Regex filters and whitelist rules
Given the widespread use of hostile bots on Twitch right now by bad actors, particularly those targeting LGBTQIA+ and BIPOC communities, it would be great if we could have a RegEx pattern block/ban/follow restriction option. A number of the accounts being spun up for the purpose of harassing and stealing the identities of streamers have consistent name patterns, and being able to issue a regex block/ban on those patterns would be useful.
This sort of feature would obviously be something that streamers would have to use at their own risk. Overly inclusive block/ban patterns could potentially limit legitimate viewers. But for those of us who are more tech savvy, this could be a massive time-saver.
On the same broad thrust, the ability to add users to a whitelist so as not to unfairly impact existing supporters of a channel, and either a temporal or behavioural filter which automatically adds users to a channel's whitelist would be useful. This could be a more effective approach than the current "followers only" or "subs only" chat options. More fine-grained control on the white-listing process would be appreciated.
For example:
Whitelist on account age
Whitelist on x minutes/hours/days follow age
Whitelist on x minutes/hours/days without mod action in <max> other channels
Whitelist on x minutes/hours/days without automod action in current channel
Whitelist on x minutes/hours/days without automod action in other channels
This could perhaps be branded as an "advanced mod tools" option which can be unlocked by the streamer at the account level so those comfortable with the existing system could use that, but those needing more nuanced tools could have such systems made available on request.
StoryGirl
shared this idea
-
posstoma
commented
I don't like racists either, but I also hate people who are happy to hate others. I don't like the act of continuing to act or act even though people dislike it. I thought it was a great proposal and would like to agree with it.
-
志摩あんず
commented
This is such a brilliant idea!
Japanese users have been trolled by just one malicious user for more than six months with very obscene language consisting of romaji and Japanese.
We can protect ourselves if this feature is implemented.
This feature needs to work for both IDs and usernames (display name). Spammers use direct words in their usernames that Japanese people would dislike at first sight.
I think it needs to be addressed as soon as possible.
Thank you in advance. -
zoumotu3
commented
This is such a brilliant idea!
Japanese users have been trolled by just one malicious user for more than six months with very obscene language consisting of romaji and Japanese.
We can protect ourselves if this feature is implemented.
This feature needs to work for both IDs and usernames (display name). Spammers use direct words in their usernames that Japanese people would dislike at first sight.Also the opinion of liliths_fane in the comments section is very nice.
There needs to be control over who can follow, raid or host. -
AnarchistLaboratorySean
commented
It's absolutely abhorrent how many BIPOC and Queer streamers have been harassed off the platform because YOU refuse to address it, twitch. YOU refuse to implement tools or even listen to anyone that comes to you about the problem. This is ridiculous; get your **** together
-
FaeAlicia
commented
I would also like to demand an overhaul of Automod to, for example, not block us trans people for literally just mentioning "trans" or for associating completely irrelevant terms to "sexism" or some ****.
I literally have to disable Automod and use third party tools for this like Nightbot simply because Automod would force me to add so many terms to the allow list, with the allow list not even working half the time.
Beyond that, there's a lot that can be done to make the experience more enjoyable.
- Give us the ability to disable stream chat for non-moderators after X minutes after our streams end
- Fix the appeals system so that it doesn't take longer for the appeal to go through than the fake suspensions we get from the hate raids reporting our channels for the hate raids existing in our channels. Appeals should take less than 48 hours no matter what.Among many other things.
-
liliths_fane
commented
All of this should be implemented.
Additionally, control over who can follow, raid or host needs to be made granular as well. With as many options as possible. I'm talking account age, whether or not they stream, how recently they've streamed, and these should have their own whitelists. For instance, I might whitelist everyone who's been following me for a month already, but I wouldn't whitelist all of them to host me.
Similarly, it would be great if we didn't need to use outside tools to do what twitch could do internally just as well. I could whitelist creators, for instance, and automatically anyone they banned would be banned on my channel. A simple tool to help keep communities safer with creators we trust.
The regex feature is a good idea, and it should be implemented to give users as much control as possible. But, they could have a very simple slider that prevents interactions within a certain timeframe (also modular) from any accounts with too many similar characters. They aren't clever. Today I had a guy go from 1 to 4 before switching anything up and adding anything else to it. This wouldn't be perfect, and it would only serve to force them to be more clever, but at the very least it would prevent troll1 to troll23 from following you after you ban troll. Anything that makes their job harder.
Lastly, they need to get in contact with people in our communities already running large shared global ban lists so that they can implement controls to prevent those accounts from cropping up at all.
-
TheClosetForge
commented
This would be extremely useful! Yes please!
-
ancomlibra
commented
Superb idea. Can’t believe this isn’t already in place! Twitch, get on this ASAP.
-
shyredfox
commented
This is an outstanding idea! Please implement it!
-
Maverynthia
commented
Why isn't this a thing already.