Thread 5538

This is an idea I've been thinking about for a while. I don't know if it's a good idea or not, and I can see why the suggestion could be alarming, but I think it's worth talking about at least.

Whenever all of the janitors on a board are offline, shitposters have free reign to derail threads with roll images and "^- This poster is..." images, and CP tends to stay up longer than it should because it takes 3 illegal reports for anyone else to be able to deal with it. The point has been made that, if you just add an image MD5 to a banlist and prevent people from posting it, they'll change one pixel in Paint and post it anyway, which makes the process of blocking the image in the first place a waste of time.

My suggestion for a workaround would be something which automatically bans these posts after a short time.

Before I continue I just want to make something clear: it would be a big problem if this were to look for post content (like regex-ing "if you don't reply to this post..." which would run the risk of catching good posts which just happened to contain those phrases. In my mind, it wouldn't even be for ban-evaders like Barneyfag who routinely post the same images, because someone else could post those images without being Barneyfag, and the images themselves aren't rule-violating. It would be exclusively for things like roll images or CP, which are, in and of themselves, always rule-violating. After posting images on the auto-ban list, a short timer would run down, and the poster would be banned. It simulates good janitoring and makes sure these posts always get dealt with quickly, without giving the poster any indication that a janitor isn't actually active, or giving them a reason to modify the image in Paint.

This would improve the quality of the boards by quickly getting rid of the big derailing posts which never fail in accumulating hundreds of replies, which human janitors can't hope to do with the same speed and regularity.

How would it identify these posts then? If it's only supposed to detect the picture, it needs something to compare it to or some sort of filter. You'd have to manually add all of said pictures to the filter and sooner or later people will catch on then change one or two pixels like you said to get by it.


While I really like the idea, I'm still not sure how it would be implemented. Also not sure if having an automated system like that that deletes the post AFTER it's posted is a good idea.

We already have image filters. The more things added to them, the slower posting is due to more things being checked.

Changing a pixel removes the block.

The filter is used for CP of course. Blocking the problem images before posting is better than letting them sit for a time and then getting them. I don't know why would need to simulate janitoring for this.

While it would be nice to have bots and programs that can do our work for us, I've been told it's too intensive for our current system previously, at least for "minor" things like the non-illegal reposts.

I've wanted this before, but for text. An auto-janitor would make it easier to go to sleep even when someone keeps posting the same stupid shit from their phone at idiotic hours of the morning.

One of the problems with a fake auto-janitor is that it makes it look like a janitor is online, which might be useful for a few reasons, but it's going to make things look extremely stupid if the bot continues to knock out roll posts in individual threads while a porn thread stays up on the front page.

>>5540
>Blocking the problem images before posting is better than letting them sit for a time and then getting them.
Informing people that they have been filtered is an invitation to change a pixel. People don't resort to that kind of behavior if they are being suppressed by a person.

The hard part is making sure that people don't ever figure out that there is no man behind the curtain, either by identifying the pattern of deletions or simply due to an information leak, because after that the game returns back to the way it was before.

>>5541
I want to add that I'm speaking here only in the context of rule violations other than CP. That should obviously never make it onto the site.

Unless the image is specifically illegal there really should not be any kind of auto-banning happening. Besides, most bots already modify their image to spam repeatedly.

The boards have survived this long with random off-hour shitposting, it's not going to burn down in the few hours you're asleep. If there's a particular spammer you can always ask RA to set up a custom filter or a temporary rangeban if its easter-tier.

There are some things that probably could be added to the spam filter that would make all our lives much easier, but I don't know how much you can just pile onto it for frivolous auto-assist purposes rather than conserve resources for any more illegal or ad spam shit that comes in the future.

Board-specific block/spam filters would be another pie in the sky idea, to deal with local problems without putting on needless site wide burdens, but I don't know if that is a feasible option or not dev-wise. I remember, a while back, there were attempts to auto-filter block some basic shitposting/flaming tactics on /v/ so I think it is possible, although I'm not implying anything remotely as heavy-handed as that.

If feasible, such a thing would be appreciated in combating the cunnyposting bot on /tv/.

>>6230
If you waited two more days you could have bumped it exactly a year later.