

Building trust is hard. It’s easier to trust a few companies than to trust a million unknown servers. It’s why I prefer Wikipedia over amazingnotskgeneratedatalltopicalinformarion.biz when I’m looking up simple facts.
Furthermore, Facebook isn’t selling data directly. At least, not if they’re following the law. They got caught doing and fined doing that once and it’s not their main mode of operation. Like Google, their data is their gold mine, selling it directly would be corporate suicide. They simply provide advertisers with spots to put an ad, but when it comes to data processing, they’re doing all the work before advertisers get a chance to look at a user’s profile.
On the other hand, scraping ActivityPub for advertisers would be trivial. It’d be silly to go through the trouble to set up something like Threads if all you want is information, a basic AP server that follows ever Lemmy community and soaks up gigabytes an hour can be written as a weekend project.
Various Chinese data centers are scraping the hell out of my server, and they carry referer headers from other Fediverse servers. I’ve blocked half of East Asia and new IP addresses keep popping up. Whatever data you think Facebook may be selling, someone else is already selling based on your Fediverse behaviour. Whatever Petal Search and all the others are doing, I don’t believe for a second they’re being honest about it.
Most Fediverse software defaults to federation and accepting inbound follow requests. At least, Mastodon, Lemmy, GoToSocial, Kbin, and one of those fish named mastodonlikes did. Profiles are often public by default too. The vulnerability applies to a large section of the Fediverse default settings.
I’d like to think people would switch to the Fediverse despite the paradigm shift. The privacy risks are still there if there’s only one company managing them, so I’d prefer it if people used appropriate tools for sharing private stuff. I think platforms like Circles (a Matrix-based social media system) which leverage encryption to ensure nobody can read things they shouldn’t have been able to, are much more appropriate. Perhaps a similar system can be laid on top of ActivityPub as well (after all, every entity already has a public/private key pair).
Mastodon is just one of many applications that uses AP for their own custom purposes. MissKey and derived software has some kind of emoji response feature to posts that’s basically unimplemented anywhere else. Lemmy’s boosting trick to make comment sync make interoperability with timeline based social media a spamfest.
Maybe I should check again, but last time I looked into it there were no commonly used ActivityPub compliant servers. Everyone does their own thing just a little different to make the protocol work for their purposes. Even similar tools (see: MissKey/Mastodon, Lemmy/Kbin) took a while to actually interoperate.
As far as I can tell, the idea behind the original design, where servers are mostly content agnostic and clients decide on rendering content in specific ways, hasn’t been executed by anyone; servers and clients have been mixed together for practical reasons and that’s why we get these issues.