Either you misunderstood me or vice-versa. I'm not exaggerating, i quite literally mean "no miners." If mining is not profitable, and we discount fetishists who enjoy spending huge sums of money on specialized silicon in hopes of eventually breaking even, we can assume there'll be no miners. Don't you agree?
No that is simplistic.
Higher difficulty = Less miners = less hashing power = longer block interval = lower difficulty
Lower difficulty = more miners = more hashing power = shorter block interval = higher difficulty
The idea that you would ever get to no miners as in literally 0.0000000 GH/s mining is just nonsense. The first miners have sunk cost. Once you have hardware your ongoing cost is just electricity. So long before existing ASIC miners become unprofitable new miners will stop deploying additional ASIC rigs slowing the rise in difficulty. Even if difficulty so overshoots to a point that some ASIC miners are unprofitable those with lower electrical costs will still be profitable. The highest cost miners will shut down their rigs, difficulty will fall, and the margins for remaining miners will rise. It is likely this will also overshoot, difficulty will fall far enough that the profits for remaining miners is "excessive" this will encourage more miners to go online (possibly buying rigs second hand) raising difficulty and lowering profitability all over again.
Please describe this scenario where all miners simultaneously stop mining and network hashpower goes to 0.000 GH/s.
You're absolutely right, above quote was a reply to a user who was arguing that if mining profitability tends toward zero, mining is a "zero-sum game." My response is a hypothetical, highlighting the absurdity of that assumption. By definition, a zero-sum game's profitability is provably 0,* which makes it as sound a business as tossing a coin & expecting to profit by betting heads.
The miners/ASICs scenario you're describing is also not without flaws, though these flaws are rooted in different assumptions.
You assume the smooth negative feedback model. At least that's the model that comes to my mind, the standard market self-regulation. For me, that's similar to a feedback loop in an op-amp, or a centrifugal governor on a steam engine (Yeah, i know, but bear with me
I'm guessing you're familiar with at least one of the two, they both are straightforward examples of negative feedback, and really handy for modeling feedback systems). IRL, both fail to stabilize unless all the conditions are just right ( otherwise the op-amp will oscillate, the steam engine will hunt). For mining example, this implies a bunch of stuff. The "corrections" in mining also won't necessarily mimic damped oscillations, the swings will increase in amplitude 'till the system self-destructs.*
TL;DR: There are sound models in which ASICs destabilize & crash Bitcoin value. Being concerned by the impending flood of ASICs is not unreasonable.
*An aside: Any analog electronics guy will tell you that *everything* desperately wants to become an oscillator