"Spiraling Transaction Fees Destruction" of bitcoin

[AnonyMint]

Here is a new vulnerability I just realized exists in Bitcoin. This is pretty fucking bad news unless someone can refute it. I am very surprised no one thought of this before, or did they?

Note I am very sleepy (been awake 24 hours) and I just thought of this, so I could possibly be mistaken.

This goes in the Economics thread, because it an economically driven attack, not a protocol attack.

This is going to be counter-intuitive to your instincts, so stop and think deeply rather than rashly posting some nonsense that I will easily refute.

This is basically a Tragedy of the Commons effect, or perhaps Tragedy of the Anticommons.

The problem is that as more people buy mining hardware they increase the difficulty. This dilutes the profitability of everyone. At some point, any miner can become unprofitable and they must decide to continue losing money or quit.

But they have another option. They can see the available transactions and fees offered for the block before they start computing the proof-of-work solution. So they could optionally only attempt to compute it when the fees are high enough to make them profitable. In other words, they idly listen and filter the transactions by fees and only burn big electricity when fees are adequate.

Thus the collective difficulty is not a barrier to buying more mining hardware. If the miner sees that percentage of blocks of transactions are sometimes sufficient to mine, they have an incentive to add more network difficulty and mine part-time as explained above.

This effect is exasperated by the exponentially rising value of Bitcoin, which makes ROI very high even if part-time, because of future expectation of rising returns on mining. But the price drops, then miners have to stay alive until it gets back up to expected exponential trendline again.

So this is a spiral upwards. The customers can't choose to not pay transaction fees, they will offer fees high enough to get their transactions processed. But the percentage of the network which will process for any fee will always be falling off due to the asymmetric relationship between the collective difficulty and the individual part-time miner's incentive of when to add more mining hardware.

Thus transactions fees will perpetually rise as a percentage until eventually it becomes too costly to do any transactions.

And then Bitcoin dies.

End of story.

P.S. There may be evidence this is happening already. Coin rewards are still high at 12.5% per year, so the transaction fee effect is only just minimal so far, but this will worsen as coin rewards decline in 3 more years, if I am correct in my theory above.

P.S.S. The basic problem is that there is no limit on mining hardware, and no limit on transactions fees. Customers don't have a choice. And the miner's individual profit calculation is not symmetric with the overall difficulty scaling. And the adding of hardware is a one-way function, you can't return it after you already bought it. So much better to part-time mine and filter on fees. The Bitcoin prices sometimes drop below trend too. But the hardware can't be returned.

I guess a rebuttal could be that as they part-time mine the difficulty drops but not immediately, and still more will add hardware when the difficulty drops since that is the new level and fees have increased due to the filtering before the difficulty dropped, which drives demand for more hardware.

See the basic problem is the asymmetry both in the collective difficulty versus individual part-time, and the irreversible (coinductive) function of buying new hardware. And the oscillations always push up the hardware on the upside, but can't remove the hardware on the downside. So miners filter mine once they realize what I have written if they are not already doing it. They may not be doing this yet, because of the still high coin rewards.

[Impaler]

Theoretically valid, but I think it unlikely that the cumulative fee volume will vary by enough each block to make a part time approach attractive.  The miner also has to be worried about their overhead costs (costs incurred even when not mining), Moors law will gradually make any piece of hardware obsolete so it may be uneconomical to mine in a part time fashion even if one is capturing higher then normal fees on the blocks one tries for.

It is possible that even if part-time mining occurs that the effect may simply be a kind of network wide 'Drinking Bird' effect.  Mining pools basically all wait until fee accumulation reaches some threshold before mining begins en mass.  Different cost structures would lead miners into pools based on their electricity costs, low cost pools would begin mining earlier at a lower reward then higher electricity cost pools which would turn on only as rewards pass their desired threshold, the pool would essentially provide the on/off signal to all it's clients to maximize their efficiency.  The overall effect is that hashing ramps up each block but at a rate proportional to transaction fee accumulation, and then crashes after block completion.  If this ends up forcing more fees from users or not I'm not sure.

http://en.wikipedia.org/wiki/Drinking_bird

For further analysis I would examine the Fee per block history and see what the variance is, dose it look high enough to effect miners incentives, has the variance been growing or shrinking over time so that an extrapolation out to the 2030 time frame what would we expect it to be.

[Peter Lambert]

Why is your proposal a bad thing? It will just help the market balance supply and demand issues. If fewer people are sending transaction fees, then that is a signal to the miners to slow down. If people want their transactions processed faster they will add more fees. It is a give and take that was built into the structure of bitcoin from the beginning. It is a feature, not a bug.

[AnonyMint]

Why is your proposal a bad thing?

Because the theory is that there is no equilibrium, and that the transaction fees will increase to far beyond what is reasonable, e.g. imagine 10% transaction fees. Spenders don't really have a choice, they can't spend if they don't pay it.

At some level, the spenders will stop spending and consider their Bitcoin to be lost. Transaction fees of 50%? 75%?

I am still trying to catch up on sleep. Once my full mental facilities return, I will try to explore the theory more mathematically if I can.

The theory may be flawed. Let's try to mathematically prove it is incorrect...

[Peter Lambert]

Why is your proposal a bad thing?

Because the theory is that there is no equilibrium, the transaction fees will increase to far beyond what is reasonable, i.e. imagine 10% transaction fees.

I am still trying to catch up on sleep. Once my full mental facilities return, I will try to explore the theory more mathematically if I can.

I don't think that is possible. If some people refuse to mine transactions under 10%, there will be others who include those transactions to get the fees. Having a low fee might make your transaction slower, but as long as you include some fee the transaction will eventually get picked up by somebody. Bitcoins is still open source, you will never have 100% of the miners agree completely.

Already you run into people who say "we miners are getting abused, let's all refuse to include no-fee transactions", and they can modify their software to only include fee-carrying transactions, but there will be other people who just laugh at them and keep putting the free transactions in.

[BitcoinTraderFX]

@Anony I strongly agree with you. The low transaction fees are one of the main things that first attracted us to bitcoin as traders. Every other market has ridiculous transaction fees (unless you are trading on the floor).

I also fear government involvement in the exchanges. Once they regulate, next comes fees and taxes. Governments shouldn't  be able to touch bitcoin, but they do it by attacking the exchanges located in their countries.

I hope bitcoin stays independent as long as possible, and the technical mining difficulty issue somehow resolves itself. The new ASIC hardware that will be hitting the market over the next 6 months spells lots of trouble.

[Timon2010]

A miner always is a customer and has some coins,
and coins are money.

The miner will always willing to mine to keep
his or her transact. fast.

The miner will mine to protect his or her economy in life.

The miner knows that to protect his or her economy
is to don't do what you described in op.

Remember, a miner is a customer and for his or her best
thus will just keep mining.

[AnonyMint]

A miner always is a customer and has some coins,
and coins are money.

The miner will always willing to mine to keep
his or her transact. fast.

The miner will mine to protect his or her economy in life.

The miner knows that to protect his or her economy
is to don't do what you described in op.

Remember, a miner is a customer and for his or her best
thus will just keep mining.

Arguments of this type can fail to factor in the Tragedy of the Commons or Anticommons. Individual self interests can override the interests of maintaining the shared part.

I haven't yet had time to think deeply about Peter Lambert's posts. I am putting out fires in other threads first. Will be back. Thanks for all the help on this one.

[laowai80]

I wouldn't worry about such issues. Free market will balance everything.

If users are too greedy (low fees), miners delay their transactions.

If some miners are too greedy (forcing higher fees), other miners will take over the work.

If all miners somehow agree to be too greedy, users switch to an alt-coin.

If that alt-coin's miners become inadequate, there is a myriad of other alt-coins already created, there is a choice.

Competition is the best thing since sliced bread.

[jubalix]

Why is your proposal a bad thing?

Because the theory is that there is no equilibrium, and that the transaction fees will increase to far beyond what is reasonable, e.g. imagine 10% transaction fees. Spenders don't really have a choice, they can't spend if they don't pay it.

At some level, the spenders will stop spending and consider their Bitcoin to be lost. Transaction fees of 50%? 75%?

I am still trying to catch up on sleep. Once my full mental facilities return, I will try to explore the theory more mathematically if I can.

The theory may be flawed. Let's try to mathematically prove it is incorrect...

this is why you have peercoin, solves this exact problem

[sdp]

This kind of thing has already happened to 'dust' transactions.  Fees are not normally a percentage of the value of the transaction but in proportion to the size of the transaction.  I also worry about the fees creeping higher and higher until only banks and big companies can use the thing.  I can always sell my stake as the price approaches its value though.  I can buy litecoins or ppcoins.

[AnonyMint]

Thanks for the encouragement. Have you noticed what I realized in the other thread? 

Bitcoin is sending the coin rewards for mining to 0 over time. Thus transaction fees must rise a lot to compensate for the loss of coin rewards. Thus in the future, you may not be able to cost effectively use Bitcoin for transactions, because the fees will be so high.

Bitcoin is for people that fiat currency does not provide an adequate medium of exchange.  Drugs, gambling, remittances, avoiding exchange controls - these are all areas where Bitcoin is useful.

And if you don't get widespread use as currency, then government can pigeon-hole you. Because the masses won't care if the government fucks your currency which they don't use any way.

And as such, Bitcoin users don't give a damn if the value goes back to where it was in 2009.  The transactions will still go through.

No either the security will decrease so it can be 51% attacked easily or the transaction fees will be going very high such as 50% (especially if happens when coin rewards have diminished).

You didn't realize that your future depends on Bitcoin not declining in price!

So now you see the currency is lost when the price crashes! NO INTRINSIC VALUE.

[Peter Lambert]

Bitcoin is sending the coin rewards for mining to 0 over time. Thus transaction fees must rise a lot to compensate for the loss of coin rewards. Thus in the future, you may not be able to cost effectively use Bitcoin for transactions, because the fees will be so high.

But remember also that the network will reach an equilibrium scaled to the size of the demand as demonstrated by the transaction fees. If transaction fees do not rise to replace the block subsidy, then the inefficient miners will turn off their machines. Mining will be left to the people with the most efficient energy uses and for people willing to mine at a loss for ideological reasons or by companies who profit more from the bitcoin network than they lose by mining to maintain that network.

[AnonyMint]

Bitcoin is sending the coin rewards for mining to 0 over time. Thus transaction fees must rise a lot to compensate for the loss of coin rewards. Thus in the future, you may not be able to cost effectively use Bitcoin for transactions, because the fees will be so high.

But remember also that the network will reach an equilibrium scaled to the size of the demand as demonstrated by the transaction fees. If transaction fees do not rise to replace the block subsidy, then the inefficient miners will turn off their machines. Mining will be left to the people with the most efficient energy uses and for people willing to mine at a loss for ideological reasons or by companies who profit more from the bitcoin network than they lose by mining to maintain that network.

And that is why I wrote "or 50+% attack". Drastically downscaling hashrate is very dangerous.

[AnonyMint]

And that is why I wrote "or 50+% attack". Drastically downscaling hashrate is very dangerous.

I don't see where you wrote that, but this is indeed a problem, and that's what it is. When (If) the world become inundated with ASICs which can't make a profit if they're all used at once, the risks of a 51% attack increase.

I can't come up with a solution. It might be to just move to an altcoin with a different POW before the risks get too high. Alternatively maybe it's feasible for Bitcoin to change its POW, but if that's done before the block reward is nonzero it's going to be controversial, as it basically makes all those ASICs nearly worthless.

Well I was writing more about the risk when price crashes and the network hashrate crashes to match.

I can't think of how ASICs makes that worse.

[AnonyMint]

And that is why I wrote "or 50+% attack". Drastically downscaling hashrate is very dangerous.

I don't see where you wrote that, but this is indeed a problem, and that's what it is. When (If) the world become inundated with ASICs which can't make a profit if they're all used at once, the risks of a 51% attack increase.

I can't come up with a solution. It might be to just move to an altcoin with a different POW before the risks get too high. Alternatively maybe it's feasible for Bitcoin to change its POW, but if that's done before the block reward is nonzero it's going to be controversial, as it basically makes all those ASICs nearly worthless.

Well I was writing more about the risk when price crashes and the network hashrate crashes to match.

I can't think of how ASICs makes that worse.

ASICs don't make things worse, per se. The problem comes from reaching the limits of increasing technology.

To see what happens when the SHA256d hashing power of the world sits idle or is used for something other than extending the blockchain, just look at Testnet.

As far as taking "when price crashes" as the assumption, I don't see the basis for that.

If BTC crashes, then less funding for mining (coin rewards worth less), thus difficulty must crash when many miners stop mining, thus network hashrate crashes.

[AnonyMint]

No argument from me on that.

[jaysone]

Why is your proposal a bad thing?

Because the theory is that there is no equilibrium, the transaction fees will increase to far beyond what is reasonable, i.e. imagine 10% transaction fees.

I am still trying to catch up on sleep. Once my full mental facilities return, I will try to explore the theory more mathematically if I can.

I don't think that is possible. If some people refuse to mine transactions under 10%, there will be others who include those transactions to get the fees. Having a low fee might make your transaction slower, but as long as you include some fee the transaction will eventually get picked up by somebody. Bitcoins is still open source, you will never have 100% of the miners agree completely.

Already you run into people who say "we miners are getting abused, let's all refuse to include no-fee transactions", and they can modify their software to only include fee-carrying transactions, but there will be other people who just laugh at them and keep putting the free transactions in.

Do you really think that transaction fees will increase forever?  Why would the bitcoin way of sending money not be subject to competition just like other things in a free market?  Why would I pay a huge premium to send bitcoin if I can send funds via wire, credit card, or bank for around 3.5%?  Pretty sure the miners will need to be competitive to alternatives way of transferring money. 

I know you are a big proponent of free markets (as I am too), so I am interested on why normal supply/demand theory does not work in this instance? I feel that miners will have to competitively price fees to acceptable alternatives.  I think if they raise them too high, they drive away business.  I think they will price their product at the point which supply/demand will allow the maximum amount of fees to be generated.

[Peter Lambert]

Remember that the fees will increase both by amount per transaction and also by increasing the total number of transactions. If enough people use bitcoins, the fees can be low and still support mining by sheer volume. This might require increasing the block size?

[AnonyMint]

I finally had a spare moment to contemplate the variables.

A key factor is the block size. If the block size is unlimited (and bandwidth is an insignificant cost), then unless miners have a monopoly they will accept transactions with fees as low as don't constitute a DoS attack, in order to maximize revenue.

In other words, they would have no pricing power at all (a Tragedy of the Commons) and the system would devolve into a partial-monopoly in order to gain pricing power.

A partial-monopoly in this case is enough % of the network hashrate to delay transactions (by that % of blocks) which do not include a sufficient fee.

If we limit block size, then the system doesn't scale.

If we let the Bitcoin foundation decide when to increase block size, then they control the economic market function, i.e. we've centralized Bitcoin.

Let us assume unlimited block size and partial-monopolies. Thus the transaction fee can always be forced higher in order to generate more revenue for the miners. Thus Bitcoin devolves (as coin rewards diminish) to a system that presents spenders with a choice between include a very high transaction fee or accept an ever increasing delay for confirmation. This will exacerbate as coin rewards diminish and volume of transactions increase.

If we instead assume limited block size, then the Bitcoin foundation will set the transaction fees, not the market.

Bitcoin is a broken design.