Economically Unspendable Outputs: A Problem On The Radar

[misterbigg]

It's not likely that this "somebody" would ever reach the kinds of volumes that SD has, so it wouldn't be much of a problem if this "somebody" would do so if their operations were minimal.

Yeah the combination of SatoshiDICE's reluctance to change their losing bet confirmation along with Blockchain.info giving them tons of free traffic through direct links from wallet page is a one-two punch to the face of not only everyone running a full node, but any users who have to pay fees (because the fees are higher).

This is hitting legitimate businesses like bitpay or even mtgox directly in the wallet through higher fees.

I'm not saying that gambling is not a valid use of the block chain, nor am I saying that we should block SatoshiDICE specifically from the network. But we desperately need to make it impossible or impractical to send amounts that are below the cost of transmission and storage.

[1713958897]

1713958897

[1713958897]

1713958897

[1713958897]

1713958897

[Amitabh S]

How about SD send the 1 satoshi (or 1000 satoshis) to a common "donation" address instead of back to the sender?

Senders can verify that the donation indeed contains funds from at least one of their original outputs. Also, SD can bunch a lot of such payments into one tx.

Its still a "band aid" until the "problem" (as many say) is fixed.

[Luke-Jr]

i would rather have SDICE exploit this flaw now rather than somebody else with really bad intentions some time in the future.

SD has clearly established, through their refusal to stop attacking Bitcoin, that their intentions are bad - or at the very least indistinguishable from what someone with bad intentions would do.

misterbigg, thank you for raising awareness of the problem

For the thousandth time: There is NO "problem". Just a bug in the software.

No, there is no software bug. The bug is in the human component of the protocol.
Satoshi was intelligent enough to recognize that no anti-spam software can ever be perfect or address future spamming techniques, so he designed the Bitcoin protocol to make use of humans to manage the job.
The mistake was that he assumed that miners would always care about the long-term welfare of Bitcoin, and neglected to give them any real incentive to do so. As a result, the bigger miners - who apparently only care about their short-term profits - neglect and/or refuse to do their part in filtering out this garbage.
The only bug here is in these humans - not the software.

[misterbigg]

How about SD send the 1 satoshi (or 1000 satoshis) to a common "donation" address instead of back to the sender?

The reason SatoshiDICE sends these betting loss confirmation is because Blockchain.info plays a sound when your wallet receives a payment. The cycle goes like this:

1) user logs into their Blockchain.info wallet
2) user clicks a SatoshiDICE betting link
3) SD sends them a loss confirmation
4) Blockchain.info AJAX web page plays a sound
5) go to step #2

If the 1 satoshi goes to a donation address then the player doesn't hear the sound and there's no Pavlovian conditioned response taking place.

Also, SD can bunch a lot of such payments into one tx.

Not really because all the dust comes from different places so will require many txin, typically 1 for each loss notification, wherein the fees for each additional txin will be many times higher than the amount sent.

[Mageant]

If this really is a problem for Bitcoin, then couldn't somebody with bad intentions use it to try to destroy the network?

After all, it wouldn't cost very much, only the minimal transaction fee.

If that is the case then it isn't solved by just persuading SatoshiDice to stop doing this.

[misterbigg]

If this really is a problem for Bitcoin, then couldn't somebody with bad intentions use it to try to destroy the network? After all, it wouldn't cost very much, only the minimal transaction fee. If that is the case then it isn't solved by just persuading SatoshiDice to stop doing this.

As has been stated over, and over, and over, and over....persuading SatoshiDICE is only one prong of a multi-pronged solution. Of course the other part is to implement technical measures to make the creation of economically unviable transactions expensive.

Why won't it read?

[cypherdoc]

Not really because all the dust comes from different places so will require many txin, typically 1 for each loss notification, wherein the fees for each additional txin will be many times higher than the amount sent.

its much more than that.  SD wants and needs to give immediate loss feedback with 0 confirmations.  they do that to encourage repeated and impulsive betting.  they get away with this b/c if someone is trying to do a double spend, the bet tx won't confirm and thus neither will SD's loss notification.

[Jutarul]

... Of course the other part is to implement technical measures to make the creation of economically unviable transactions expensive.

Let me officially endorse this statement.

However, let me also remind everyone that there are legitimate uses for creating dust outputs (e.g. colored coins, contracts). Thus some transactions may "appear" economically unviable, however they there may still be a strong inherent incentive to spent it.

Ultimately what should be the deciding factor is the intent of spending any output at a later stage. One way to give the intent a monetary value is the use of deposits. But there may be other, more refined measures to capture intent in terms of transaction fees. I'd like to see more proposals.

[solex]

... Of course the other part is to implement technical measures to make the creation of economically unviable transactions expensive.

Let me officially endorse this statement.

However, let me also remind everyone that there are legitimate uses for creating dust outputs (e.g. colored coins, contracts). Thus some transactions may "appear" economically unviable, however they there may still be a strong inherent incentive to spent it.

Ultimately what should be the deciding factor is the intent of spending any output at a later stage. One way to give the intent a monetary value is the use of deposits. But there may be other, more refined measures to capture intent in terms of transaction fees. I'd like to see more proposals.

I have been thinking about this and the distinction between spam and virus-like behavior is critical here. Just as your PC/laptop needs anti-spam for email, it also needs anit-virus software for the thousands of more "intelligent" threats.

While the existing Bitcoin mining filters are excellent at transactional anti-spam, they have not been designed for, and tested with, virus-like transactions. SD is showing that this threat is out there. It is doing a huge favor to Bitcoin because anti-virus protection needs to be built-in as well. This would "capture intent" and permit the legitimate uses Jutarul mentions.

From what I hear, SD has already implemented a 0.5% min loss payout, Thank-you Erik! Great news, it helps a lot on the blockchain bloat front.
The larger threat is from external virus-like applications - and this is still out there...

[misterbigg]

Ultimately what should be the deciding factor is the intent of spending any output at a later stage.

Wow...good luck writing that function! I don't think it's possible to encode this in computer code.

[Jutarul]

Ultimately what should be the deciding factor is the intent of spending any output at a later stage.

Wow...good luck writing that function! I don't think it's possible to encode this in computer code.

Yes - there is no way to measure intent. But we could resort to game theory to investigate the necessary rules for an environment in which people are encouraged to remove unspent outputs timely.

[ShadowOfHarbringer]

SD has clearly established, through their refusal to stop attacking Bitcoin, that their intentions are bad - or at the very least indistinguishable from what someone with bad intentions would do.

No, there is no software bug. The bug is in the human component of the protocol.

This is such a terrible bullshit. Can you all please stop creating this bullshit and spreading FUD ?

You simply don't get the point: If the HUMAN behavior is a problem which the network cannot solve itself using its own algorithms, then the network is useless and doomed to failure.
There can be no such thing as human component-based bug in the protocol, because the protocol is designed to cope exactly with that.

Bitcoin is still on very early stages. If BTC at this stage cannot cope with such a minor disturbance as SatoshiDICE currently is, then it cannot cope with reality at all and should be either fixed or redesigned from scratch.

The point in short:
The whole Bitcoin network exists exactly for the reason of coping with human imperfections (like forgery, double spending, thievery and other cons), so if it cannot do that, that means it is *completely useless*.

Did i make myself clear this time ?

[solex]

SD has clearly established, through their refusal to stop attacking Bitcoin, that their intentions are bad - or at the very least indistinguishable from what someone with bad intentions would do.

No, there is no software bug. The bug is in the human component of the protocol.

This is such a terrible bullshit. Can you all please stop creating this bullshit and spreading FUD ?

You simply don't get the point: If the HUMAN behavior is a problem which the network cannot solve itself using its own algorithms, then the network is useless and doomed to failure.
There can be no such thing as human component-based bug in the protocol, because the protocol is designed to cope exactly with that.

Bitcoin is still on very early stages. If BTC at this stage cannot cope with such a minor disturbance as SatoshiDICE currently is, then it cannot cope with reality at all and should be either fixed or redesigned from scratch.

The point in short:
The whole Bitcoin network exists exactly for the reason of coping with human imperfections (like forgery, double spending, thievery and other cons), so if it cannot do that, that means it is *completely useless*.

Did i make myself clear this time ?

+1

[Jutarul]

The whole Bitcoin network exists exactly for the reason of coping with human imperfections (like forgery, double spending, thievery and other cons), so if it cannot do that, that means it is *completely useless*.

Did i make myself clear this time ?

Nobody is arguing against that. The point raised by the OP is that the current structuring of transaction fees only takes into account the current network cost of a transaction and doesn't capture the (indeterminate) future cost of keeping that unspent output available.

If you want to understand it in software terms: intentionally unspendable outputs have the same devastating effect as a memory leak. Thus one has to try to either tame them or define a scheme by which the blockchain can be cleared from them once in a while.

[justusranvier]

Thus one has to try to either tame them or define a scheme by which the blockchain can be cleared from them once in a while.

Miners have complete control over which transactions they include in a block. They don't need to use the default rules, nor do they need to change the protocol. A couple lines of code could create the necessary incentive to clean up all of these "unspendable" dust outputs.

https://bitcointalk.org/index.php?topic=150726.0

[misterbigg]

Miners have complete control over which transactions they include in a block. They don't need to use the default rules, nor do they need to change the protocol. A couple lines of code could create the necessary incentive to clean up all of these "unspendable" dust outputs.

Miners have complete control but they also act in their own economic self-interest (just like S.DICE). Cleaning up dust would cost caring miners money:

Not just care, but care enough to forgo some income for it when they take a txn consuming more utxo paying lower fees per byte over one consuming less utxo paying more.

There is no incentive to clean up dust.

[zebedee]

There is no incentive to clean up dust.

Then let's not worry about it!  Great.

[misterbigg]

There is no incentive to clean up dust.

Then let's not worry about it!  Great.

You didn't read. I quoted gmaxwell's explanation for why miners are not really incentivized to clean up dust (hint: it would cost them money).

We do have to worry about it, because the dust makes transactions cost more than their value. I'm not sure I understand your point, do you not get that unspendable outputs are bad?

[justusranvier]

I quoted gmaxwell's explanation for why miners are not really incentivized to clean up dust

And you refrained from quoting the explanation of why they are incentivized.

[misterbigg]

I quoted gmaxwell's explanation for why miners are not really incentivized to clean up dust

And you refrained from quoting the explanation of why they are incentivized.

Okay, I have to admit that I do not fully understand the exact mechanics of your proposal. Are you saying that there is a way to financially reward miners to collect the dust? Because that wasn't apparent from your post, and I think it is what gmaxwell was saying. I would love to be proven wrong of course, because this would solve the spam problem in a way that fixes it for anyone who exhibits the bad behavior and not just targeting one company specifically.