There is no Tragedy of the Commons. What there is, is an underprovision of a Public Good.
 I meant to begin by pointing out what an idiot the quoted poster is. Timo Y, you are confused. Please consult wikipedia.
The stable long-run equilibira are:
a) fully monopolized mining under a benevolent 51%er
b) fully monopolized mining under a malicious entity (perhaps a competitor like paypal)
An additional unstable equilibrium is:
c) competitive mining
The equilibria (a) and (b) are absorbing states. The equilibrium (c) is unstable because it is profit-maximizing for one miner to buy out all the other miners and organize a monopoly.
Here are some rank orderings of the equilibria along different dimesnions.
In terms of supply of security (higher is better):
In terms of txn fees (higher is worse):
In terms of incentives of the monopolist to invest in complementary technologies (higher is better):
In terms of incentives of other actors to invest in complementary technologies (higher is better):
It is ambiguous whether c or a is the first-best outcome for bitcoin. In any case, we are likely to end up with a.
Thanks for illustrating the problem. Picture this scenario:tl;dr
- if I'm right, then the total cost of mounting a 51% attack in the "stable steady state future" is negatively affected by added competition between honest miners ... so someone trying to mount a 51% attack can just build/acquire more mining corps (at 0 net cost) in order to lower the cost of attack). Please read onward and tell me where I made a mistake:
- Assumptions: In 40 year(or 140, doesn't matter), block reward is essentially zero, and the major income of miners is from transactions fees.
- There are N different non-monopolistic competing mining cooperation (this is the situation most beneficial to users - c).
- For sake of simplicity, assume that all mining corps are equal in size, and neglect any freelance miners.
- This is a steady state - the margins that these cooperation make are very close to zero, and there are almost no fluctuation in the price of BTC (compared to say the Consumer Price Index).
Now, every block (10 minutes) there is some average amount of transactions, for which the sum of the market price for the transactions fees is T. So, every block, a mining corp will make T/N in gross income, and so his expenses are very close to T/N per 10 minutes. We can see that the market worth of such a mining corp will be proportional to T (some constant times T), assuming N is fixed, because the amount of net revenue for a mining corp is proportional to T, and it's not economical to be mining if your net worth (or cost of construction of the corp) is many times higher than T X constant_multiplier.
Note that this is unlike the situation today, where a miner or mining corp's "cost of construction" or net worth is roughly proportional to the changing block reward (soon 25) and the fluctuating rate of BTC ... besides we're far from any kind of steady state. But in this theoretical future, I don't see any reason why the economical cost to construct a mining corp will be very different than T X constant.
Now let's say some rich entity wants to gain 51% of the hash power in order to perform double spend attacks. The cost of building N+1 mining corps is T X constant X (N+1). These don't just have to be built, parts or full existing mining corps can be taken over by roughly the same cost expenditure.Now here is the potential problem
. From the above calculations, the cost to performing a 51% attack is proportional to the average market price of transaction fees in a block, times some constant. If a lot of transactions are not conducted on the blockchain, but instead move to 2nd level networks, this number drops ... Note that miners are competative (our assumption), so as N gets larger, the TX fees drop ... and the cost of mounting a 51% attack drops. A new honest mining corp has incentive to start mining if it can get efficient electricity, good mining gear, and other reasons, all unrelated to the fact that the very act of opening a new mining corp lowers the transaction fees and lowers the cost of 51% attacks