Topic: [If tx limit is removed] Disturbingly low future difficulty equilibrium - page 11. (Read 37667 times)

My prediction was correctly restated by Holy-Fire, so I continue without a repetition.


We are talking about a high block size limit with large amounts of low-fee transactions allowed to be processed by miners. I would call it "eliminating the arbitrary magic value linking market size to transaction fees", but your proposal sounds good too.

Holy-Fire:
I think the argument was that if the transaction confirmations make up a large portion of power demand, a large miner is at a massive advantage. He gets to use his confirmations more often, since he finds more blocks. Competing small miners have to calculate them too, but they find fewer blocks and thus have to throw them away most of the time.

Wait, are we talking about raising the hard blocksize limit, or the free transaction size limit?  Or eliminating the fee scale?


Assuming that this is a correct assesment of the situation, there are other reasons that people mine than direct profit.


A cartel isn't required, as this is how things function right now.  It would take at least 50% of the generators to agree to a rulechange (by upgrading) in order to lift the limit at present.  Even if Gavin Anderson said, "I'm raising the blocksize limit" and changed it in the vanilla codebase, if more than 50% of the miners (by hash power) just refused, there would be nothing that Gavin could do.  I don't agree that eliminating the blocksize limit would create a "tragedy of the commons" scenario, but nor do I expect that the Bitcoin mining collective are going to simply decide to drop the only source of artificial scarcity in the system without great consideration.

Basicly, this could only be a problem if the future miners were generally inclined to act contrary to their own interests.


ASIC's wouldn't raise the barriers to entry in the long term.  Only increase the overall security of the system for roughly the same cost.  ASICs continue to exist in the modern world because they are cheap in quantity.

I don't know much about ASICs, but I can't think of any reason they can't be put on a PCI-e card and sold to consumers to add to their home computer, allowing them to mine with a pool.

Probably because there will be no competition for space in blocks, as it would be eliminated.  There would be no reason for anyone to reject any transactions, they may as well accept all transactions because there's no incentive to only fill a limited amount of space with the highest paying transactions.  Someone could come along and accept all transactions regardless of fee, which in turn would lower people's willingness to pay fees.

Others including [Mike] have made a fairly good point worth considering as a counter-argument to "tragedy of the commons" I brought up before: that a "cartel" of miners who collectively do more than 50% of the mining can simply reject blocks that are too generous in terms of low fees.  They could effectively make it impossible for miners to include transactions on terms they didn't like, because if they did, they would find their blocks quickly undone.  There is possibly no analogy to bus driving or airlines, except perhaps to suggest that if the majority of buses go empty, the operators of those buses have magic powers that could automatically teleport riders on the "cheapo" bus company back to where they started from as a way of forcing people to pay for rides.  A potentially scary power on one hand, especially if future ASICs raise the barrier of entry to mining so high that only well-connected people can get in...kind of like banking now...

And why do you think they will be too low?

Being one of the people who agrees with him, I'd say it's this:

If the limit on block size is lifted (or relaxed too much), the equilibrium reached with respect to transaction fees, and consequentially the incentive to mine, will be such that the total hashing capacity of the network, and thus its resistance to being undermined by rewriting the block chain, will be too low.

I like this statement:

Either it won't be a problem, or if it does become a problem, it will be solvable.

I think that you are projecting now.  I guess I'll say that I don't understand what you think the problem is.  Could you restate your prediction?

I'm kind of relieved to see more people agree we do have a problem.

Now, while I'm glad this thread is being used to clarify a part of the situation, this is touching a topic on which a theoretical discussion no longer suffices. IMO, we need a clear, widely accepted statement on the issue.

Sorry for making yet another thread, but I feel a serious need to do so; here I want to discuss possible solutions:

https://bitcointalksearch.org/topic/necessary-protocol-improvement-dissent-on-future-mining-configuration-6576

Ok, let me untwirl this - I disagreed with you in the first sentence only to agree with in my next sentence: Product heterogeneity and capital requirements will ultimately produce a stable price, you are correct in saying that miners won't drive themselves out of business.

However you miss a more important point: Miners can lower fees substantially in a sustainable fashion. Why? Because other miners who are less efficient will go out of business, which lowers the hashing rate and gives the remaining ones a larger chunk of revenues. Just like competition ensures that there aren't ten bus companies servicing the same route, miners will compete until the equilibrium number of miners is reached.

That means: The invisible hand will make sure 1. all transactions get processed and 2. as cheaply as possible. But there is no intrinsic mechanism to take network security (hash rate) into account. That means the hash rate will be whatever it will be, it could be high enough or it could be too low. If it is too low a mechanism to raise it is provided via the block size limit.

Bus drivers will not drive around empty buses while their competitors take all the passengers. We don't see infinitesimal bus fares on real bus markets because there are few practically infinite buses around.

Also, look at all the airlines selling last-minute and price discrimination seats at below operating costs.

That's exactly my point. The last statement is not true because once a miner does that, fees will fall across the board (assuming they are significant enough to be noticeable). Why pay more when you'll get security anyway?

Bus drivers set one price to get on their bus which is not negotiable because that's the best way to run their business. If they have a big bus (so capacity is not a constraint) then they could easily do what you're suggesting miners will do and accept passengers regardless of how much they're willing to pay. After all they'd be making the trip regardless so once they fire up the engine, extra passengers are just extra profit. But that would be commercial suicide so nobody does it.

I've explained this as clearly as I'm able to several times now, so I won't be trying again. I've satisfied myself that BitCoin will be fine in the long run and haven't been convinced by the arguments otherwise, it's now up to others to decide if they want to continue with the project or not based on the arguments they've read here.

This is true up to the block size limit.

This is only true if there is no block size limit.

How about this one: Mining fees are subject to first degree price discrimination, whereas bus fees are subject to third degree price discrimination. In plain English: When a bus driver lowers his price he lowers it for a whole group of people (standard, senior, children, etc.), so at some point the additional revenue from more passengers is less than the lost revenue from lower rates for his existing passengers. In contrast, a miner can accept transactions with lower fees without losing any revenue from transactions with higher fees.

Generally the factors acting against price discrimination are product heterogeneity, market frictions and high fixed costs. (Shamelessly copied from Wikipedia ) Product heterogeneity applies to Bitcoin (tx with higher fees get accepted faster), as do high fixed costs.

That said, price competition does not care how secure the Bitcoin network is. It will be purely a function of the factors stated above and it will lead to some price, which may or may not be high enough to make Bitcoin robust against takeovers.

My prediction is that the ratio of the network hash rate to the size of the Bitcoin economy will lower quite a bit each time the minting rate lowers. Once minting ends it will be at a level dramatically below where it is now. That alone does not mean however that it will be so low that block chain takeovers become viable. That depends on way, way too many factors to predict decades in advance. We don't know what the overall size of the Bitcoin economy will be, how large the biggest transactions will be, whether large transactions will use new, additional methods of confirmation, what the incentives of taking over the block chain will be, whether there will be terrorists/governments with enough resources to shut down Bitcoin for purely ideological reasons, etc, etc. It is downright ridiculous to speculate on whether the hash rate 50 years into the future will be sufficient imho.

However, what we can predict is that the block chain size limit provides a mechanism for future generations of Bitcoin users to intervene if the security is deemed too low. There will likely be different clients and their developers will likely disagree on the exact security/fee tradeoff (lower block size limit => higher security, but also higher fees), but they will have a very strong incentive to find a compromise. Otherwise whatever client doesn't agree to the consensus will end up running on a separate block chain.

TLDR; Either it won't be a problem, or if it does become a problem, it will be solvable.

I think that this premise is wrong.  I don't believe that individual miners will ever cease to exist no matter how corporate or consolidated professional mining becomes.  For a couple of notable reasons.

First, and foremost, because mining does not have to be profitable to be rational.  I use the example of the lone gaming geek in a cool climate, who lives in a small flat that is heated entirely by electro-resistive heating.  This geek doesn't have to live in a cold climate, just one that has a fairly long heating season.  He is going to own a high-end gaming rig regardless of the existance or success of Bitcoin, and has to pay the electric bill to heat his tiny apartment anyway.  So why not mine while away?  The capital costs are covered by other endeavors already, any bitcoins said lone geek were to get from mining are effectively free.  This doesn't consider the wear that heavy usage imposes upon a GPU, but odds are high that his GPU is going to get upgraded before it dies anyway.

Second, in a successful Bitcoin future, financial institutions are not all going to be into mining for the profits of mining, but also for the digital equivalent of a bank spending half a million on a bank vault.  For the security this contributes for their own membership.  Nor will all these institutions be banks, as credit unions and non-profit institutions exist now and therefore it's reasonable to expect them to exist in the future.  These institutions will participate in the Bitcoin infrastructure for their clients' comfort and benefit moreso than the possibility of (direct) profits.  And these same institutions are likely to develop interlinking agreements to process the transactions of each other's clients for free even if they were to refuse to process free transactions in general.

I contest this premise as well.  The only requirements a mining operation for profit will be reliable and relatively cheap Internet access and electric service.  Both conditions favor developed nations presently, but I can foresee no reason that such operations couldn't cluster in developing nations if the economies of scale would permit said mining operations to bring
these things with them.

I suppose you haven't heard about AWS and similar cloud services? That was rhetorical, of course you have and of course now they're a terrible way to mine, but I don't think that will necessarily remain the case.

I think I understand your argument now and my disagreement with it is off topic for this thread, so I'll leave it at that.

Renting capacity from who? Somebody who has 100,000 unused cpu cores kicking around and will happily rent them to you for a very short time?

I doubt there'd be a big pile of transactions with fees higher than zero but low enough that they never successfully get into a block. I mean that's a possible outcome of my proposed model but there isn't much incentive for anyone to actually create such transactions. If you trust your counterparty just exchange a transaction outside of the network and avoid fees entirely. If you don't then send a fee high enough to buy enough computation to get into a block (you can send this one in a bundle with some free transactions if you need to). It should be easy to calculate such a fee.

Let me state my assumptions clearly:

• Artificial scalability limits will have been removed because most users are on SPV/light clients and BitCoin usage will be growing.
• By the time inflation is low enough that fees become the main driver of the network, mining will have largely consolidated into a number of professional mining conglomerates (this has already happened).
• Individual miners at home with their gaming rigs will cease to exist even in pooled operators because the difficulty will be so high that the returns aren't worth it (same as for CPUs today).
• Mining companies will be legitimate, regulated organizations that choose to locate in developed countries for the usual reasons.
• Mining companies will be using clusters of specialized hardware housed in professionally managed datacenters.
• These firms will be like many others: they will have startup costs (purchase of the ASICs and supporting hardware) and ongoing costs (power, salaries). They will be run in a sustainable manner if only because that'll be a requirement to get the credit needed for setup.

In such a world I see no obvious reason why mining would behave differently to other businesses. People have proposed various arguments for why mining will be different to running a bus company, like "mining isn't local" (so?) or "mining has no startup costs" (it already does). But I don't buy it.

@creighto - If the block size limit is maintained there's no problem. The premise of this thread is that there will be a problem if this limit is lifted.

Also, currently there are no fees because what could be gained by them is negligible compared to generation reward. In the future that will change.

You still haven't explained what stops someone from renting capacity for a short time. It doesn't have to be cutting-edge mining-specific efficiency, the bulk of low-fee floating transactions can make up for the inefficiency.

If you are right and this will be impossible, it means Bitcoin has failed its role as a decentralized currency.

No, transaction inputs are phrased as txhash:output index. You can easily reference transactions that were not included in a block yet.

I think there's some confusion here about how this part of BitCoin works. Any number of transactions can be in the memory pool and they can depend on each other (grep main.cpp for mapDepends). The block chain is not valid unless a transactions dependencies are also in the chain, but BitCoin will happily accept transactions that depend on things it has not seen before. These are referred to as orphan transactions and they wait in the memory pool until their dependencies have been satisfied. Then they are laid out in the block in dependency order.

Mining (in the future) will not have low barriers to entry. The barriers to entry are already non-trivial .... if you want to mine at anything approaching a competitive hash rate you need to either have lots of people in your pool, or build clusters of a very particular type of graphics card. In future you'll need to invest in special hardware (mining ASICs), contract out datacenter floor space and power capacity etc. It will cost you time to set up. In the long run it will be at least as expensive as buying a few buses.

Big miners will understand the dynamics of their own business. There is no need for "mining police" for the same reason there's usually no need in other markets.