> that scraping essentially requires creating an in-memory copy of this HTML before the user-produced content can be extracted, and thus is a violation of their copyright.

Isn't copyright a bound on distribution? How would this work unless the HTML itself was distributed by the scraper? If you use SAX processing to stream it, does that change interpretation? (Not sure about how to interpret the in-memory copy.)

If they violate my copyright, can I sue them for royalties every time they rendered the image they never got a license for and distributed it to someone? I'd imagine their TOS/EULA precludes this by just equating uploading to refusing my license, but how enforceable is that? It neatly bypasses any consumer protections.

If facebook is acting this way, maybe I've been wrong all this time--maybe I shouldn't be paying for their service to avoid ads, maybe they should be paying me for the license to my content AND not showing ads.