The scrapers will not attempt to discover and use an efficient representation. They will attempt to hit every URL they can discover on a site, and they'll do it at a rate of hundreds of hits per second, from enough IPs that each only requests at a rate of 1/minute. It's rude to talk down to people for not implementing a technique that you can't get scrapers to adopt, and for matching their investment in performance to their needs instead of accurately predicting years beforehand that traffic would dramatically change.

I challenge you to take a critical look at the performance of things like PHPBB and see how even naive scraping brings commonly deployed server CPUs to their knees.