But even if it happens, all you have to do is update the firmware again, and keep mining.
But let's imagine someone actually does shut down 70% of the mining power. What's the real consequence? Blocks will be mined somewhat slower (but not even close to 3 times slower) till the difficulty self-corrects. In the very worst case that will take 2016 blocks, or 14 days.
The "enemy" fork loses massive hash power at a critical moment.
It is also a network security concern. Bitcoin's 50% security definition is already pretty weak compared to most crypto security definitions. If you have a seperate attack that can take out 70% of the network, then you only need to have power equal to that of 30% of the network to launch a double spend attack.
By itself this might not be enough to make an attack feasible, but if the network is vulnerable to enough of these attacks at the same time, they would compound and a succesfully double spend becomes possible.
If bitmain (or an attacker) took down 70% of the mining power but one miner either didn't use any bitmain hardware or was otherwise protected from the shutdown they would very likely have over 50% of the remaining hashrate. This could give them control over the blockchain until other miners got back online and their share of the overall hashrate dropped below 50% again, worse yet a difficulty correction won't fix things if other miners are slow to recover.
https://blockchain.info/blocks
Which one is it?
I'm not them, so I don't know.
That sort of mechanism doesn't build itself, though, so I have to imagine they have at least one scenario in mind...
Also, reason #547 I don't take Bitcoin seriously. One vendor gets popular, and 70% of your infra depends on a watchdog not malfunctioning. That is not a sign of a stable ecosystem.
Most likely this is an attempt at DRM, an ability to remotely shut down any units they've deemed infringing or violating their TOS or haven't paid a renewed licencing fee or whatever (I don't know what their business model is)
The difficulty self-corrects every 2016 blocks, or 14 days if there's no major change in the hashrate. If the hashrate drops by 70% then the worst case is 14 / 30% = 46 days of slow mining until the next correction.
I agree with your overall point; just here to point out the math error.
Think of it as a race, many participants are running the distance with slightly different speeds. You randomly remove 70% of the runners, and your new top runner will be likely just a tiny bit behind the previous top runner, if at all.
The better analogy would be a race to turn over stones looking for the one with the secret on the back but the number of stones was adjusted for the previous number of searchers and now with fewer searchers it will take longer for any of them to find it.
P.S. And btw, it's >50% attack, not 51% attack. You don't need that whole extra percent.
If you can shut off 70% of all the miners (because they run backdoored hardware), then you only need >50% of the remaining 30%, no? Which should be a lot easier.
Shorting 100x before shutting them down sounds like a good incentives for those people.