Soft block size limit reached, action required by YOU

[Jutarul]

When a bet goes to SD, it normally has a fee (unless it's not required).  Same when the bet result comes back from SD.  Sure, that transaction fee cost is built into their profit model, so it "comes from the gambler".  Just like the fees that VISA charges the merchants filters its way into how they price their goods.

SatoshiDice pays its fair share of fees on both sides.  Stating otherwise is a lie.

I agree.

Discriminating transactions based on anything other than fees uses social criteria. That's a slippery slope gentlemen.

[1714986800]

1714986800

[1714986800]

1714986800

[1714986800]

1714986800

[1714986800]

1714986800

[Raoul Duke]

Code:

diff --git a/src/main.cpp b/src/main.cpp
index 9a06dbf..d3fba73 100644
--- a/src/main.cpp
+++ b/src/main.cpp
@@ -384,8 +384,16 @@ bool CTransaction::IsStandard() const
     BOOST_FOREACH(const CTxOut& txout, vout) {
         if (!::IsStandard(txout.scriptPubKey))
             return false;
+        if (txout.scriptPubKey.size() > 6
+         && txout.scriptPubKey[0] == OP_DUP
+         && txout.scriptPubKey[3] == 0x06
+         && txout.scriptPubKey[4] == 0xf1
+         && txout.scriptPubKey[5] == 0xb6)
+            return error("CTransaction::IsStandard : ignoring transaction with 1dice output");
         if (txout.nValue == 0)
-            return false;
+            return error("CTransaction::IsStandard : ignoring transaction with 0 value output");
+        if (txout.nValue <= 10000)
+            return error("CTransaction::IsStandard : ignoring transaction with dust output");
     }
     return true;
 }

You may not be interested in the if (txout.nValue <= 10000)  test, though it also gets the dice you-lost transactions and other UXTO set bloating flood.

This will make the node not relay or mine these transactions. It will, of course, still accept them in blocks.

Is that what I asked?
If so, can I send the 1 BTC to the address on your sig or do you prefer to PM me a new address?
That rule you mentioned, it drops all tx's which are less than 10k satoshis, right?

[gmaxwell]

That's a slippery slope gentlemen.

Ehh... I don't agree— when Bitcoin is used correctly you can't distinguish users like that.  Address reuse is a hazard to all bitcoin users: It hurts everyone's privacy, it will prevent the introduction of post-quantum address types, and it makes people get crazy ideas about what kind of blocking bitcoin users could be forced to do.  SD's position when being asked to make their system less abusive has basically been "lol no. Fix your stuff" ... well, turning that around if SD's use of consistent addresses gets them blocked "lol. fix your stuff". The slippery slope doesn't happen because in the limit if things can be blocked they will be, might as well assume that and adopt behavior that doesn't facilitate the creation of blocking infrastructure.

Yes, it's better if people don't block things, but right now over 90% of many blocks is being taken up by transactions involving _one_ user of the network.  People would be crazy to not block that if they can, as it's monopolizing a shared resource.

I don't agree with some of the arguments presented here that it's in miner's interest to take these fees.  The fees are not very large and while they add up, I don't believe they compensate the miners adequately for the future storage costs they represent (e.g. they cost less than S3 will charge you for the data, even in just the UTXO set assuming that the dust 'communication outputs' are never redeemed)... nor do they compensate miners for the reduction in the value of Bitcoin we get from premature centralization or reduced adoption because of delayed transactions.  Taking them is a certainly short term good thing, but not a long term one.

[gmaxwell]

Is that what I asked?
If so, can I send the 1 BTC to the address on your sig or do you prefer to PM me a new address?
That rule you mentioned, it drops all tx's which are less than 10k satoshis, right?

It is. It blocks all txn that pay to 1dice addresses or which create outputs with equal to or less than 10k satoshis. Your node will fetch them and then drop them without further validation. It will not relay them to other nodes or attempt to mine them... but will accept blocks that have them.

And sure, you can pay to the address in my sig... or if you like send me a PM and I'll give you a fresh address to pay.

[Raoul Duke]

Is that what I asked?
If so, can I send the 1 BTC to the address on your sig or do you prefer to PM me a new address?
That rule you mentioned, it drops all tx's which are less than 10k satoshis, right?

It is. It blocks all txn that pay to 1dice addresses or which create outputs with equal to or less than 10k satoshis. Your node will fetch them and then drop them without further validation. It will not relay them to other nodes or attempt to mine them... but will accept blocks that have them.

And sure, you can pay to the address in my sig... or if you like send me a PM and I'll give you a fresh address to pay.

1 BTC sent. Thanks!

Now let's put the free market at work

[Daily Anarchist]

That's a slippery slope gentlemen.

Ehh... I don't agree— when Bitcoin is used correctly you can't distinguish users like that.  Address reuse is a hazard to all bitcoin users: It hurts everyone's privacy, it will prevent the introduction of post-quantum address types, and it makes people get crazy ideas about what kind of blocking bitcoin users could be forced to do.  SD's position when being asked to make their system less abusive has basically been "lol no. Fix your stuff" ... well, turning that around if SD's use of consistent addresses gets them blocked "lol. fix your stuff". The slippery slope doesn't happen because in the limit if things can be blocked they will be, might as well assume that and adopt behavior that doesn't facilitate the creation of blocking infrastructure.

Yes, it's better if people don't block things, but right now over 90% of many blocks is being taken up by transactions involving _one_ user of the network.  People would be crazy to not block that if they can, as it's monopolizing a shared resource.

I don't agree with some of the arguments presented here that it's in miner's interest to take these fees.  The fees are not very large and while they add up, I don't believe they compensate the miners adequately for the future storage costs they represent (e.g. they cost less than S3 will charge you for the data, even in just the UTXO set assuming that the dust 'communication outputs' are never redeemed)... nor do they compensate miners for the reduction in the value of Bitcoin we get from premature centralization or reduced adoption because of delayed transactions.  Taking them is a certainly short term good thing, but not a long term one.

I'm really against the idea of "shared resources" as it leads to a tragedy of the commons. Exactly what is the "shared resource" being monopolized here?

[gmaxwell]

Some fun stats I posted in another thread earlier:

Here are some recent blocks, their sizes, and the percentage of transactions which are SD involved (meaning they either pay SD or spend input from transactions which paid SD)

Code:

Height  Size    Percentage
224737 163012 82.0755%
224736 498888 94.9111%
224735 163124 27.1875%
224734 249140 93.4021%
224733 351883  6.77507%
224732 498991 85.5789%
224731 249033 67.8657%
224730 169849 55.9459%
224729 487840 76.87%
224728 249091 80.2395%
224727 249204 64.2512%
224726 187733 69.8947%
224725 121400 49.0196%
224724 248878 69.899%
224723 248923 72.5989%

You'll note that there are a few blocks here with lower than typical numbers, these come from miners which are blocking the transactions— they aren't zero because the "involved" definition I'm using is somewhat broader than would be reasonable to use for blocking.

[Jutarul]

That's a slippery slope gentlemen.

Ehh... I don't agree— when Bitcoin is used correctly you can't distinguish users like that.  ... The slippery slope doesn't happen because in the limit if things can be blocked they will be, might as well assume that and adopt behavior that doesn't facilitate the creation of blocking infrastructure.

So you're proposing arms race instead. Let them struggle to get into the blockchain - even if they offer high fees...
Well that's fine with me. Game on! The battle for transaction volume has begun.

Ultimately each miner can set their own policy. Some may have a FIFO in place, others may sort based on transaction fees, others may sort by social criteria (no SD transactions e.g.).

I may have a preference for a pure fee-based policy, but that doesn't prevent others from trying to discriminate differently, nor should it.

[Luke-Jr]

OK, there is a patch for miners to blacklist SD tx's from being included in blocks...
What about a patch for normal nodes to blacklist the relaying/verification of SD tx's and drop them without relaying/verifying?

The patch I posted should prevent relaying, but still checks one direction so it can detect the other.

[gmaxwell]

OK, there is a patch for miners to blacklist SD tx's from being included in blocks...
What about a patch for normal nodes to blacklist the relaying/verification of SD tx's and drop them without relaying/verifying?

The patch I posted should prevent relaying, but still checks one direction so it can detect the other.

Where'd you post?

Edit: Sent you half PSY's bounty.

[Luke-Jr]

OK, there is a patch for miners to blacklist SD tx's from being included in blocks...
What about a patch for normal nodes to blacklist the relaying/verification of SD tx's and drop them without relaying/verifying?

The patch I posted should prevent relaying, but still checks one direction so it can detect the other.

Where'd you post?

https://bitcointalk.org/index.php?topic=149668.msg1595233#msg1595233

[DoomDumas]

A few years (or maybe even months) from now, we will look back at blocking SD like rearranging deck chairs on the Titanic. SD may be shady and abusive, but WELCOME TO THE JUNGLE BABY! Personal feelings and vendettas about inappropriate uses of the blockchain are going to look really silly when serious volume and attacks are unleashed at a dizzying pace in this exponential growth environment.

Indeed, I dont think SD is causing problem, just an idea of waht's incomming, as BTC will gain more and more users.  IMO, SD TX represents a very small fraction of the TX volume the BTC network will have to deal with.  BTC must evolve, because as it is right now, it will not be able to deal with the exponential growth comming.  Another tought, fees will be more and more important as the growth occurs, and halvings happens.

[HappMacDonald]

I think that Satoshi Dice is being used as a red herring and a scapegoat to kick the can down the road.

Luke accuses them of abusing the network. Our bandaid solution is to encourage miners who are amenable to blacklist their transactions (which is their prerogative, sure), and to take other steps to hassle SD (client changes to encourage higher fees to leave SD behind in priority, etc), so that maybe they'll become unprofitable enough that they stop representing such a high percentage of transactions and maybe we'll gain some breathing room before ordinary txn start fighting over their placement in blocks.

The question I am forced to ask is: are we really solving a problem this way, or are we spending our time on whack-a-mole?

SD is unquestionably profitable right now, and the allegation is because they are offloading much of their expenses to the rest of the network (blockchain bloat, premature block size limits reached, etc). So is it our job to classify this as a moral failure in SD and in the thousands of users who play it, to be punished by edict, or as a systems failure whereby we've built a transaction system that allows users to offload their expenses in this allegedly dangerous way?

Since SD is profitable and has to be targetted with filters to resist, what happens when a dozen SD knockoffs launch that evade filtering attempts? If they have profit incentive to do so, what will deter them? Should we really be blocking "1di" address prefixes (give or take), or should we instead be building something into the txn priority protocol that either discourage address re-use or re-spending of funds not yet confirmed?

If people want to spend their individual participatory power to hassle SD, that's not my business. But that probably belongs in a different thread. The problem of txn rate scalability and altering the system if possible to naturally resist anyone from being able to offload their expenses by making the txn fees you pay finance the costs you incur should be what our discussion is about. Be it because we've hit soft limit with SD, soft limit without SD or hard limit.

SatoshiDice is nothing but a symptom of an imbalance in our incentive structure. The sooner we solve the root of that problem, and enshrine this "social contract" luke holds so dear into the protocol itself somehow (where it can actually be enforced by technology instead of whinged about in forum drama), the more time we have to implement the solution before crashing into a wall or losing ground to a competing currency system.

[DoomDumas]

I really think it is better to rise the transaction size, the idea to filter someone like satoshiDice goes directly against all our freedom philosophy. everyone should be able to use the network like if they where anonymous. Whe just can't use information about satoshi to arbitrary censor it. That's just wrong.

thx for reading! i'm new here because i love the principles of bitcoin.

+1

SD's TX should not be filtered out !  Against BTC philosophy to filter/censor normal use of the network.

[Face]

Just noticed a 429kb block by Slush come by as this discussion is ongoing.
That makes BTCGuild and Slush that have doubled the defaults so far.

We talk they get on with it...

[AbsoluteZero]

Just one more Bitcoin user added to the consensus of increasing (or removing) block limit.


More opinions:

* I like Galvin's (half baked) idea of accepting bigger blocks provided that there is 50BTC fees per each MB of block size
This is effectively a minimum transaction fee of close to 0.01BTC/TX.
It is good now but probably too expensive when and if BTC becomes much more valuable.

* Miners who only mine for the reward and generate "empty" blocks should be orphaned by other miners.
If every miner mined empty blocks Bitcoin would have no value

* Satoshi Dice is not bad for Bitcoin, we should let the free market decide what works and what does not.
I do not care if free transactions are included in the block or not.
But for fee paying transactions Bitcoin should be transparent.
We want a new, clean payment system not censorship.

* People suggesting to leave the limit in place and volume can move to other peer to peer networks is very stupid.
So to prevent filling my hard drive with Bitcoin I will fill it with more currencies.

* Satoshi IS a Genius. Both in the technology that he developed and in economics.
But NOT DIVINE. (This is not opinion just a fact)


BLOCK INCREASE SHOULD BE DONE ASAP
It has to be effective in a future block to give a vast majority of users chance to upgrade and prevent a fork.
And at the rate Bitcoin is going we will hit 1MB before this is implemented.

So we are already late.

[gmaxwell]

to take other steps to hassle SD (client changes to encourage higher fees to leave SD behind in priority, etc),

Thats not a step to hassle, it's just free market competition for space. Of all possible solutions it is the most socially neutral in my opinion, but it has an unattractive cost: The 0.005 BTC/KB needed to reliably get ahead of SD would imply a 0.5 BTC fee on a 100KB transaction...

Should we really be blocking "1di" address prefixes (give or take),

The filter on that patch is on bytes of the binary hash160, so it's actually blocking the whole base58 prefix (1dice), not just 1di.  I agree that the blocking is ugly.

or should we instead be building something into the txn priority protocol that either discourage address re-use or re-spending of funds not yet confirmed?

For a long time— even before SD existed— I ran a patch which only permitted two transactions from or two any address per block. I like it: Beyond the issues with privacy, security, and censorship that reuse has, people who reuse are voluntarily identifying which txn are theirs— so we might as well use that data to give more equitable access to the blockchain by rate-limiting reusers.  But other people didn't like it, and I got called names over it. ::shrugs::

There is a fundamental imbalance though: Priority vs fees. Long term costs and profits vs short term profits.  Because no one knows how to evaluate the long term impact of low priority txn blocking out high ones picking things based purely on fees is surely attractive.  The first time a miner misses out on a crazy high fee because of the above mentioned rate-limiting, I expect they're going to rip that crap right out. Thoughts?

[niko]

I don't get it. SD transactions include fees, correct?  What is the rationale behind blocking these transactions as opposed to, for example, all transactions with more then 13 inputs, or transactions below BTC0.00020477???  All of these would lower the "bloat". What am I missing?

 

[CIYAM]

We do not have blockchain bloat due to *all tx's with more than 13 inputs or tx's below 0.0002x* but we *do* have blockchain bloat due to SD (i.e. the source of the current soft limit needing to be raised is just them).

Of course if SD would like to buy me a new hard drive then I would be more than happy for the (hard) limit to be increased just for them.

[Largo]

I dont think what SD does is good, but blocking addresses isnt really in the spirit of bitcoin