What if scrapers ips are millions of smartphones? If I was as evil as an AI scraper company that is not obeying robots.txt I would totally build/buy thousands of small games/apps for mobiles to use them as jumphosts to scape the web. This is probably happening already.
in my case my application does not use pagination, it uses infinite scroll, even if you had a million devices that use google chrome, they would all load page 1 and if that req/minute progressively decreaasing thing is implemented, once they start scrolling endlessly they would all hit the rate limits sooner or later, the thing is a human is not going to scroll down a 100 pages but a bot will. once this difference has been factored it, it wont matter how many unique devices they bring into the battle