[deevolution]

A healthy coin depends on a balanced ecosystem of miners and users. If mining becomes unprofitable then the ecosystem will collapse. But if its a coin that miners have significant interest or investment in and they see profitability decreasing due to energy costs, they will be incentivised to find cheaper sources of energy. This likely explains why so many bitcoin mines in china are located near renewable energy sources.

[pests]

It doesn't collapse. Mining might become unprofitable for some miners. They will drop out and the difficulty will decrease. Other miners are still profitable. Once decreased they or other miners might become profitable and start mining thus increasing the difficulty.

It's an equilibrium.

[phoyd]

Mining will be in an equilibrium, but the equilibrium might be at a point, where the network can not be protected from a 51% attack.

Currently, it will cost around $2bn to buy the hashing power to control Bitcoin - that's the price of one Stealth Bomber.

[pests]

Any point could be the point it can not be protected from a 51% attack.

So what if the bitcoin hash rate drops to say even 10% of current levels. Yes, it'll be way easier for other actors to come back and control 51% but you made my point: that can be done right now for $2bn.

No one ever talks about the fallout effects if a 51% attack does occur.

First of all, we can detect double spends. We have the complete history and just because the head chain was changed doesn't mean the old chain was forgotten.

The mere act of performing a 51% attack becoming public knowledge will crash the price and now the attacker gets devalued coins. Might not even be worth it.

Secondly, 51% attacks are hard. I don't know where you got the $2bn number above but I'm guessing its the equivalent of the current network hash rate.

To double spend a transaction 6 blocks back that means you need to mine 7 blocks before the other 49% of the network mines 1 block.

I choose 6 blocks back as thats the number of blocks most software/people consider needed to consider a transaction confirmed. With bitcoin's target of 10mins/block that is one hour of transactions.

It seems that 51% attack really requires a 7/8 = 87.5% attack. *

* Technically this only needs to be sustained until caught up with the current head chain length and then you can resume 51%.

[phoyd]

The network hash rate is roughly equivalent to 1M Antminer S9 rigs, which would cost something around $2bn, perhaps $3bn.

Your attack scenario assumes an attacker who wants to extract a financial gain from his advantage while keeping a long position.

However, there are scenarios where an attack might benefit from a crashing bitcoin or the ability to deliver a plausible threat to trigger that crash

An example could be a nation state who would like to remind its negotiating partner, how easily they could establish control over a network that holds a substantial amount of foreign wealth. China obviously. Another one would be an investor who wants to short bitcoin in a massive scale (This has been done before, think of George Soro's 10bn GBP bet against the Bank of England)

Crashing seems a lot easier that trying to double spend. A majority hasher could simply ignore all blocks that are generated from the minority and mine only their own blocks, containing nothing but dummy transactions between the attackers wallets. The minority miners would at a 50% chance mine a new "honest" block and at 50% mine an attackers block, giving the attackers blockchain an 75 over 25 advantage. Since there are no real transactions in the attackers blocks, there would be no way to move bitcoins around anymore, holding them literally hostage.

If there is a way to make a profit from a large scale attack, then that large scale attack will be made some day.

[icelancer]

Several profitable altcoins are attackable right now if you count a pool controlling 51%. It's not like they're going to flex that power, or even know how to.

Difficulty curves and such cause great opportunity for smart miners, especially now that Nicehash is offline.