Are we stress testing again?

[medUSA]

Yeah, someone told me that too today. I think thats unbelievably stupid. How can they filter out the transactions? It does not make sense at all.

Blockchain.info is connected to many nodes, mostly 500, sometimes over 1000. I believe they are right in filtering out spammy transactions and refuse to relay them to other nodes. Would it make a difference to the network if they filter out spammy transactions? No, not really. There are many other nodes that will relay spam.

You are correct in saying they give a false picture of the mempool, which can lead to user's transactions getting stuck in limbo. I believe they could change the layout of the page and add info about filtered transactions, but filtering them IMO is a right move.

[1714953159]

1714953159

[1714953159]

1714953159

[1714953159]

1714953159

[tspacepilot]

Yeah, someone told me that too today. I think thats unbelievably stupid. How can they filter out the transactions? It does not make sense at all.

Blockchain.info is connected to many nodes, mostly 500, sometimes over 1000. I believe they are right in filtering out spammy transactions and refuse to relay them to other nodes. Would it make a difference to the network if they filter out spammy transactions? No, not really. There are many other nodes that will relay spam.

You are correct in saying they give a false picture of the mempool, which can lead to user's transactions getting stuck in limbo. I believe they could change the layout of the page and add info about filtered transactions, but filtering them IMO is a right move.

Indeed, for better or for worse, bc.i is the best-known block explorer and while I don't begrudge them for doing what they have to do to keep their service responsive, it's better to be more transparent rather than less.  Since so many people are going to them for information about the blockchain (ahem, it's kinda in their name!) they really ought to be upfront about what information they're actually providing and what they're ignoring/not-providing.

[amaclin]

they really ought to be

they ought nothing to you and me. point.

[Quickseller]

I dont think anything is going on at the moment. Unconfirmed transactions are below 2000 and thats normal. https://blockchain.info/de/unconfirmed-transactions

do not trust bc.i
they do not show you all network transactions even the number of them
now they have some anti-spam filters which ignore incoming chained unconfirmed transactions

Yeah, someone told me that too today. I think thats unbelievably stupid. How can they filter out the transactions? It does not make sense at all.

At least you can still see when the network gets spammed. When the unconfirmet transactions are higher than 2000 then its most probably the case. I mean real transactions are stuck and still get counted.

But a stupid move of them for sure.

They setup their nodes so that they will reject certain transactions that do not meet certain criteria. For example, the last time someone was spamming the network, they were rejecting transactions with small inputs that had a fee that was under a certain amount.

Other transactions they will reject would be double spends (which is similar to what the majority of nodes will do).

[Cryddit]

I'm still of the opinion that someone is doing it because it's profitable.

And if it's profitable, they won't stop.  Ever.

[bitcoinmasterlord]

I'm still of the opinion that someone is doing it because it's profitable.

And if it's profitable, they won't stop.  Ever.

I agree. The thread with the golden ratio attack was killed but i think the arguments against aren't good enough. The attacking party only needs to have a way to create miners cheaper than the competition and the arguments against vanish. And for a chinese miner consortium it should be easy to create the cheapest way.

[RoadTrain]

I'm still of the opinion that someone is doing it because it's profitable.

And if it's profitable, they won't stop.  Ever.

I agree. The thread with the golden ratio attack was killed but i think the arguments against aren't good enough. The attacking party only needs to have a way to create miners cheaper than the competition and the arguments against vanish. And for a chinese miner consortium it should be easy to create the cheapest way.

Still, this consortium would be subsidizing other miners. This might be a drop in the bucket, but still it would shift balance towards other miners.

[oblivi]

I'm still of the opinion that someone is doing it because it's profitable.

And if it's profitable, they won't stop.  Ever.

I fail to understand how attacking the network with spam is profitable in any way. Anyone cares to elaborate on this hypothesis? The only way I see this being profitable is if you were spamming a website with every transaction as some sort of massive advertisement but it's not the case, and everyone would hate you anyway.

[var53]

I'm still of the opinion that someone is doing it because it's profitable.

And if it's profitable, they won't stop.  Ever.

I fail to understand how attacking the network with spam is profitable in any way. Anyone cares to elaborate on this hypothesis? The only way I see this being profitable is if you were spamming a website with every transaction as some sort of massive advertisement but it's not the case, and everyone would hate you anyway.

What if you had a giant stash of Litecoins you bought dirt cheap right at the bottom and you were also running a Chjinese Litecoin ponzi scheme? You could pump the Litecoin price using the ponzi scheme, then spend some Bitcoins to spam the Bitcoin network and slow everyone's transactions down to a snails pace.

People would be likely to convert their Bitcoins to Litecoins if they wanted to quickly move their money between exchanges. That would help pump the Litecoin price even more and the Litecoin ponzi scheme operators could dump their Litecoins for a bigger profit.

[tspacepilot]

I'm still of the opinion that someone is doing it because it's profitable.

And if it's profitable, they won't stop.  Ever.

I fail to understand how attacking the network with spam is profitable in any way. Anyone cares to elaborate on this hypothesis? The only way I see this being profitable is if you were spamming a website with every transaction as some sort of massive advertisement but it's not the case, and everyone would hate you anyway.

At least in the original "stress-test", part of the context was trying to "prove" that the blocksize limit needed to be adjusted sooner rather than later in order to accomodate more transactions.  The idea here is not exactly that it's profitable in the short term to spam, but that the spam proves a point which you wanted to prove.  I don't really think this succeeded, btw, and I'm not 100% sure that was the actual motivation, nevertheless, it's one hypothesis.  Here's another one though, again, it's not for profit, but you can imagine that there are trolls out there who want an idea to fail.  Maybe you're a banker and you're threatened by bitcoin the same way hotels are threatened by airbnb.  You want bitcoin to fail so you spend money to make it hard to use.  Again, this may not be the case, but it's another plausible explanation.

I agree with you that the idea that someone is somehow profitting directly by spending a lot of coins to slow down Bitcoin seems a bit farfetched to me.

[medUSA]

I fail to understand how attacking the network with spam is profitable in any way. Anyone cares to elaborate on this hypothesis?

During the attack, the spam created a transaction fee floor. Legit transaction with fees below the floor will be severely delayed. This prompted users to use a higher than standard fee if they want faster confirmation. After the attack(s), I have heard that some dice sites and exchanges have increase the standard fee of withdraws. If the new transaction fee stays semi-permanent, the attack IS profitable for miners (no evidence they are behind the attacks).

[Cryddit]

Still, this consortium would be subsidizing other miners. This might be a drop in the bucket, but still it would shift balance towards other miners.

If a consortium has more than 38% of the mining power, subsidizing other miners and getting higher fees is more profitable than not doing so.   And whenever there is legitimate demand for more than half the space in the blocks, you can get higher fees by spamming the block chain.

[JorgeStolfi]

Another "stress test" just started.  The backlog is already 60 MB of unconfirmed transactions, and the input traffic is ~ 3 MB every 10 minutes, or ~5 kB/s .  So far, the tester is paying 0.2 mBTC/kB, like the last one.  Will it keep at that, or increase the fees later?

[tspacepilot]

Another "stress test" just started.  The backlog is already 60 MB of unconfirmed transactions, and the input traffic is ~ 3 MB every 10 minutes, or ~5 kB/s .  So far, the tester is paying 0.2 mBTC/kB, like the last one.  Will it keep at that, or increase the fees later?

It seems like it's going on weekly at this point.  I really wonder how long the "testers" can afford this.

[RealMalatesta]

Another "stress test" just started.  The backlog is already 60 MB of unconfirmed transactions, and the input traffic is ~ 3 MB every 10 minutes, or ~5 kB/s .  So far, the tester is paying 0.2 mBTC/kB, like the last one.  Will it keep at that, or increase the fees later?

It seems like it's going on weekly at this point.  I really wonder how long the "testers" can afford this.

Let's assue the attacker is a bank which tries to establish its own coin / blockchain and first wants to test and eventually permanently attack Bitcoin... they would be able to afford this for years. So I guess the "cheap days" are over...

[talkbitcoin]

it doesn't feel like any more stress testing is happening . maybe the spammers ran out of money and dind't achieve their goal

[Cryddit]

it doesn't feel like any more stress testing is happening . maybe the spammers ran out of money and dind't achieve their goal

Betcha they did achieve their goal.  Check what the average fee per kilobyte being paid is now vs. when the "stress tests" started. 

[Amitabh S]

That was not a stress test but a spam attack, and I don't thinking increasing block sizes will prevent it.. Well the debate goes on

[OROBTC]

...

For everything but one bitmixer.io payment (a signature payment, one of which WAS delayed for over 40 hours), I have had no troubles at all with any transactions since I started changing my miner's fee payment to BTC0.0003 (roughly eight cents).

No other problems.  EIGHT CENTS is perfectly OK with me for a Bitcoin transaction.

[JorgeStolfi]

For everything but one bitmixer.io payment (a signature payment, one of which WAS delayed for over 40 hours), I have had no troubles at all with any transactions since I started changing my miner's fee payment to BTC0.0003 (roughly eight cents).

No other problems.  EIGHT CENTS is perfectly OK with me for a Bitcoin transaction.

Indeed most (all?) of the ~80'000 "test" transactions in the queue have 0.2 mBTC/kB fee, so if you pay more than that you should get confirmed in the next block or two -- as used to be the case before the stress test, even with minimum fee.

However, there are several hundred transactions with fee 0.4 mBTC/kB, and another few hundred paying 0.5.   It is not clear whether they were issued by the tester, too, or by normal clients who raised their fees to bypass the backlog.

That was not a stress test but a spam attack

The fact that most transactions pay the same low fee seems to make it a "brutal test" rather than an outright attack.  That is, the tester is not really trying to block the network (he woudl have to kep raising the fees to do that), but does not give a damn about the ~12'000 transactions that are caught in the backlog with fees below 0.2 mBTC/kB, whose issuers probably have no idea of what is going on.

I don't think increasing block sizes will prevent it.

There is no way to prevent a spam attack by someone who has a large enough budget.  All one can do is make such attacks more expensive, hence less likely.

Increasing the block size to 8 MB would make a similar attack 15-20 times more expensive.  It would also clear the backlog 15-20 times faster.