14

Ignore obnoxious comment signatures, make dogwhistles more transparent, remind yourself not to interact with certain topics, uncensor common celf-censorship euphemisms, or just apply some good 'ol censorship of your own. What I am imagining is a system that allows the user to set custom regular expression rules that get applied to all comments. There could be both global rules (that apply to every single comment) and user rules (that apply to all comments from just that user, similar to custom user labels).

What are your thoughts on having a feature like this in whatever app you use? Would you even find it useful at all? If the change is only visible to the user, is there any abuse potential?

Pictured here is a proof of concept showing Thunder with a simple "Cloud To Butt" function applied.

top 8 comments
sorted by: hot top controversial new old
[-] olorin99@kbin.earth 1 points 12 hours ago

A feature that should be able to handle this will be in the next version of Interstellar. The feature proposal is here if you have any extra ideas/comments.

[-] wizardbeard@lemmy.dbzer0.com 3 points 1 day ago

The abuse potential would be that I think it opens the door to more easily create convincing doctored screenshots of comments and interactions. It's technically possible already, but I don't think we've really seen it on lemmy yet. There's ways around that, like formatting replaced text differently or marking those comments with a clear icon to indicate a regex replacement was used on them, but I still don't particularly like the idea.

I'm thinking that it would also make it easy for people to have confusing interactions, if somehow a regex ends up changing the meaning of a statement or something, and the user forgets they have it set up. I guess the "big icon indicating regex was applied" idea would work there too though. Maybe the icon could even be a button to toggle it on/off per comment?

The common use case I see for regex features on social media apps and addons is for hiding entire comments that match one of your patterns, and I generally think that's the best choice for implementation.

That said, with a clear indicator it had been applied it probably wouldn't be awful.

[-] admiralpatrick@lemmy.world 4 points 1 day ago* (last edited 1 day ago)

But yeah, your other points are valid. Anything replaced needs to be indicated / highlighted / otherwise marked as not the original text.

[-] admiralpatrick@lemmy.world 2 points 1 day ago* (last edited 1 day ago)

I think I could pretty easily shim that into the filter policies in Tesseract (that's the big project for this upcoming release). I already have customizeable regex filter lists for keywords, but those only set the "filtered" flag if they're detected and return the list of regex patterns that were matched.

Since the filter policies can be either global or per community group, you could get pretty granular with it. I'll look into it and see if it's feasible to shim in.

I jokingly thought about adding something like that already.

Basically:

(^Paywall(ed)?$) -> I'm going to contribute nothing to the discussion except whine that you didn't cater to my laziness and make my laziness everyone else's problem until someone else replies with an archive link I could have easily gotten myself

I thought the feature was too absurd to flesh out, but if there's interest, I'm open to adding it lol.

[-] SatyrSack@quokk.au 1 points 1 day ago* (last edited 1 day ago)

I jokingly thought about adding something like that already.

Basically:

(^Paywall(ed)?)$ -> I'm going to contribute nothing to the discussion except whine that you didn't cater to my laziness and make my laziness everyone else's problem until someone else replies with an archive link I could have easily gotten myself

This is very close to what I had in mind when saying "remind yourself not to interact with certain topics". That any time a comment contains a certain keyword/phrase, it could replace the entire comment with "I am a person not worth talking to".

What I have in mind would be a simple pair of textboxes for each filter policy: one for the regex pattern to search for, and one for the string to replace it with. Then a checkbox or something to toggle whether just the matched string gets replaced or the entire comment gets replaced when the pattern matches anywhere in the comment. Then, the user can add as many policies as they like.

[-] admiralpatrick@lemmy.world 2 points 1 day ago* (last edited 1 day ago)

What it does now (well, in the dev version), is evaluate the whole comment for various rules you can set. Here, I added a keyword regex filter (words|phrases) and applied it to the "Lemmy Apps" group.

This is how your comment shows up when it hit on those keywords:

When content is filtered, I don't even let it show the creator (because sometimes you just see a name and you know it's gonna be something braindead lol).

So I feel like that takes care of the "entire comment" portion of your idea (or at least that's as far as I'd want to implement it). If the comment hit on any other policy rules, those would also show up in the list you see.

For the replacement text, I'm taking WizardBeard's advice (below) and would somehow visibly identify that the original text has been modified by a user preference. Not sure how, exactly, I would do that. Maybe strike-through the original text and insert the replacement string italicized, in a different/larger font, or something like that.

~~Good morning~~ Hello.

~~What's up?~~ I am not a person worth talking to.

The only complication would be that the indicator method, whatever it may be, has to be valid markdown and, thus, could theoretically be something the original person could have said.

Filtering policy editor looks like this (this is scoped to a group, so the "Communities" tab at the top is hidden):

[-] SatyrSack@quokk.au 1 points 1 day ago

As for indicator, I think what I would prefer as a user is a little icon next to the edited indicator, and then the ability to view comment source (which is already a feature on many clients). Though that is not really addressing the problem very well, as people viewing the screenshot need to understand what the icon means.

[-] admiralpatrick@lemmy.world 2 points 1 day ago* (last edited 1 day ago)

I've got an indicator row between above the comment text and the username (shows things like "deleted by creator", "banned from community", etc"). I suppose that could indicate whether the item was altered by a user filter.

But now I'm having second thoughts about this whole idea. Originally, I felt validated because I had kicked the idea around already, at least as a joke. But now I'm remembering YPTB, Fedi-drama, and similar communities exist and there are frequent flyers there who have nothing better to do than present things out of context and stir shit up and how this would be like giving them a loaded weapon.

Lemme (re) think on this.

this post was submitted on 06 Sep 2025
14 points (100.0% liked)

Lemmy Apps

6683 readers
8 users here now

A home for discussion of Lemmy apps and tools for all platforms.

RULES:


An extensive list of Lemmy apps is available here:

LemmyApps.com

or lemmyapps.netlify.app


Visit our partner Communities!

Lemmy Plugins and Userscripts is a great place to enhance the Lemmy browsing experience. !plugins@sh.itjust.works

Lemmy Integrations is a community about all integrations with the lemmy API. Bots, Scripts, New Apps, etc. !lemmy_integrations@lemmy.dbzer0.com

Lemmy Bots and Tools is a place to discuss and show off bots, tools, front ends, etc. you’re making that relate to lemmy. !lemmy_dev@programming.dev

Lemmy App Development is a place for Lemmy builders to chat about building apps, clients, tools and bots for the Lemmy platform. !lemmydev@lemm.ee

founded 2 years ago
MODERATORS