Olives @olives@qoto.org

It's not like the U.S. could possibly be racist. Not when the government passed laws where one group of people was punished much more harshly for doing the same thing.

It's ironic when someone makes a meme making fun of people who scapegoat minorities, then someone replies in a ridiculous fashion to basically scapegoat minorities, essentially proving the point of the person who made the original post.

"Twitch claims their policies haven't changed."

Big Tech companies (#Twitch is owned by Amazon) often have two sets of policies. A vague public set of "guidelines" which are there to justify a decision they've made.

And then, there are the more detailed internal policies which are distributed to moderators.

I can't speak for Twitch specifically but there are other companies which are like this.

And hey, I think their rules are silly and censorious, I'm not defending them. #anime

Of course, someone might then ask a question like "what to do if an index from the origin server is corrupt?". As I said, this one is just an idea off the top of my head. It's here to think about how it might be done.

Practically speaking, it might be done differently. For instance, one server? Multiple servers? How might it work? What if a server or set of servers fails?

I dunno how content curation could work. One idea off the top of my head might be to have servers which act as gateways and servers which act as indexers.

An individual server might have some sort of compact index which the indexer could index. The indexer wouldn't store the actual data so to save as much on disk / database space as possible.

The user could have an app (maybe a web app) which queries a set (i.e. for load balancing purposes) of indexers for public posts (the query might denote something like what interests they subscribe to, maybe also negative prompts, maybe someone doesn't like AIArt) and the gateway for visibility restricted posts (and anything else, if the index isn't adequately complete).

They could then hand the results to the gateway to get the actual data.

The idea is that rather than having every gateway having to maintain a complete index of the fediverse (or the accompanying data), the logic of figuring out what is on the fediverse could be off-loaded in a more efficient manner to an indexer. Alternatively, they could do both, if resources permit.

This would still be fairly decentralized, although I suppose that someone might prefer the model of each server being it's own unit of decentralization. If an indexer, or a set of indexers, aren't functioning properly, pick another one / set. I dunno.

Part of the "content problem" might not even be a problem of users or content. It might just be basic content curation / collection. Right now, it is a bit of a hose pipe. Information is indiscriminately pulled in, if anyone shows a sliver of interest in it. Conversely, information that might be interesting is not pulled in, if someone doesn't already know about it.

Practically speaking, there is no way to subscribe to a particular related interest or anything. And that is one of the things mainstream social media kind of does. They guess at what someone is interested in, they guess at what content they might like, and they serve that. Of course, this model has it's flaws.

The fediverse is kind of anti social media in design because it goes out of it's way to *not do that*. It's not that particular users or content don't exist on social media. They do. But, generally, you might not see them and they might be buried by what someone is *actually* interested in finding. Likewise, particular types of users might engage with their own content, and that might keep them busy, rather than the subject being some sort of drama.

Then again, that is just a theory.

What I suspect happens is that someone markets the fediverse as some sort of "safe space" (rather than a "pick your own server" social network*), then someone comes in with that expectation, and that leads to more toxicity than is really needed.

* An idea which is interesting in and of itself.

Even if marketing it as a "safe space" gets a few more users, the so-called growth is not worth the toxicity it creates. Is indiscriminately trying to pull in more users even what is really wanted? Chances are it is never going to be Facebook.

"how to deal with block list mania"

For starters, stop marketing the fediverse as some sort of "safe space" unique among social networks, as if that is ever a thing.

At the very least, the Recall thing which Microsoft wants to add should be opt-in (and the user should understand how it works). Don't show lots of prompts to try to get the user to turn it on.

One of the cardinal rules of privacy is that someone is supposed to *minimize* the amount of data held. This is the exact opposite.

So, that is already one red flag against it.

While it is true that there might be stuff like cached data, and maybe that data should be easier to manage, this is also far more indiscriminate in what it collects.

The utility of it at this time is also largely hypothetical, although even supposing it is plausibly useful to someone, it is very over-engineered for the utility it actually provides and it's not clear why someone needs to use *this particular tool*. But, you know what, let's give it the benefit of the doubt momentarily in this post.

It is a treasure trove of information. If someone lives with an abuser (or with some sort of abusive dynamic going on), that might be troublesome.

It could be legally hazardous. What if someone inadvertently encounters illegal content? What if someone does moderation? Is that content going to be "saved"?

How might the data be managed? Can it be managed? Is it clear what is in it?

https://www.japantimes.co.jp/life/2024/10/10/language/weights-exercise-japanese-grammar/ Interesting post.

Someone tried to explain to me how a location tracking app is supposedly good, but the more I heard, the creepier it sounded. #privacy

I could be more specific with A"I" but that would look weird.

Remember, I say "AI" in quotes. That roughly means artificial "intelligence".

Just because someone *can* do something (although, since it involves "AI", who knows what weird way it might fail) doesn't mean that they *should*.

If you are wondering, the bugs in the instance haven't been fixed yet. I still don't see the local feed.

I'm surprised that Microsoft is still at it with Recall.

I'm trying to use replies to the post instead of QTs here because these can render as links and I don't think people click on links as much.

One reason I don't do this is because I don't really want to create a lot of noise on tags by pulling in every tag that might be vaguely relevant to a post.