The "attacker" can rent 1000X more botnets than you can. That was the entire point of the "attacker's" blog article.
'NotBack you are going of the deep end a bit here.
The difficulty stratosphere attack you've described doesn't work very well against Monero if you work out actual numbers.
If you wanted to drive up the difficulty 1000x then you would need something approaching a billion typical (old, insecure, low powered) botnet nodes. If you wanted to use higher performance computers say from cloud computing you'd need 10 million or so, which is good portion of the capacity of the big cloud computing vendors. To rent that you would have to displace most or all of their other paying customers (or in the case of vendors such as Amazon or Google, their own usage). That won't happen.
Now lets say you did manage to, somehow, drive the difficulty up 1000x. You would drive the average block time from 2 minutes to 2000 minutes which is around a day and a half. The chain would not completely stall, it would continue to generate blocks at this slow rate. Those blocks would feed into the difficulty adjustment and after a few days the block time would rapidly begin to come down. It would still be slow for quite a while, but the severity would subside. Meanwhile, the blocks would be full of high-paying transactions and the block size would increase. The network would hobble along until it self-healed.
If you tried the possibly more plausible 100x version instead of the ridiculous 1000x version, then the block time only goes to 200 minutes, which hardly slower than Bitcoin on a bad day (I've personally waited over an hour for a block). Again, block size adjustment would start to kick in and clear the transaction backlog. Over time (hours to days, not years) the block time would start to come back down pretty fast anyway.
This ignores that Monero with a billion market cap would probably have a much higher baseline hash rate, meaning not only is 1000x implausible but 100x would probably be as well. And a 10x difficulty attack is just purely money for basically no purpose.
Nothing about this is specific to Monero's algorithm, which probably isn't even all that good. This sort of attack won't work against any current alts with faster base block times and difficulty adjustment algorithms that have been battle-tested not only by malicious parties but by auto-switching pools which do this form of "attack" automatically and routinely by rapidly moving massive amounts of hash rate between coins.
The attack works much better against Bitcoin-style coins (1st gen alts mostly) that start with a higher block time and that maintain a fixed difficulty for a (reasonably long) cycle. You drive up the difficulty during one cycle and the difficulty never adjusts at all until the end of the next cycle so there is an unacceptable wait for the whole cycle (2016 blocks in the case of many early alts that just copied Bitcoin).
Shebly spends more time off the deep end than on land; his distinguished history of eccentricity one of his more endearing attributes.
But after a spanking such as smooth just administered, he won't be able to sit down for a week.
The term "codemonkey" exists to describe the end result of hyperspecialization in software (and concomitant atrophy/retardation of other disciplines).
Physicists have much more of a "full stack" understanding of the world, as software (in all of its complexities) is merely a tool in their belt, to be used in multiscale modeling from below to above.
EG, SETI@HOME/BOINC were seminal in the development of distributed computing, yet developed by astronomers and cosmologists.
Shebly may have once been a whiz at writing printer drivers and pondering esoteric nuances of type theory, but utterly fails to contextualize software within the socioeconomic matrix of Actual Organic Reality.
Thus he thinks it's legitimate to ponder attacks on $10 million market cap Monero as if they seamlessly/transparently apply to $1 billion Monero.
The (self-modulating/regulating/defeating) of Shebly's proposed attack are so far outside his narrow field of expertise, explaining them to him is like trying to explain graphic design theory to a Dunning-Kruger afflicted blind mole.
Notice he still hasn't specified how for how many blocks his attack endures. What intellectual dishonesty!