What protects the stability of Bitcoin's value if supercomputers or some other technological breakthrough allows them to be mined at rates we have not anticipated?
The system is protected by an incentive mechanism. As Satoshi said in the paper itself:
He ought to find it more profitable to play by the rules, such rules that favour him with more new coins than everyone else combined, than to undermine the system and the validity of his own wealth.
This means that if someone has access to a supercomputer, more powerful than the entire network combined, it would make more sense for them to use such computer to mine in a honest way and earn more reward than any other miner. Rather than actually use the power to do dishonest things, like to double-spend or to invalidate transactions, which would diminish the network's validity and therefore the value of the attacker's mining reward.
Also, if such computers are ever built, many people would want them, which would eventually redistribute the power. In that sense, ASICs could be considered tiny supercomputers.
Actually, there is a limit of difficulty growth. It is 4x. Therefore it is possible, but highly unlikely, that one connects machine so powerfull, that diffuculty won't "catch up", with the increased hashrate. This being said, block will be generated more often than 10 minutes, it is possible that block will be generated every second, adding 25 BTC to the network any second. This will more likely crash the BTC/USD price.
There is a second thing: Luca Matteis is wrong. He has stated "it would make more sense for them to use such computer to mine in a honest way and earn more reward than any other miner".
Let's consider following scenario:
- bitcoin still exists
- bitcoin is worth 442 USD a piece
- Current network hashrate is 64 PHps
- I have 128 PHps on my own (twice the btc network speed).
I use my awesome hashrate to generate the coins for a 14 days (timespan between difficulty increase). There is 2016 blocks to mine, each awards 25 BTC (tx fee is ommited). It means, there are 50 400 coins (22 276 800 USD) to be generated. I have 2/3 of total network power, so I get 2/3 of the reward, this means, I have $14 851 200. A lot.
Step two: Diffuculty increases 2x.
I am swithing to solo mine. No txs are included in my blocks. None, whatsoever. I do not publish my blocks. Meantime, I am sending those 50 400 to exchanges, shops, casinos, hookers, etc. I am cashing out my money - nearly 15 000 00 USD.
After two months of such behavior, we had 4 difficulty changes. Bitcoin network does not know about this yet.
I am releasing the Kraken^H^H^H^H^H^H Blocks I have held hostage! All transactions from past 2.5 months has been invalidated! Since nobody else worked on my blocks, I have at least: 25200 + 12600 + 6300 + 3150 = 47250 BTC on my own blocks. Moreover! Because the past 2.5 month are erased, my coins mined in first step (33 600 BTC) are back in my possession! So, I have 80 850 BTC in total. I am cashing them out NOW! on multiple exchanges, and request payout immediatelly.
This is the part I have killed bitcoin, btw.
BTC is deep. But I have just sold bitcoins worth $35 735 700. Moreover, it will be pretty obvious, that BTC has been >51%ed. Nobody will trust the BTC anymore, major exchanges will crash, price will hit $ZERO. I have $35M. I can live with that.
Conclusion: Currently, difficulty rises so high, that my >51% rigs will not be worth enough in the next 2.5 months to keep being honest. Satoshi didn't expect ASICs. It is more viable for me to crash BTC in 2.5 months, than to mine.
Actual answer to your question: Nothing protects the stability of bitcoin's value if supercomputers or some other technological breakthrough allows them to be mined at rates we have not anticipated.
Landauer's principle gives some very weak security for Bitcoin mining since Landauer's principle states that erasing a bit always costs k T ln(2) energy where k is Boltzmann's constant and T is the temperature. Erasing a bit of information reduces the entropy of the information so it must generate at least as much entropy elsewhere, and therefore erasing a bit always consumes energy. In reality, an irreversible logic gate for cryptocurrency mining will cost at least around 20*k*T energy in order to overcome random thermal noise. Therefore, as we approach Landauer's limit, Bitcoin mining on a conventional computer must consume energy in proportion to the number of hashes computed.
While Landauer's limit provides some (very weak) security against a sole miner controlling a majority of the mining power, I can think of two ways to get past Landauer's limit.
- Reversible computing: Reversible computing is computation that does not in any way erase any information. For example, since the AND and OR gates have one output but two inputs, the AND and OR gates are irreversible and are therefore not allowed with reversible computation. Reversible computers must therefore use specialized gates. Any computation that can be done by a conventional computer can also be done using a reversible computer using a technique called uncomputation (uncomputation is essentially “erasing” garbage data produced by a reversible computer by running the computation in reverse.). However, since uncomputation requires one to run a computation in reverse, in general, it will take more steps to mine using a reversible computer than it would to mine with a conventional computer. Since reversible computation has this overhead, Bitcoin mining with reversible computers will only be profitable once reversible computers are several times more efficient than conventional computers. Reversible computers whose energy efficiency exceeds that of conventional computers have not been constructed yet, but expect to hear more about them in the upcoming decades when the limits of conventional computation have been reached.
The best way to combat reversible computers mining cryptocurrencies (I am strongly against this approach) is to use reversible computation resistant mining algorithms (SHA-256 is not designed to be reversible computation resistant) which are functions that delete a lot of data and for which it is difficult to use techniques such as Bennett's pebble game to mine these coins using reversible computers any time soon.
- Quantum computing: The quantum algorithm, Grover's algorithm, could be used to speed up the rate and efficiency of Bitcoin mining beyond Landauer. This scenario however is currently a long way off since Grover's algorithm does not offer as much of a speed-up as other quantum algorithms.