I totally understand your perspective, but I approach this from the opposite direction.

From my perspective, there's no "at least" here. My Lemmy posts are public. I have no control over what is done with them after I post them. I am comfortable with that.

The difference between Reddit and Lemmy is not that one protects privacy and they other doesn't. NEITHER is a platform for private discussion.

The difference is that with Lemmy, public means PUBLIC. Reddit, Twitter, and Facebook are also "public" in the sense that there can be no expectation of privacy. But they're "private" in the corporate sense — a single corporate entity retains control of the data. They can, at will, restrict access to that data, without the consent of the users who created it.

And that's not just theoretical; all of those companies have literally restricted access to content that users meant to be public. People can't read the Twitter posts that I made with the intention of them being public, because Twitter now requires an account to read posts and comments. Reddit has restricted access to posts I made with the intention of them being public and readily accessible, because they killed apps and integrations, and implemented onerous access control in an attempt to hoard my data.

They altered the terms, and I, for one, got sick of praying that they would not alter them further.

Lemmy is public. You cannot control who can read it, and you cannot control what they do with it. The difference is that with a truly public platform like Lemmy, my data can benefit the whole world, instead of just some corporation.

If you are looking for a platform for private discussion, Matrix is probably it. But even then, the concept of data privacy only makes sense if you trust all the people that ever have access to the data. If I'm in a Matrix room with hundreds of strangers, I wouldn't consider that "private" either, regardless of the protocol's encryption.

Bad actors will always have access to the posts I make public. On Lemmy, good actors do, too, and nobody can take that away from us. THAT'S the difference.

An instance isn't required. It's not like the current generation of generative AI wasn't trained from web scrapings

Free and open information, like Wikipedia, used to be an ideal. I have used Reddit since 2008 or earlier because it got on search engines and shared information consistently on precise topics. Twitter used to also be this way, but now mostly only puts paid subscribers on search engines.

If you are to organize information around topics, such as a Commodore 64 community, and the protocol openly allows copies to be made via federation, I encourage people to have the attitude that information be treated like Wikipedia content. It sucks now that so much information from 10 years ago has been just entirely lost now that so many deliberately purged their Reddit comments, etc. Tragedy of the commons. And it drags down the entire planet that people squirrel away discussions on topics that are generally public. It's like now everyone wants to monetize even their discussions on Commodore 64 or automotive repair / have behind absolute control or paywalls /etc.

The instance would likely just act as a regular instance and allow normal users on, you couldn't even tell they were using it to scrape data at that point.

There are already a few instances that ignore delete requests

Legally, in EU, you probably cannot scrape an instance of someone else because of the database copyright law. But I have no idea if that applies to being part of the network. Since the other instances send you their content willingly.

Maybe someone should make a license extension to ActivityPub, where instances can communicate what can and what can't be done with the information they publish. Then at least there would be legal clarity. If it can be enforced is another question.

People can already do that without an instance, the same way google indexes the site.