A solution to the “Tragedy of the Commons” problem in Bitcoin ?

[Sergio_Demian_Lerner]

There is a theoretical problem (https://en.bitcoin.it/wiki/Tragedy_of_the_Commons) that may affect Bitcoins future.

From the article:

The relevance to Bitcoin is a hypothetical market failure that might happen in the far future when the block reward from mining drops near zero. In the current Bitcoin design, the only fees miners earn at this time are Transaction fees. Miners will accept transactions with any fees (because the marginal cost of including them is minimal) and users will pay lower and lower fees (in the order of satoshis). It is possible that the honest miners will be under-incentivized, and that too few miners will mine, resulting in lower difficulty than what the public desires. This might mean various 51% attacks will happen frequently, and the Bitcoin will not function correctly.

There is a possible solution for this problem (if it really is so). The solution is to gracefully push transaction fees to match electricity cost. I think SolidCoind tried that, by manually adjusting the block reward. This is obviously completely undesirable and opposed to the “not centrally controlled” idea behind Bitcoin.
A realistic and simpler solution is to give some extra reward to miners proportional to the number of bytes unused of the 1M available bytes in a block. This extra reward should be fixed forever, and should be much lower than the initial reward of 50 BTC. Obviously this will break the deflationary base of Bitcoin. But anyway, the effect of very little inflation would be almost imperceptible for users, so from any practical point of view the coin would still be deflationary.

Example:
   Initial reward (prize) = 50 BTC, halved every X blocks.
   Extra reward “e” = 0.000005 BTC for each unsent byte. (Maximum = 1M*0.000005 = 5 BTC )

If a transaction of size “s” specifies a fee lower than (s*e) then no miner will include it, since they would prefer getting the extra reward for those bytes unused. The extra reward can reach a maximum of  262800 BTC/year, which is 1.2% of the 21M maximum coins created by prizes (but it will be probably much lower, at 0.12% since blocks will be almost full, e.g. 90%). This is a very low inflation rate.

Let me show how equilibrium is reached when prize goes down to almost zero.

1. If the coin value goes up →
   transaction cost (in USD) goes up →
      people sell coins because fees are to high to transfer money  →
         coin price goes down
         (equilibrium is reached)

Obviously the opposite direction implication is also true, inverting the actions.

2. Too many coins in circulation →
   merchant increase prices since people are willing to pay more →
   coin price goes down →
      transaction cost (in USD) goes down →
         new people buy coins because it's cheap to transfer money →
            coin price goes up
            (equilibrium is reached or price goes down very gracefully)      

Please forgive me for my little knowledge in economics, but this implications seems to be for me very straightforward.

It would be very easy to modify the behavior for Bitcoin  (1 line of code). But it's obviously a hard-fork and needs large consent.

What do you think?

Best regards, Sergio.

Note to the moderator: Fell free to move this thread to the Economics section if you think it should be there.

[d'aniel]

If demand is there, I like Stefan Thomas' idea of lifting the block size limit, and scaling the network up:

Processing more transactions means that hashing is a smaller part of the
overall cost for miners. For example, paying for 50 BTC worth of hashing
per block costs 0.05 BTC per tx at 1000 tx/block, but only 0.0005 BTC at
100000 tx/block.

Number of transactions is a lever that lets us have lower fees and more
network security at the same time. Like Greg correctly pointed out, this
is not worth having if we have to sacrifice decentralization. But if we
don't, it becomes a no-brainer.

Just because I think the inflation schedule should be thought of as set in stone for psychological reasons, I'm hoping for something more along the lines of a block discouraging scheme https://en.bitcoin.it/wiki/Discouraged_block to enforce a minimum tx fee, or Mike Hearn's assurance contract idea https://en.bitcoin.it/wiki/Contracts#Example_3:_Assurance_contracts to directly fund hashing.  If the fraction of miners discouraging blocks that undercut a minimum fee is large enough, it would be uneconomical to go against them.  And joining this "cartel" would be The Right Thing To Do for a miner who has an interest in health of the network.

[Sergio_Demian_Lerner]

The advantage of my proposal it that it generates controlled and predictable inflation without government intervention.

When the blocks are not completely filled, it's is a sign of recession, so the extra reward creates money and stimulates the economy by generating inflation.

On the contrary, if blocks are completely full then not inflation occurs.

[Realpra]

This will not be a problem, if people want to make a secure stable transaction they can put a huge fee on it and be sure someone will prevent it from being reversed.

Inflation is market noise and dangerous.

[SimonL]

This seems to come up time and again. People constantly trying to use excuses to insert an inflationary model into Bitcoin. Now I'm no expert on economics either, but the whole "Tragedy of the Commons" scenario is impossible to apply to the Bitcoin system as it stands simply because Bitcoin doesn't work in a way where this situation will occur. TOTC works on the premise that a free service can be over used, leaving nothing to sustain the system over the long term. But Bitcoin is not, and never has been, a free service.

Let us assume that the demand (or value) of Bitcoins remains the same, since the fear the OP posits is that, all other things being equal, the block halving can't be replaced by fees sufficiently, hence causing a vacuum of incentive to mine thus plummeting of difficulty and failure of the network. As the block rewards halve every 4-ish years, there are two forces at work. Transaction fees are one, miners WILL become more discerning of which transactions are included, and if the demand (or price) of Bitcoin doesn't change, miners will recoup the absence of Bitcoins after the halving by preferring the higher fee paying transactions. So this leaves it to users to cover the cost of the block reward halving, and miners will still get paid. Aditionally, block halving means less Bitcoins are entering the network and will become more scarce, thus driving the price up, even if demand remains the same, hence even when the block rewards halve, the value of the Bitcoins will rise proportionately to meet demand, thus miners get paid.

Either way miners get paid, hash rates will remain the same (thus difficulty will remain unchanged), and the Bitcoin network remains secure. Many people assume Bitcoin miners were meant to be a free service hence applying the TOTC scenario. The Bitcoin network as I see it has no TOTC scenario so your "solution" is moot.

[DeathAndTaxes]

So miners can mine empty blocks and get paid extra for it?

The proposal seems overly complex.  Setting a minimum mandatory fee equal to 0.000005 BTC per byte enforced by the protocol would have the same effect while not rewarding miners for doing less work.

As time goes on I am less convinced that a Tragedy of Commons even exists.   While it is true that block rewards could be driven towards zero it is also true that for some tx the fee (and high assurance of inclusion in the next block) has higher utility than free or nearly free tx.

There are also alternative models.  If Bitcoin fails due to insufficient security then everyone loses.  Large stakeholders lose the most.   I could see a scenario where MtGox provides hashing power to the network.  They could provide it directly or they could simply create a pool and pay miners a fee per GH/s to include all tx.  Since MtGox shareholder equity is directly related to the utility of Bitcoin it is a reasonable expense.

You could also see scenarios where companies which need fast processing (like our company) could pay a pool a flat monthly fee, provide the pool a list of addresses, and the pool gives tx to those addresses highest priority even without a fee included by the sender.  Our company directly benefits from fast confirmations (as it reduces our variance risk/cost) so it is potentially profitable to pay for it.

[nedbert9]

Fee for use model is simple, works and allows the mining operator the latitude to define their own economics.  It allows for no fee charity mining ops which would likely represent a limited hashing capacity relative the the entire network.  No need for an inflationary grand scheme based on velocity. 

D&T's effort to bounty a rational FFU metering scheme in bitcoind was a good start and everyone should start using it.

No free rides unless you're running a charity.

[JoelKatz]

The OP wants to pay miners to *not* include transactions in their blocks? It seems pretty obvious to me that this makes things much worse.

[Sergio_Demian_Lerner]

The OP wants to pay miners to *not* include transactions in their blocks? It seems pretty obvious to me that this makes things much worse.

Strange, it's it?
But is works as good as setting a minimum fee/byte, plus a little automatically-controlled inflation....

[SimonL]

The OP wants to pay miners to *not* include transactions in their blocks? It seems pretty obvious to me that this makes things much worse.

Strange, it's it?
But is works as good as setting a minimum fee/byte, plus a little automatically-controlled inflation....

Stop kidding yourself, inflating the blockchain based on this method means spamming the blockchain with worthless transactions is going to be rife. This is why fees exist, so that miners can add transactions based on a priority that benefits them.

[grau]

There is a different break-even point for each miner, that will become more and more apparent with decreasing bounty and diversity of technologies (FPGA, GPU, ASIC, CPU botnet) deployed. Most miner will soon be out of business. Mining will centralize to those able to keep up with innovation and investment required, those few commanding the majority hashing power will soon build an Olygopoly http://en.wikipedia.org/wiki/Oligopoly and ensure transactions deemed too cheap will not be included to the blocks they create and have therefore marginal chance of being included by the few challenger outside the ologopoly.

The oligopoly will manage transaction costs high enough for their profit but low enough not to provide sufficient incentive for real innovation compared to their technology.

Think of OPEC oil price management vs. green energy.

[DeathAndTaxes]

There is no evidence of Monopoly/Oligopoly being able to sustain themselves in a free market without either government intervention or natural monopolies and bitcoin mining is neither.  There are really no economies of scale, no high barriers of entry.  No mechanism to artificially constrain supply.  No international borders or duties to prevent global competition.

TL/DR:
If Bitcoin mining becomes a Monopoly/Oligopoly it would be the first of its kind.

Think of OPEC oil price management vs. green energy.

OPEC wouldn't exist if it weren't for governmental interference in the free market (constraining supply, subsidizing the cost of fuel, taxpayer support of military intervention).  In the absence of that, the real cost of oil based products would be much higher today (these externalized costs would be internalized which would make oil less attractive).

[grau]

There is no evidence of Monopoly/Oligopoly being able to sustain themselves in a free market without either government intervention or natural monopolies and bitcoin mining is neither.  There are really no economies of scale, no high barriers of entry.  No mechanism to artificially constrain supply.  No international borders or duties to prevent global competition.

Monopolies or oligopolies do not last forever but arise and fail for an other to shortly emerge, there is no point arguing if they last. There will very soon be an entry barrier in form of substantial investment needed to keep up with the bigger player. CPU was "free" for most, GPU soon required incremental investment to what was there, but was still comodity with resell value, FPGA can still be reused eventually but market is no longer huge, then ASIC is a special purpose machine. There you have the entry barrier.  The cheap USB ASIC will not long break-even after a few deployed racks full of those. Most will have to decide to be either in the business of mining with quite a few thousands of BTC or not at all. Sure some idealists will still run USB powered single ASICS but they will not get anything in solo. The pools (those survive) will soon join the oligopoly in interest of their shareholder.

[JoelKatz]

The pools (those survive) will soon join the oligopoly in interest of their shareholder.

The pools have a huge interest in keeping the Bitcoin system vibrant and robust. Otherwise, Bitcoin itself will fall and all their hardware super-optimized for Bitcoin hashing will be sold by the pound for scrap.

[DeathAndTaxes]

The price of BFL hardware is roughly the same (MH/$) as the high end TH/s monsters.  If the USB unit won't break even then neither will the large one.   Still even if it is dominated by commercial entities that doesn't mean a monopoly will form.  Almost nobody mines their own gold however the distribution of supply is so fragmented no single player has sufficient marketshare to control the market (I think the largest player has about 8% of global production).

Also not everyone will mine for profits from tx fees:
* Those invested in Bitcoin may mine simply to protect their investments
* Merchants and other enterprises who rely on Bitcoin may mine to ensure their companies have a network to profit from
* Users seeking anonymity will mine because it allows them to acquire coins which have no "trail"
* Some users may mine for break even in cold climates as a source of "free heat".
* Some companies may mine to gain better hardware utilization (imagine if Amazon put a PCIe ASIC miner in each EC2 server).  When the server is idle it mines and thus any profit improves the overall efficiency of Amazon datacenter.  When the instance is needed it stops mining until it goes idle again.

[grau]

The pools (those survive) will soon join the oligopoly in interest of their shareholder.

The pools have a huge interest in keeping the Bitcoin system vibrant and robust. Otherwise, Bitcoin itself will fall and all their hardware super-optimized for Bitcoin hashing will be sold by the pound for scrap.

I do not see the system faling or not being vibrant nor robust because mining is more centralized. It is just specialization. Some will be merchants, some provide services and some will mine, but all will have entry barriers and giants.

Think of the early internet days where everybody could have as shiny webshop as any big name, then came amazon, but the net did not fail.

Or just think of Satoshi Dice...

[grau]

The price of BFL hardware is roughly the same (MH/$) as the high end TH/s monsters.  If the USB unit won't break even then neither will the large one.

This current pricing. It would be rather unique if there would be no benefit on scale, that BFL will have to give to customer as soon as they get competition.

[DeathAndTaxes]

The price of BFL hardware is roughly the same (MH/$) as the high end TH/s monsters.  If the USB unit won't break even then neither will the large one.

This current pricing. It would be rather unique if there would be no benefit on scale, that BFL will have to give to customer as soon as they get competition.

Why? 

If you can get 1TH/s for $x and it is the best deal are you not going to take is simply because someone else can 1/100th of TH/s for 1/100th of $x?

The idea that they must make the smaller units a worse deal is false.  They *MAY* make the smaller units a worse deal but they cartainly don't have to.
The economics of ASIC (high fixed cost and low per unit cost) make it most profitable for them to offer a competitive price for both small and large miners.

Still even if there was no $150 unit, a $5,000 rig isn't an impossible barrier for a hobbyist. 

[grau]

The idea that they must make the smaller units a worse deal is false.   

cmon, the incremental production cost of a further chip on the board is less than building its own box, interface, cables around it. That BFL prices them equally on $/hash today is just their idea of how to maximize profit until they get competition.

[SimonL]

There is a different break-even point for each miner, that will become more and more apparent with decreasing bounty and diversity of technologies (FPGA, GPU, ASIC, CPU botnet) deployed. Most miner will soon be out of business. Mining will centralize to those able to keep up with innovation and investment required, those few commanding the majority hashing power will soon build an Olygopoly http://en.wikipedia.org/wiki/Oligopoly and ensure transactions deemed too cheap will not be included to the blocks they create and have therefore marginal chance of being included by the few challenger outside the ologopoly.

The oligopoly will manage transaction costs high enough for their profit but low enough not to provide sufficient incentive for real innovation compared to their technology.

Think of OPEC oil price management vs. green energy.

This is a fallacy simply because every miner's circumstances and operating conditions are different. Some places will have cheap power, but expensive hardware, others will mine for a few quick bucks, while others are trying to make solid profits. ALL however realise that any monopoly on transaction processing means they lose, BIG time. No miners will seek a monopoly since to do so will endanger the value of the very coins they are collecting. Cheap or near zero fee transactions will be added for many years to come, and frankly if the fees are too cheap or non-existent on a transaction then the person making the transactions needs to suck it up and pony up the dough for what the majority of miners consider a reasonable fee. Fees in transactions don't in any way mean the blocks get solved faster so your assumption that an oligopoly based on the majority of miners only processing transactions with fees included is completely wrong.

Competition in mining is definitely going to get fiercer as new technologies come on board but everyone is in the same boat, just because one group is surging ahead with a faster, cheaper solution doesn't mean everyone else is standing still, nor does it mean new techs have that much of an edge. GPU mining is cheap and losses are hedged because their equipment can be sold on, it's a low risk and flexible investment, FPGAs or AISCs, very specialised, high cost to implement, long lead times, cheap operating costs. Each has their advantages and disadvantages.