking them the way we talked then hiding them in small batches would be the easy part.

With more than 20M profit a day at stake compared to the old 1M or less, they can surely afford the proper code to manipulate these transactions, let's pretend its Foundry and Antpool alone doing this, this brings them 10M a day instead of the average 0.5M they used to make on fees, Antpool keeps all fees on their PPLNS (which they call 0% fee ) and own a lot of their own mining gears, and foundry is the same, so it's safe to assume that they would gain at least half of whatever they manage to make on fees, make that 5M a day, if they do this for 10 days every month that's 50M a month, that's a lot of money, they won't even need to play with 50M to spam the blockchain, one-tenth of that is enough.

I would also guess that many of these would be naive users who overpay their fees, and wallets that overestimate fees also add to this mess.

fees stay high and frankly I see them staying high for months maybe till July only a price crash will drop the fees.

last 10 hours and the 80 sat level is unchanged.  and we made a lot of blocks in those 10 hours

821,892 12-19 19:47:10
821,908 12-19 09:18:49    so in 10 hours and 30 minutes we should do 63 blocks

we did  84 blocks . So 84/63 = 1.3333 vs 1  so  miners whaled like mad (well foundry did) and fees at the 80 sat level were not changed.

[mikeywith]

Fees now have dropped to 80~ sats/Vbyte, if nothing much changes, in a few more hours you may get away with 40 sats transactions, there is a large wall at around and above 40 sat (almost a whole day's worth of transactions) so I think 40~ sats would be a good level to consolidate if you are rushing and don't want to wait, I think we would eventually in a few days get to 10-20 sats, but probably never again see 1-2 sats unless something major happens.

[apogio]

Fees now have dropped to 80~ sats/Vbyte, if nothing much changes, in a few more hours you may get away with 40 sats transactions, there is a large wall at around and above 40 sat (almost a whole day's worth of transactions) so I think 40~ sats would be a good level to consolidate if you are rushing and don't want to wait, I think we would eventually in a few days get to 10-20 sats, but probably never again see 1-2 sats unless something major happens.

Thanks mikey.

Personally I took the chance to consolidate some inputs and also make some pending payments at 63.5 sats/vB.