Block 209999 - a short history of the Bitcoin civil war

[BubbleBoy]

As you probably know, the block reward is set to halve somewhere around late 2012 - early 2013. It's impossible to say with precision when this will happen because block generation time is somewhat unpredictable, but the first block to feature a 25BTC reward has the number 209999.

We should expect that the mining market will be very mature by that time, the profit margins very thin and the price fairly stable. Since miners can't control the price on the market the halving of the block reward and thus of the mining revenue will be seen as a negative effect by the miners, many will no longer be able to cover their expenses. A period of cutthroat competition should follow, where the last man standing can hope to return to profitability, but not before the difficulty also drops by half. This is necessary since where you previously needed one mined block to cover your expenses, you now require two blocks, so a single block should twice as easy to found as before. In other words, half of the hardware will be tossed before the bloodshed is completed.

But it doesn't have to be this way. Since the miners are more or less unanimous that the change hurts their bottom line, they might chose to dissent from the planned schedule and keep the reward at 50BTC. If they simply patch their clients to remove the halving, it has little effect; as soon as the first fake 50 BTC is claimed the blockchain will split:
 

• the original blockchain will continue with a 25 BTC reward, as implemented by all clients, exchanges, merchants and other parties who have no incentive to patch the reward schedule; this chain will continue to be protected by the fraction of the miners that chose not to dissent and play by the rules
• the split chain of the mutineers with the fixed 50 BTC reward; since only the mutineers accept the patch, the coins mined here are not tradable on any exchange and are effectively worthless.

The revolution has failed and the miners are forced to return to the legitimate chain and cannibalize each other as planned by Sathoshi-san

This is where the plot gets thicker. Assuming a high fraction of miners decide to accept the patch, they can attempt to force everybody else to accept the change: they can fork the blockchain and attack the legitimate chain at the same time. The attack will consist of extending the original blockchain past block 209999 with empty blocks. The mutineers will hardcode their miners to refuse any reorganization and hammer away at extending the legitimate chain with empty blocks only. As we know, if they have 50%+1 of the computing power in the legitimate chain,  this will prevent any transaction from clearing in the original chain. Meanwhile a small fraction of mutineers will mine the alternate blockchain to keep it somewhat secure.

I would expect such a move to be widely advertised beforehand and alternate client builds to be made available so that users can "upgrade" to the fixed reward algorithm. If someone refuses or forgets to upgrade he will find that his client is no longer able to send or receive any transactions. Once most users and services have switched to the alternate chain, the attack on the original chain can decrease in intensity. The minority honest miners will eventually stop defending the original chain since they are unable to get any block they mine to stick in the chain - the majority evil miners refuse to extend their blocks.

What I like about this attack is that miners are not risking too much by attacking the original chain: if the revolution fails they have not wasted resources by mining empty blocks. They are anonymous and it's very hard to prove that mining for an empty block on purpose is a security attack from a legal p.o.v - it's certainly allowed by the protocol. There's also a strong incentive for miners to conspire to change the rules, the individual benefit is in line with the collective benefit, unlike the normal mining game. So if the mutineers can amass a  majority of the hashing power of the network, there's little doubt they can force a rule-change.

Ok, so what stops miners from doing this right now, and increase the block reward to 5000 ? Nothing really. I just believe the block 209999 shock to be a strong catalyst, maybe enough to make the majority of miners participate, and change the rules only slightly so as to maintain overall confidence in the system.

[Disposition]

firsts of all it's >50% and it's not history, good insight never the less, especially to those who are not aware.

Nitpicks aside, I believe we'll cross this bridge when it comes. There are more pressing matters I am annoyed by that's inherent in the Bitcoin network that needs to be addressed.

[evoorhees]

Here's an alternate theory:

If anyone cares about Bitcoin by then, the transaction fees will be a large portion of each block reward, perhaps more than 50 btc. This means the drop to 25 btc reward will not actually be a halving of reward for the miners, making the event less dramatic.

Further, because the drop to 25 btc is totally predictable, we will see miners start to drop out long before the switch (those miners who are least profitable). As this happens, difficulty drops... and this will occur until some equilibrium is reached, again making the event even less dramatic.

In practice then, miner profitability will resemble much more of a curve, not a huge step drop. As such, there will be little incentive to orchestrate a global Bitcoin civil war =)

[mndrix]

So if the mutineers can amass a  majority of the hashing power of the network, there's little doubt they can force a rule-change.

If rogue miners amass a majority of the hashing power, an inflated block reward is the least of our worries.

If the miners can't amass a majority, the worst they can do is slow down transaction confirmations while denying themselves transaction fees. As transaction confirmations slow, spenders will pay higher transaction fees to compensate.  A rogue miner's opportunity cost to stay with the rogue cartel increases and the most efficient miners defect.  The cartel falls apart and Bitcoin survives.

Of course, my analysis could be wrong. It'll sure be fun to watch when block 209999 arrives.

[iamzill]

Wouldn't this make the bitcoin supply infinite? IIRC the geometrically decreasing rewards is the only thing limiting the total bitcoin supply to 21 million.

At least it'll shut the "the hard limit of 21 million limits bitcoin's usefulness" crowd up.

[FreeMoney]

getting 50BTC for block 200000 is just as much part of bitcoin as getting 25BTC for block 300000. If you think miners will want to double the payout for future blocks why wouldn't they double their payment now? And why not 10, 100, 1000x? Maybe they understand that merchants provide 100% of the backing and won't accept non-Bitcoins?

Miners don't change the code I run. They can choose to mine bitcoins or something else, I'm only accepting bitcoins though.

[Littleshop]

I am both a miner and a merchant.  I understand the benefit of the mining reward cut, and actually am LOOKING FORWARD to it.  I would not mine with a pool that was trying to re-write the rules, and I think most pool operators get the logic as well. 

When the block reward goes down, the price of bitcoin will probably rise.  I personally do not think it will double, but it will cause real upward pressure. 

Like the stock market, I believe the bitcoin market will 'anticipate' this change and the price of bitcoin will rise BEFORE that reward change arrives.  There will be no huge spike at the point of change.   Just my thoughts....   

[FreeMoney]

I am both a miner and a merchant.  I understand the benefit of the mining reward cut, and actually am LOOKING FORWARD to it.  I would not mine with a pool that was trying to re-write the rules, and I think most pool operators get the logic as well. 

When the block reward goes down, the price of bitcoin will probably rise.  I personally do not think it will double, but it will cause real upward pressure. 

Like the stock market, I believe the bitcoin market will 'anticipate' this change and the price of bitcoin will rise BEFORE that reward change arrives.  There will be no huge spike at the point of change.   Just my thoughts....   

+1 to all that

[memvola]

I don't get the scenario though. If it doesn't benefit me as a holder of this currency, why would I switch? If the miner crowd, whoever they are, would think that destroying the main chain would make sense, "they" would've already increased the reward, wouldn't they? Let's make the reward 500000 per block, this way we'll all be Bitcoin millionaires, yippee!

Don't worry, great majority of the miners will always be sane enough to know that you can't make any profit after effectively destroying trust in the currency that you want to generate and exchange.

[TiagoTiago]

That interesting attempt at scifi...

[fivebells]

Actually, what will happen is that the hashing rate will continue to drop, and when it reaches 1TH/s, the NSA, CIA and FBI will decide that they can cost-effectively take control of the block chain and use this power to cause further mayhem and disorder in the bitcoin markets.  My crystal ball told me.

[freequant]

This sudden cut of the miners reward is likely a backdoor in the protocol.
It creates a singularity at a point in time where the whole consensus that unites miners will be blown away and replaced by a new one.
Satoshi of course knows that. And he knows that all he had to do to avoid generating this singularity was to model the miner reward as a function of the block number that converges progressively to 0.
Yet he *chose* to put this singularity there.
What could be his motivations? Satoshi knows that his creation is not perfect and may need to be upgraded to adapt to the evolution of technology and society, but he also knows that he designed the network to be a Nash equilibrium and that any divergence from the consensus by less than the majority of the miners will always lead to a negative outcome for the mutiners and a favorable outcome for lawfuls, making it a better startegy to wait for the majority to become mutineers before becoming oneself a mutineer. This behavior makes the network change avert and tames any divergence from the status quo.
The only way to upgrade the network is to disrupt the Nash equilibrium and let the chain fork multiple times. This may seem like a point of no return, but it is not.
The original chain will continue past the singularity and return to equilibrium with the remaining miners that decided to keep mining it. If the new breed of blockchains that were spawned at the singularity fails to lead to a new dominant Bitcoin fork offering the properties of a Nash equilibrium, all the forks will degenerate and  the miners will revert to the only possible fallback: the original blockchain.
At the time we reach the singularity, there will probably be Satoshi around with a new version of the client, but also some powerful entities hidden behind some charismatic personnae trying to push their own version through the singularity in the hope to get Bitcoin fit better with their own agenda.     

   

[theymos]

Halving the subsidy impacts mining profitability in exactly the same way as doubling the difficulty. Many miners will drop out, the difficulty will be reduced, and then things will stabilize again at a lower network hash rate. This is not a huge deal, since the network already has way more computational power than it really needs.

[Clark]

This thread is so refreshing! An actual conversation instead of just a bunch of trolls.

I agree completely with Littleshop's analysis of upward price pressure.

[istar]

The effect of a "war" like that would only destroy the trust and faith in Bitcoin, so it would ruin the value of it.

Since hopefully many miners also save Bitcoins, they dont want those to decrease in value.

[BubbleBoy]

Here's an alternate theory:

If anyone cares about Bitcoin by then, the transaction fees will be a large portion of each block reward, perhaps more than 50 btc. This means the drop to 25 btc reward will not actually be a halving of reward for the miners, making the event less dramatic.

The current fees per block  is around  0.05BTC. Keeping all other things the same, an 100 fold increase in the transaction number will bring the fees revenue to 5BTC per block. Even in this highly unlikely scenario, the miners still stand to loose 45% of their revenue when the algorithm hits (55 -> 30 BTC per block).

Further, because the drop to 25 btc is totally predictable, we will see miners start to drop out long before the switch (those miners who are least profitable). As this happens, difficulty drops... and this will occur until some equilibrium is reached, again making the event even less dramatic.

Why should miners drop out while their are still profitable ? You can argue that they are "morally prepared" for the upcoming onslaught, so they expect to pull the plug on the farm when the 25BTC hits. But to stop in proactive fashion ? That's not rational. The only way a smooth transition could happen is via a price increase as suggested by Littleshop: market anticipates the lower daily liquidity injections by the miners and increases prices, creating a temporary high profit margin. New miners are discouraged to seize this profit margin since they know new equipment can't recover it's cost before the fee drops. So the higher prices are not associated with a difficulty increase.

Wouldn't this make the bitcoin supply infinite? IIRC the geometrically decreasing rewards is the only thing limiting the total bitcoin supply to 21 million.

At least it'll shut the "the hard limit of 21 million limits bitcoin's usefulness" crowd up.

They can change  the algorithm to keep the 21 million limit, and the block reward close to 50BTC for the foreseeable future, for example by adopting Pieter Wuille's curve. Instead of halving to 25 the reward drops gradually to 44 BTC after a half a year, and 39 BTC after a year. So they skew the curve to mine the coins faster than normal, and leave a smaller profit for miners of the future.

Also, assuming the reward remains constant, an infinite expansion is not that bad. After 25 years of constant 50BTC/block expansion the total annual liquidity injected by miners falls bellow 4% of the outstanding monetary base, and that's much less than the FED is printing nowadays. The rebels could also implement a Friedmanian exponential expansion: 50 BTC for 25 years, and then 4% per year expansion after that: 52BTC in the 26th year, 54.1 in the 27 year, etc. This keeps the mining revenue a constant fraction of the monetary base and it's actually quite sound from a macro perspective, albeit inflationary.

Miners don't change the code I run. They can choose to mine bitcoins or something else, I'm only accepting bitcoins though.

That's the whole point: they don't need your collaboration, they can freeze the original blockchain for good. So it's either you "upgrade" or you walk away from your bitcoins and technical investment in the system. Most users and merchants would be economically forced to accept the rule change.

Halving the subsidy impacts mining profitability in exactly the same way as doubling the difficulty. Many miners will drop out, the difficulty will be reduced, and then things will stabilize again at a lower network hash rate.

It's not exactly the same. When the difficulty doubles the miners have no option but to watch their profits margins erode. There's no way they can block new entrants in the game. When the difficulty remains the same and the reward drops they are fully equipped to fight it: by simply mining empty blocks they can keep the community hostage until their demands are met. Empty blocks fetch the same 25 BTC reward they would have earned anyway.

[Ayo_4_Yayo]

What civil war? Confederacy at it again?? 

[theymos]

It's not exactly the same. When the difficulty doubles it the miners have no option but to watch their profits margins erode. There's no way they can block new entrants in the game. When the difficulty remains the same and the reward drops they are fully equipped to fight it: by simply mining empty blocks they can keep the community hostage until their demands are met. Empty blocks fetch the same 25 BTC reward they would have earned anyway.

The fees on queued transactions will pile up and eventually be taken by someone. It might take some time for fees to fund competition, but it will happen eventually.

[P4man]

The current fees per block  is around  0.05BTC. Keeping all other things the same, an 100 fold increase in the transaction number will bring the fees revenue to 5BTC per block. Even in this highly unlikely scenario, the miners still stand to loose 45% of their revenue when the algorithm hits (55 -> 30 BTC per block).

If we dont get a 100x increase in transaction fees by 2013, none of this really matters, as bitcoin will remain an obscure nerd experiment. Note I say transaction fees, not transactions. The current transaction fees are ridiculously low, which is whats you'd want for speculation, which is the vast bulk of current transactions, but if you do actual trade, few would mind spending 10x as much or even more to get fast confirmations.

Why should miners drop out while their are still profitable ?

Because the smart miner will understand that he can get better prices for his hardware if he gets out early.

[Elwar]

So...if all of the miners switch to a new client...

would that mean that I could mine on the existing client and get all of the booty for myself on a network that is used by everyone else?