Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
How much longer is Twitter going to last?
A few weeks
A few months
A few years
About as long as the rest of humanity
View Results
 
  • Post
  • Reply
Morroque
Mar 6, 2013
Thinking on it, that probably means there isn't enough physical room at the Twitter HQ itself to actually have a non-remote workforce.

... is what I would say, but he's already fired a lot of people...

Edit: Looks like I'm not the only one who thought that...

Adbot
ADBOT LOVES YOU

Morroque
Mar 6, 2013
I think the comparisons are unfair. Mastodon isn't a social media platform unto itself, but instead an implementation of a network API. Such is also the case with the larger internet. Websites with child porn on them statistically exist, but it is within every other website's rights to simply deign not to acknowledge them. That article expects the Fediverse to function differently, when that's not really possible for anyone. If the API is publicly known, and a child porn distributor chooses to make use of that public information, "hiding" that API connection in other areas really is the only available response.

The parts of the article going into the social dynamics of why the sudden influx of digital contrabanding happened was interesting enough, but it shouldn't be Mastodon's problem. It's like saying WordPress-dot-com should be responsible when some darkweb host installs a WordPress locally to liveblog all the crimes they're committing.

Morroque
Mar 6, 2013

oliveoil posted:

I got the impression the founder thinks search shouldn't be allowed to reveal it though, which makes me wonder how much of it was designed with support for this material in mind.

Suppose for a moment that it did reveal the scope of the issue; the only response would be people thinking the search function is broken because it keeps spitting out CP, and then there would be requests to revert it back to what it was before. This might be an extreme example, but it is hardly a different problem from other types of spam and undesirable content cluttering things up.

I don't think it is really possible to interpolate if the Mastodon creators are secretly pro-CP or actually rabidly anti-CP on this matter, because their actions would be the same in either instance. Even Twitter had functionality to exclude certain accounts from search results, which they frequently applied to people posting NSFW materials on the reg.

Morroque
Mar 6, 2013
In my experience, submitting CSAM activity reports is usually handled by the moderators specifically, and usually only after it is either reported to them by other users or detected automatically via an anti-virus-like tool such as PhotoDNA -- after which the reporting process is usually automated. If the police do or do not follow up on it is usually handled by a law enforcement clearinghouse whose acronym I forget, so it is all rather standardized. (In the US, anyway.)

The public availability of the CP in question afterwards is the subject to the whim of the Internet Janitors on duty, so it wouldn't be available long enough to really benefit from the general public being able to report on it. This stuff is the mod team's job, not the general user's.

Morroque
Mar 6, 2013
For the period of about two years, I was using a browser plug-in that automatically removed all algorithm-suggested content and bruteforced the timelines into chronological sorting with retweets and regular tweets in a separate tab. The reason I installed it was because my friends posted irregularly, but the suggestions were near-constant, that it buried the connections I cared about and I didn't learn about the fact that a friend of mine was sick in hospital until two weeks later. (Despite him posting about it.) It was also "suggesting" to me all the porn that one specific contact of mine was Liking, even after I had their retweets turned off. Still don't know why it was snitching on them specifically and not anyone else. The algorithm was getting some weird ideas and I couldn't get it to stop, so I got a plug-in which just nuked it all.

Anyway, because of that plug-in, web Twitter has remained remarkably stable for me in the face of all of this. Either everything getting affected is the stuff I never liked, or the Github developer running it is doing some black magic to keep fixing all the mistakes client-side. It's ironic, really; for all Musk bloviates about wanting to remove all the bloat, there does theoretically exist a paired-down version of Twitter that actually is much nicer to use. ... it's just too bad that version is the one where paid-promotions and advertising do not exist, and all forms of engagement-manipulation are serriptiously removed.

... and those forms of engagement-manipulation did have legs. My Twitter usage went from being on constantly to only needing to check the site once a day.

Morroque
Mar 6, 2013
I've seen the article posted elsewhere. The jist of it is that it was intentional, but not even Twitter's own employees know why. They received the order to do it, and Comms was told to draft up language for it, but that second part either kept getting delayed or just didn't even know what to say.

Morroque
Mar 6, 2013
It's also possible that employees sent the request to Comms regarding "3party vendors" (as the article said) only for that request to remain unanswered because the process does not account for a crucial department no longer being there. That would also be an appropriate reading of the presented items.

Morroque
Mar 6, 2013
While that might be comparatively true, it begs the question of who owns the debt, currently.

Morroque
Mar 6, 2013
The one thing a return to non-Musk ownership will do for Twitter is that the advertisers will likely attempt a comeback. What they want can only be done on a centralized platform. If any one Mastodon instance implements ads, and people don't like it, then those users will just move accounts to an instance that doesn't. There genuinely is no room for advertisers anywhere but Twitter.

Morroque
Mar 6, 2013
As much as I agree that "algorithm driven politics" was a cause for concern, there does still seem to be remnants of it on a few Mastodon instances I know which have a separate feed for "popular, federated" posts. Right now, those feeds are dominated almost entirely by posts complaining about Republicans rolling back healthcare access for women. I have to ask this question: is what I am looking at a genuine response to world events, or is it only there because it agrees with the sensibilities of the instance runner?

Mastodon currently only seems like a more pleasant place to my eyes because fascist content thus far is comparatively difficult to find using any instance main page as a starting point. Is that only because I am not necessarily seeing stuff that I agree with, but rather is not disagreeable enough to be outright repulsive? Would that begin to change the moment even one major conservative content creator/aggregator shows up with a portion of their audience? If the common feed is in any way numbers-driven, then their presence will be immediately felt. Right now, those types seem happy to remain on Twitter, but if the company can't make its interest payments, would that force their hand into the Fediverse? If that happens, would the current systems necessarily survive as they are?

My fear is the fediverse systems I am aware of only appear reasonable because nobody has had a pressing need to game their systems yet. Meanwhile, you can't walk two steps without YouTube "recommending" yet another Jordan Peterson video no matter how much I refuse to watch them, and I don't really see Mastodon having any protections against that sort of thing. Or worse, maybe it is already that way, and I simply don't notice because what it has thus far thrown at me hasn't been toxic enough to make me feel unwelcome there. ... I certainly have been exposed to far more Former US Labour Secretary Robert Reich in the past few months, and while that's not terrible, it is weird.

I'm trying to resist the idea that web software can have editorial biases, and I'm resisting it for some reason I can't quite articulate, but I can't ignore that the evidence is there for it.

Morroque
Mar 6, 2013
It's strange. It is not like Twitter ran their own translation department, so it was probably another external service they were being invoiced for, which the new management decided not to pay. I would have to wonder how long it was that Twitter was stiffing them on it, given it lasted this long.

... but at the same point, when they published the algorithm's source code, the language of a given post was a key point of categorization. That implies language detection is part of the process, which likely requires some same part of the translation toolkit, which might not have happened in-house.

I'm just wondering what else might be broken now. A lot of things, obviously, but I'm just anticipating the knock-on-effects which Musk is obviously blind to.

Morroque
Mar 6, 2013
I guess the thing about doing development directly onto to prod is, even if everything still works, it gives the impression that everything's falling apart anyway...

Morroque
Mar 6, 2013

That Italian Guy posted:

Since they are going by regdate with this, would you mind sharing your (approximate) regdate for Discord?

I thought this was true, but it doesn't seem to be actually the case across the board. I've seen paid subscribers with '18 get the update, but other friends of mine with late '15 regdates have only gotten their chance within the past week.

It is very clear that the new system will likely not keep to the scale that Discord requires. While the old disambiguator system may have been strange and unusual, it at least solved the problem of conflicting usernames as a matter of course. The new system feels automatically flawed in this way. The public namespace of allowable usernames is something which these systems tend to quickly exhaust after 10 years. Discord was free of those problems, but it will quickly develop them now.

The stated reason was that most users didn't know how to use the old disambiguator system effectively; to the tune of 60% of their sampled user base, anytime they asked. That is an issue that could've just as easily been solved through better education on the subject, not by throwing the old thing out entirely.

Morroque
Mar 6, 2013
The worrisome part is that some subreddits do plan to shut down permanently. They are supposedly the ones who have unique moderation concerns, where they need to rely on third-party tools to moderate large subreddits at scale. If the API changes go through, the tools they rely on would probably go offline due to the pricing demands, and Reddit still doesn't have any ready-made equivalent toolkits to replace them. (Especially since any new tools would now need to be made entirely in-house by Reddit, as external developers would need to pay for privilege.) The quality and safety of those subreddits would no longer be ensured, especially if circumstances have them already vulnerable to ongoing troll incursions.

Morroque
Mar 6, 2013
500 post per day wouldn't be so bad if a large majority of the posts weren't so useless. Given how much of the underlying infrastructure is the same, I almost wonder if advertisements and other unprompted "recommendations" would count towards that limit, making the actual number that much less.

Thinking on it, post distribution is also uneven due to the social inequalities of the platform, which means the rate limiting would actively disincentivise looking at and responding to high-engagement tweets. In an ironic way, this would actively cut into whatever any clout-chaser could hope to achieve. The bots and reply guys one has, the more likely any fresh audience would actively avoid you; may have been economically harmless before, but now it is actively costly to so much as look at it.

Morroque
Mar 6, 2013
Early previews suggest that Threads is 100% algorithm-controlled with (currently) no way to force chronological feed. Even following someone is no guarantee it will serve you anything they post, making follows more of a suggestion than a rule.

Morroque
Mar 6, 2013

FlamingLiberal posted:

I haven't been on Facebook in ages but isn't that the same thing they do there

If so, not interested

The thing is, from a social network engineering standpoint, I can almost see why they'd do it. Supposing their prespective, you're going to have millions of users making database requests that need to be sorted, which is computationally expensive -- and gets more expensive the more complex the query is. However with a "recommendation system," if you notice that some accounts are more popular than others, you can save on energy by only sorting a query exactly once and serving it to everyone even if its not an exact match. ... that's the theory at least. It's very cost-efficient for large-scale platforms, but it's also incredibly prone to abuse in actual practice.

Not only is there the whole advertising-and-nepotism angle of manually rigging the recommendation system, it also tends to price out mid-level players even at nominal operation. This was a constant complaint I heard for Youtube, where people were saying even being subscribed to a particular channel was no longer a guarantee you'll be notified when they post a new video. It's an enshittification thing, yeah, but there almost a kind of logic to it aside from that.

Morroque
Mar 6, 2013
Facebook hasn't really recovered from the Cambridge Analytica scandal in the eyes of many, even if the common understanding of it is very different from the specifics of what actually happened. It's not surprising that the Nymwars is still a millstone that hangs around their neck even now.

Morroque
Mar 6, 2013
From the sound of things, Facebook stopped being in the social network business and just transitioned over into being television instead. ... television where you're not allowed to change the channel, even.

Morroque
Mar 6, 2013
So, um... Who, exactly, owns that copyright?

Morroque
Mar 6, 2013
If so, then that's an unfortunately sensible argument.

Morroque
Mar 6, 2013
Musk is confident in not appearing hypocritical here in case he will functionally retain the ability to "block" in terms of banning people from the site, even if the actual block feature is removed for everyone else. The moves only purpose is to consolidate his own power.

Morroque fucked around with this message at 17:09 on Sep 16, 2023

Morroque
Mar 6, 2013
ActivityPub is a World-Wide-Web Consortium standard in the same way HTML is. The W3C has been made subject to the whims of powerful corporations like Microsoft and Google before, so it is not outside of the realm of possibility. However, W3C standards need to be easily-implementable on a practical basis. Any attempt at biasing the implementation of ActivityPub in their favour would be a double-edged sword. If Facebook wants to collect user information from other federations, then the same conduit would also enable those federations to collect data on Facebook users.

It's not a "new" conversation by any means. Pawoo is/was a very big Mastodon service within Japan, but it very quickly became a "federate at your own risk" server due to the raw amount of CSAM it was piping through. I'd wager most well-managed Mastodon servers would just "limit" Facebook -- connecting to it on an as-needed basis as directed by their users, but not going out of their way to promote or "suggest" Threads-based content.

If there is any inherent weakness I can see with the ActivityPub standard as I understand it, (or on the last draft of it I saw,) there's no agreed-upon standard for most forms of moderation. Even simple matters like anti-spam were left to the imagination with vague suggestions, and that was quite noticeable given how clear and practical the other given examples were.

Adbot
ADBOT LOVES YOU

Morroque
Mar 6, 2013
While certainly curious in light of recent circumstances, the anti-TikTok animus the US has is much older than the current Israeli slaughter of Gaza. The US has been trying to undermine TikTok for much longer, including all the way back into the Trump administration, and for reasons which still are still not entirely clear. They may just be using the post October 7th chaos as a retroactive justification.

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply