Kierunkowy74

smallpatatas, 10 months ago (edited 10 months ago)

A few things here.

The first one that comes to mind is that defederation DOES stop your posts from going to Meta's platform when combined with the AUTHORIZED_FETCH server setting, while a simple user-level block may not. Depending on your server's settings, your posts may or may not be available on the open web where Meta could scrape the data - but this is still very different from them appearing in the feed or search results of, say, the transphobic, racist, or antisemitic groups that call Meta home.

This has serious implications for user safety and should not be overlooked. In fact, user safety is one of the biggest issues I have seen people mention when advocating for defederation.

Second: it's not yet clear if threads will allow their users to follow people on Lemmy or Kbin servers. But if they do, their users - including, for instance, the millions of followers of some big celebrity or politician - would be able to uprank posts and influence what you end up seeing. You might have LibsOfTikTok tell their users to brigade any posts critical of them, who knows. Meta's own algorithms could end up surfacing certain posts to their users, making the post rankings here largely a reflection of what Meta wants their users to see.

In other words, there's a lot more to the story than just 'blocking their content' when it comes to why you would want full defederation.

Here are a couple of blog posts that go into more detail around some of the data & privacy issues with federation:

https://privacy.thenexus.today/just-blocking-threads-isnt-enough/ discusses why defederation is much better than user-level blocking when it comes to protecting yourself from Meta

https://www.cacherules.com/blog/2023/6/resistance-is-futile-you-will-be-assimilated-by-meta/ discusses the things that Meta can learn about you via federation that they can't otherwise.

firebat, 10 months ago (edited 10 months ago)

Politicizing the topic through the means of tribalism is certainly an objective of interests against the fediverse so undoubtedly we will see more of that happening as unfortunate as that is. There are absolutely parties interested in pushing manufactured division and creating "us vs them" camps. I certainly hope that most users steer away from the quite frankly silly discussions of why one is better than the other. At the end of the day it should simply be a personal choice based on perhaps aesthetics or some other set of minuscule factors rather than politics. In regards to kbin and it's currently limited functionality: many of today's large platforms began in a very similar fashion. I personally enjoy the quirkiness of kbin and my proverbial hat is off to Ernest for the effort he has put into this.

Niello, 10 months ago (edited 10 months ago)

Actually, the main reason they are safe to eat if you compare them to the US is because chickens in Japan are vaccinated against salmonella. This is more of a quality check and likely not used extensively everywhere, especially on smaller farms or those who personally keep chickens.

ZickZack, 10 months ago

Peertube is inherently very scalable with relatively little cost due to an artifact of all social media platforms: Most of the traffic is driven by a tiny amount of videos/magazines/etc...

For services like youtube, you can use this as a way to quickly cache data close to the place it's going to be streamed: e.g. Netflix works with ISPs to install small servers at their locations to lessen the burden on their (and the ISPs) systems.
But with centralised systems you can only push this so far since ultimately everything is still concentrated at one central location.

Hypothetically, if you could stop this super-linear scaling for each user (you need to pay per user plus overhead generated from managing them at scale), you could easily compete against the likes of youtube simply because, at sufficient scale, all the other effects get ammortized away.

Peertube does exactly this by serving the videos as webtorrents: essentially this means that for every "chunk" of a video you downloaded, you also host that chunk for other people to download. That means that peertube itself theoretically only has to host every unique video once (or less than once since the chunks are in the network for a while), meaning you rid yourself of the curse of linear user scaling against users and only scale sub-linearly with the number of unique videos (how sub-linear depends on the lifetime for your individual torrents; i.e. how long a single video chunk stays available for others).

The costs that remain for every peertube instance is essentially the file hosting costs (and encoding the video, but that also only scales in the number of videos and could be pushed onto the uploader using WASM video encoders).
Storage itself isn't cheap, but also not ungodly expensive (especially since you can ammortize the costs over a long time as you platform grows with storage prices in a continual massive decline).

Platforms like Netflix and youtube cannot do this because

1. Netflix is a paid-service and people don't want to do the hosting job for netflix after having already paid for the service
2. Youtube has to serve adds which is incompatible with the "users host the content" method

In general torrenting is a highly reliable and well tested method that scales fantastically well to large data needs (it quite literally becomes more efficient the more people use it)