You are welcome to look at the talkback but please consider that this article is over 1 month old before posting.
Forum - View topicThis Week in Anime - Crunchyroll Silences Their Viewers
|
Goto page 1, 2 Next Note: this is the discussion thread for this article |
| Author | Message | ||
|---|---|---|---|
Vaisaga
 Posts: 13235 |
|
||
|
Manga Plus just updated their comment section guidelines too in a response to a certain user spamming antisemitic comments on every chapter. Didn't go so far as axe the comment section entirely.
I am something of a free speech absolutist though. Let people posted all the hateful comments they want. All it does is expose them for what they are and let's everyone know to avoid them. This just goes for words, though. I've been spammed with guro images because I didn't support a certain ship and those sorts of images should certainly be banned. |
|||
|
|||
Joe Mello
 Posts: 2290 Location: Online Terminal |
|
||
|
I'm a little surprised that we went the entire article without saying "enshittification." I'm less surprised that everyone's favorite speculatory reason (criticism of AI) came up, mostly because I don't think a company like CR cared enough for that to be the last straw.
|
|||
|
|||
Fluwm
 Posts: 947 |
|
||
|
I definitely think that if a platform cannot or will not moderate its users, it shouldn't have a comments section/forum in the first place. The loudest and most toxic voices will always dominate those spaces -- just look at Steam.
And, I gotta say, I'm very skeptical about how much utility there actually is in AI-moderation. I suppose it would be effective at blocking the most obvious and overt problematic content, but I'm not sure it'd be any better at that than, say, a profanity filter. People, in general, have gotten very good at using implication and euphemism to launder their hate, which really does require a pair of human eyes to sift through. Like there are plenty of completely innocuous-seeming posts here that moderators have to really examine, because they'll do things like just off-handed remark or reference things from neo-nazi communities (EG dogwhistles, conspiracy theories, etc.). Things that even a human would have trouble recognizing easily -- I can't imagine AI would be any better.
I feel like this is sort of vaguely-implying that things have gotten better in this space, but it really is unfortunately how few JDramas are getting licensed. Sure, there are more available to watch through legit channels today than there used to be, but it's still a very small, limited number. |
|||
|
|||
NeverConvex
 Subscriber Posts: 2456 |
|
||
|
I really like the idea of AI as a tool for flagging content for a human moderator to review. That seems like a perfect use-case for it.
It just directly moderating I'm more skeptical of, although maybe it'd be workable if specifically (re-)trained for that task, established to have a relatively low false positive rate, and there were a human appeals system that would hopefully not get gunked up with people disputing legitimate moderation (although, I can already imagine that's not how it would go, especially in the beginning  ). And, as the column notes in a few places, stakes matter; would be more willing to accept that in e.g. Youtube comments or even Reddit than in a long-form forum environment.
|
|||
|
|||
Nebs
 Posts: 386 Location: University of Illinois |
|
||
|
This reminded me of an Anime Central convention many years ago (10-ish years, maybe more?). Miles and Victoria had a Crunchyroll panel at like 9am on a Sunday, and it was just so fun and chill. Up to that point I wasn't familiar with either of them, but I instantly followed them on Twitter following the panel. I thought it was so cool that people like them were running Crunchyroll.
I like the current team of LaurMoor and Tim Lyu, but the company definitely doesn't have the same vibes as I experienced that one Sunday morning. |
|||
|
|||
Greed1914
 Posts: 4546 |
|
||
|
It makes me a little sad to see Crunchyroll Hime's image on there around a topic involving cutting costs and reducing customer interaction. Up until the end of 2023, Hime was streaming on a regular schedule. Then, after months of not seeing or hearing anything, she had a video saying she wasn't gone, but was switching types of content. Basically, not streaming and switching to videos. But, that video saying she wasn't gone was also the last thing we've seen.
|
|||
|
|||
|
Tenebrae
Posts: 489 |
|
||
|
I've never thourght it is a good idea to let the interne have a comment section for their use in news articles, even less so with little or no moderation.
A number of years back, a local news broadcaster added comments to their news articles. Needless to say, that has been mostly walked back now, the comments section is available in very select news items only, and AFAIK works in "every post is flagged for moderation until cleared by mods" manner.
And that's just a matter of properly training the dataset. AI is very good at picking up things it has been trained to pick up, including euphenisms. Then you just push the flagged posts to human mod team. (It is also funny to call it AI, when we're actually talking about LLMs, not intelligence.) |
|||
|
|||
|
Fluwm
Posts: 947 |
|
||
I mean, that's kind of my point -- AI is very good *only* at picking up what it's trained to pick up, and humans are very good at inventing *new* euphemisms and dogwhistles, so for the AI to work, it would require constantly updating it and adjusting it, which kind of defeats to purpose of automating the process in the first place. And, as NeverConvex mentioned, there's also an (increased) risk of false-positives, who would inevitably be caught in the crossfire -- something that would give general members of these communities that much more friction. To jump back to your comment on semantics, this is something where I think a *real* AI would be enormously useful, but I'm not convinced LLMs would necessarily improve things. |
|||
|
|||
Glordit
 Posts: 606 |
|
||
|
I help out on a website which has a comment section, it has an auto-mod system, which is basically useless.
The most random posts will get marked as spam for example, "Nice show!" or "Good episode". It will at random flag accounts for no particular reason as well. It's only useful for catching the most obvious offenders. You also can't go and flag words because someone used it as euphemism for something else, due to it then flagging everything with that word in it, regardless of context. As others have said, people are extremely crafty and will easily find ways around it or to fool it. Don't except the community to chip in or be very helpful. You set the bar too low, and people work out that they can spam report posts that simply don't agree with. Make it too high and you need people to actually care in order for it to get flagged, which doesn't really happen much. |
|||
|
|||
The_Daytona_500
 Posts: 98 |
|
||
|
Having viewed CR comments a few times, 95% of comments were the cringiest stuff you could imagine so I'm not too bummed about them being gone. What sucks is the other 5% where people would comment that the subs are from the wrong show or missing that was useful for not wasting my time (was especially great in winter 2023 when CR screwed up at least 4 times).
|
|||
|
|||
MarshalBanana
 Posts: 5467 |
|
||
|
The last time I used the comment section, it was full of bots. Was that an issue that ever got fixed?
For many years now, I've only watched CR on streaming devices like Apple TV and Amazon Firestick, so i thankfully never see any comments. |
|||
|
|||
DerekL1963
 Subscriber Posts: 1119 Location: Puget Sound |
|
||
|
My guess? 99% of the Very Online people mourning the loss of the comment section have never actually read the comment section, let alone posted to them. They've been cesspools for years now. Just like forums, Crunchy didn't shut down the thriving community folks seem to believe them to have been... They were barren, abandoned, wastelands.
|
|||
|
|||
kgw
 Posts: 1137 Location: Spain, EU |
|
||
|
Not everywhere in the word have the same idea as what is "freedom of speech" as in the US, which, true to be told, feels a bit strange at this side of the "pond". Maybe because we have seen what "free speech" can hide.
|
|||
|
|||
harminia
 Posts: 2027 Location: australia |
|
||
|
I pretty much never looked at the comments on episodes so the emotional reactions people are having to this is lost on me.
I do remember looking at the comments on news articles sometimes thinking there might be some interesting comments but it was always scam bots. I think some combination of human checking and AI/LLM moderation would probably be the best way to handle it but I think it would still be a big job. I don't know what the infrastructure was for seeing new comments as they appear but it's possible they'd have to build a whole new system to really get on top of it, and we all know CR doesn't want to put effort and money into things that they feel won't help them reach their goals. |
|||
|
|||
|
nyaa
Posts: 134 |
|
||
|
The place I hang out at most of the time has a lot of different releases and doesn't have a forum per se but there is a place where one can leave comments under each release and a lot of the time it can get pretty nasty. I've gotten into some bad fire fights but I'd miss it terribly if the comments sections were removed,
|
|||
|
|||
 |
All times are GMT - 5 Hours |
|
|
|
Powered by phpBB © 2001, 2005 phpBB Group