Self-hosting by itself might make blocking marginally more difficult, but there are other reasons to do it:
- Browsers these days segment caches by origin, so there's no caching benefit to using Google as a CDN.
- With HTTP2, a first-party request is likely to immediately go through an existing (multiplexed) connection, saving a handshake.
- It's arguably better for privacy, as users and legislators seem to be concerned about links to Google leaking IPs (https://news.ycombinator.com/item?id=30135264).
ITP, ETP, and plugins that can block requests based on heuristics will make pretty short work of this. In Chrome, come Manifest v3, plugins won't be allowed to.
So... this is all uglier and more complicated than I thought.
Essentially I don't understand how possibly could free adblocking lists defeat advertisers or trackers if they truly cared about them: simply have a system running with the latest adblock lists against their test site, and if it is able to filter them, have an engineer make a modification—or have the system automatically pull up a pre-made modification or even generate a new one. In addition, the content-driving JS and the site JS could be bundled in one and obfuscated.
Best functioning filters are secret ones and thus only the technically minded minority has access to them.