Topic: Really not understanding the Bitcoin XT thing... - page 2. (Read 5820 times)

This is a great point goatpig.

The problem with relying on orphan risks are the following:

• A - The cost of orphan risks will very low, such that they are insignificant
• B - Orphan risk costs will be significant, therefore larger better connected miners have an advantage

Either A or B is true, orphan risks are significant or they are insignificant.  

There can be no balance between these two options.  We are potentially talking about 100% of mining industry revenue being financed by orphan risk, when the block reward becomes low.  It is also necessary for some miners to have different orphan risk costs.  If orphan risk is uniform across the mining industry then Greg's comment¹, becomes true.  I have come to this conclusion based on my experience as an analyst in the mining resources space.  Therefore as costs vary by miner, miners will benefit from having better propagation, which must be significant in relation to revenue, therefore the industry centralizes.

In conclusion, financing the mining industry based on orphan risk is not possible or desirable.

1 - The fact that verifying and transmitting transactions has a cost isn't enough, because all the funds go to pay that cost and none to the POW "artificial" cost. - http://sourceforge.net/p/bitcoin/mailman/message/34090559/)

Ok, my point is the fee market it may create could be insufficient and inappropriate to support mining.  Actually I don't see how this kind of fee market helps with any of the potential "reasons".


Yes, I agree with you here, as I said in my comment, I don't think Greg has shown enough evidence to support this point yet:

No you don't. If 2 solutions propagate simultaneously, as a miner you should choose the solution emitted by the larger miner. I'll make the example more extreme but it stands at any value really:

Miner M has 30% of the hash rate, Miner m has 1%.

Assuming the network propagation speed is equivalent at all points of the network, M will always get 51% hash rate support behind its solution faster than m will, for M only needs to propagate to an arbitrary group of miners controlling together 21%+ of the hash rate, whereas m needs to propagate to 50%+. The time to propagate to 51% of the total network hash rate is quantifiable, and in any scenario where propagation speed is non negligible, the difference in time to recruit 51% of the network between a large and a small miner will also be non negligible.

Now let's say m's average block propagation time to 51% of the network hash rate is t, and M's is T. We know t > T.

Now say miner N receives m's block first, then M's block next, for the same height. We have 3 situations (all timers starting when N receives m's block):

1) N receives M's block within T. N should switch to M's block simply because t > T. At this point there is a higher probability M will recruit 51% of the hash rate faster than m would.
2) N receives M's block after T but within t. N should switch to M's block, because there is a high probability M has already recruited 51% of the hash rate.
3) N receives M's block after t. N should stick to m, as there is a higher probability m has already recruited 51% of the hash rate.

The reality is a bit different however. Propagation speed isn't balanced across the network and while T and t are quantifiable, there is a lot of data to gather to even get valid estimates. What really matters is that for m < M you should always assume t >T, and that is enough information for any sane miner to switch to the M's block if it is received within a few seconds of m.

In a propagation race, the larger miner will always have an advantage, and it doesn't need a 51% cartel, it only needs to get a portion of the network behind its solution that is proportionally smaller to difference in hash rate with its competitor.

TL;DR: You don't need to control 51% of the hash rate, you only need to recruit 51% of the hash rate faster than your competitor. If you are larger than your competitor, you are essentially guaranteed to always recruit faster.

Obviously this all goes down the toilet once miners start using fast relay networks. But orphan cost will get flushed out along, so the assumption that a fee market can exists when propagation time is negligible, let alone that this fee market would be healthy, is completely undermined.

As I said, you would need a 51% cartel to do that - your competitors are the rest of the entire network on the earlier block.

If current best block is height A, and some miner/pool transmits block height A+1, everyone on the network will switch to it since it makes no financial sense to stay mining on block height A without 51%+

Along comes another block height A+1 (1ms or 15minutes later)
Unless you have 51%+ switch to the replacement block height A+1, you will be mining on the least mined block height A+1 and thus average losing out because of that.
i.e. without 51%+ doing the same, you average losing out by switching blocks at the same height.

I've skimmed it already several times, but since you insist on the validity of your claims I will give it a thorough read.

Regardless, something can be said about integrating this line to the quote if it relies on a lot more context (16 pages apparently) than the conclusion you emit right before it.

What I quoted above is the last paragraph from the paper titled "A Transaction Fee Market Exists Without a Block Size Limit."  The final sentence you guys are arguing about concludes a 16-page research paper--not just the quoted paragraph.  

Take a look at the full paper if you like: link

It is very curious how people like Peter R and friends do this.

Peter R: "proposition A supports conclusion A, and therefore falsifies proposition B. Therefore, the correct conclusion is B!"

Non-sequiturs presented as rational argument, and the expectation is that someone thinking rationally might agree.

If a node is presented 2 blocks for the same height, it will always prefer the one it received first. It will not relay the second one even though it is as valid until the next block orphans one of them. Therefor nodes are poor relay points for miners and any miner not trying to connect directly to the majority of his competitors is doing it wrong. So I'll insist, nodes aren't all that relevant in block propagation between miners.


The bold part is the issue. A weak fee market is the same as no fee market. It can't sustain mining and does not filter transactions enough. The absence of limit (or an unrealistically large limit for that matter) puts it all on Moore's law.

Orphan cost can essentially be equated to friction. For that friction to be significant, the adoption rate will have to at least constantly match Moore's law. Any engineering breakthrough or any better software propagation scheme will gravely undermine the network. Also, adoption is finite. Moore's law may or may not be infinite, that's up for debate, but certainly has more growth potential than Bitcoin adoption does.

Again, the existence of a fee market is pointless if it isn't healthy.

The cost of orphan is about as bad a metric as electricity to define mining profitability. As the cost of electricity, connectivity is subject to heavy government intervention (just look at China). Let's not reinforce the reliance of the network on yet another manipulated metric.

Lastly, technological improvement happens in thresholds, it doesn't propagate evenly or at a continuous pace. One day you have 20Mb DSL, the next day you have 250Mb fiber. One day you have 3G, the next you have LTE. Adoption on the other hand is a lot more continuous. What do you expect that will do to the fee market if suddenly the cost of orphans is so low that every miner can afford to deplete the mempool on every block for even the next 6 months? Year? 2 years?

But let's admit adoption isn't continuous, let's admit it is a bumpy as technological leaps. What happens if adoption booms and technological leaps do not occur concurrently?

The cost of orphans is not a good metric to establish a fee market because it is very inconsistent. It is poorly distributed and rather unpredictable. It acts as an extra barrier to entry into the mining market, which is meant to be defined by hash rate. Why do you think Chinese miner blind mine?

Also Moore's law doesn't regress. How do accommodate the fee market in a recession or a long term stagnation on hardware friction alone?

We need another metric to underline the fee market precisely because cost of orphans is a bad one. We shouldn't look at cost of orphans as a feature, we should try to reduce it as much as possible.


And this is what we are getting to. No miner has cause to stay out of the relay network, and XT can't avoid this network either, because it reduces the cost of orphans. Any miner that doesn't use it is at a disadvantage. However it will finish off the fee market in XT before Moore's law even gets to kick in.



Imagine Corallo's relay network doesn't exist. Imagine a couple large miners set up private version of that network for their own sake. Suddenly their cost of orphan has globally reduced.

I will not explain why this is bad for the network and why it will give an edge to these collaborating miners. I invite you to reread the previous paragraph if you want more details, it is pretty much all there.

Bottom line, either XT does not implement the relay network and a couple miners will use that edge to wear and tear their competition, or XT will implement it and friction will be so insignificant compared to demand that fees won't be able to sustain decent hash power.

TL;DR what jonny1000 said.


Not really. I am not talking about rejecting blocks outright, only about how the network tells 2 blocks apart in the case of a simultaneous solution.

Assume miner A has 10% hash rate and miner B has 30% hash rate. Assume both A and B find a solution for a block at about the same time. Let's say miner C receives these 2 blocks within a few seconds of each other. Which block do you think miner C should work on? A's or B's? Obviously B's.

Bottom line is, in the case of a propagation race, you don't need 51% hash power. You only need more hash power than your competitor and the rest of the network will prefer your block to his.


How can you conclude this from the previous paragraph. If anything it means there needs to be a primary factor in determining fees that is not the cost of orphans. If you don't want a block size limit, then your prime candidate is enforcing minimum fees, and that's a whole new can of worms.

And it would not suggest zero hash power. It would suggest blocks will be mined by those who emit the transactions directly. There is always an incentive to mine blocks, with or without a coinbase reward. The reality however is that if that incentive is proportionally too low compared to the network's market capitalization, it will lack in security and make malevolent mining profitable.

I never argued that.  I said that the orphan cost creates a fee market in the absence of a block size limit.  


As far as I know, no one has rigorously shown this to be true. The last paragraph of the fee market paper speaks to this:

"We conclude by noting that the analysis presented in this paper breaks down when the block reward falls to zero. It suggests that the cost of block space is zero; however, this would suggest zero hash power, which in turn would suggest that transactions would never be mined and, paradoxically, that no block space would be produced. Happily, questions about the post-block reward future can be explored at a leisurely pace, as we have a quarter-century before it begins to become a reality. Into the distant future then, a healthy transaction fee market is expected to exist without a block size limit."

As far as I'm aware, no one is proposing removing the block size limit. While one exists, be it 1MB or 8MB, there is an opportunity to financially exploit rarity in block space. Are you advocating that a block limit is a necessity to enable a fee market and that a fee market is a desirable feature of the network now it has arisen?

bolded is hilariously hypocritical


What I am adding to this discussion, as you well know judging by your irritated tone, is weeding out the dishonestly presented arguments. I am successful at doing so, your arguments are no exception.

I think the average (and also genuine) bitcointalker can see plainly how you're behaving. Welcome to your new reputation.

No. Anything with "market" in it I don't adhere to. A "fee market" isn't a technical point of view, rather one for game theorists and free marketeers who I would prefer to see in prisons around the world with the misery they have caused. You did do a very good job of describing an aspect that I wanted to expand on to all of bitcoin and her offspring rather than an individual client-so I quoted you. I didn't say it was your point of view, I said it was a point that should be voiced more often.

The fear on these forums is quite a sight to see when there is any slight possibility of being misinterpreted. The frantic back peddling and heavyweight interventions when something may be construed to the detriment of a particular narrative is startling. I find it all very unhelpful but oddly amusing. This is the FUD the OP talks about, which is leaking into his thread now too. I'm just a little ashamed to be contributing to it by defending myself, but hopefully this is the end of it after one last salutation to Carlton Banks.

You need to drop it and add to the discussion not continue to try and lecture me with condescension on what I should or shouldn't do to increase my reputation. Another tart, supercilious reply from you will confirm the trolling, so move along now, there's a good chap.

Please let me explain three points on the idea that orphan risk could create a marginal cost of adding transactions, which could finance mining:

• Some have argued (Gregory Maxwell¹) that if the only significant marginal costs of including a transaction is the orphan risk, then equilibrium difficulty will be too low. The reason is because average transaction fees will be equal to the marginal orphan risk increase and therefore there is no additional revenue to subsidies the cost of hashing. Therefore the equilibrium difficulty will be too low and we have a problem. I am not sure if this is correct because the marginal cost of orphan risk will be unique to each miner, therefore one could draw an industry marginal orphan risk cost curve and some miners would make more profits than others, this dynamic could be sufficient to ensure high enough equilibrium difficulty. Although it implies better connected miners are more profitable and therefore this may increase centralization. This does make the orphan risk idea idea superior to the ideas about arbitrarily fining miners a uniform amount for larger blocks, because this cost is the same for all miners, there will be no cost curve and therefore no cross subsidy for hashing, which will result in equilibrium difficulty being too low.
• Gavin and others have argued that computer science has come a long way in the last 20 years. For example bandwidth speeds and increased and storage costs have come down. This is often cited as a reason supporting BIP101, because transferring 8GB is expected to be much cheaper in 20 years time. I broadly agree with this line of thought. However, ironically this makes me more concerned about BIP101, because these technological improvements (and IBLT/relay network) may mean the marginal cost of orphan risk, when including an additional transaction into a block, will get closer to zero over time or at least become insignificant. Therefore I consider this point a large negative when evaluating the viability of a fee market without a blocksize limit.
• I am also concerned about using orphan risk as the mechanism for ensuring a fee market. Operating this way may imply the orphan risk will constantly be pushed to the limit. This is not good for the health of the network and reduces the ability of the network to effectively and decisively come to consensus about the state of the blockchain. I also consider this point a negative when evaluating the viability of a fee market without a blocksize limit.

1 - The fact that verifying and transmitting transactions has a cost isn't enough, because all the funds go to pay that cost and none to the POW "artificial" cost. - http://sourceforge.net/p/bitcoin/mailman/message/34090559/

If nodes relay block solutions between miners, then yes their connection matters.  If most miners use the Corallo Relay Network, I agree that average node connection becomes less relevant.  Nonetheless, the orphan cost still exists (it is just smaller), which creates a fee market even in the absence of a block size limit.  


What I have highlighted in bold doesn't make sense to me.  How can they agree not to orphan?  What if the block contained an invalid transaction?

seeing as the subject of your mistake agrees with my assessment (that you made a mistake), it appears that your accusation is without merit.

Be careful throwing such accusations around: if you're wrong, it will damage your own reputation, not that of your target.

This can only be done by the equivalent of a 51% cartel not accepting blocks from the rest of the network, otherwise the lower their % is the higher the risk of losing blocks ... which miners don't like doing ...
Also, the "size" of the miner isn't all that relevant to % of orphan blocks - the only obvious reduction there is when they confirm their own blocks.
... and you don't "agree not to orphan" ... either your wording is incorrect or your understanding is incorrect.

Any pool can limit the size of the blocks they are generating and thus reduce their orphans - but at the current limit of 1MB that's marginally relevant at best for any well configured miner/pool.
It will of course mean they are getting less txn fees in their blocks ... so once the block reward gets low enough, that will of course be an issue.
At the moment txn fees are averaging around 1% to 0.5% extra reward.

Right, Moore's law has almost nothing to do with the profit from mining - not sure why Gavin can't see that ... other than the fact that he cares close to zero about mining ...
Electricity costs are the biggest factor for most - and finding a source of cheap power.

Miners/Pools already softcap block size limits.
They also SPV (and supposedly non-SPV on Eligius) mine empty blocks on certain pools ...

I guess I'll repeat what Carlton said:

I was trying to present the XT point of view on the fee market, to which I deduce you adhere (if I did a good job at describing it). I was not expressing my point of view.

"Ground control to Troll Station-1. We have found your escapee."

Allow me to help you and others understand: that sentence you are quoting was spoken from the perspective of the XT design ideology, not as a representation of the views of the person that wrote it. The person you quoted does not believe the view you are attributing to them.

I'm sure you wouldn't want to look like you are misrespresenting the situation, so I thought I would inform the thread (careful, someone might accuse you of being yet another dishonest debater, and you don't want that!)

This is such an important point that rarely gets voiced. Free market evangelists look at Bitcoin and see it as a technology to serve the current status quo if only it had more "market forces" and allowed more "opportunities for profit". The technical trade-offs that are exploitable are seen as designs that should be expanded to make way for more of the same rather than weaknesses in the protocol that should be plugged.