Feature idea; check post keywords and suggest 'liking' instead

The Problem

The forum is currently showing large quantities of flagged posts simply because they're compliments or similar. While this may give the impression that the forum as a whole is against letting someone know their contribution is good, this is not the case. The only issue is that when users do this frequently, a topic becomes hard to read being cluttered with rather useless replies. For this reason, it is against the guidelines to post anything that boils down to 'nice'. This results in posts all over threads being automatically hidden due to a large number of flags, even leading to temporarily locked topics. All of that is unnecessary; the 'like' system exists and should be used instead of replies.

A Possible Solution

I've noticed that the forum already does some type of checking for keywords to see if your post matches any existing ones. To solve the issue above, I'd suggest making this system work with posts that could be summarized with the 'like' feature. When a user is setting up a new reply and it matches a few keywords like 'nice' or 'great job', the user would be alerted. A user would be encouraged to refrain from posting it and instead just hit the like button. This should drastically reduce the amount of posts cluttering threads, and as a result this wasteland of hidden posts as a whole.


This feature suggestion assumes discourse allows the keyword system to be used this way; if anyone can confirm or deny this I'd like to hear it as well.

Definitely agree. If this turns out to be too difficult to implement, I’d like to see a PSA sent to every Member and New Member instead. Empty “like” replies are becoming a problem (due to the influx of new members and the increase in posts), and I’m getting tired of explaining why it’s a problem.

I think instead of asking the user to refrain from posting, a better suggestion would be to ask them to make sure they’re giving meaningful feedback, such as explaining specifically what they like and why, instead of them simply saying ‘good job’ or ‘looks cool’. If they don’t have anything specific in mind, they should instead be told to use the like button.


CC @Lilly_S

With a plugin, yes.


Seems like a simple enough idea, yet it fits pretty well. I know GMail does a similar thing where if it detects you used the term “attached” but didn’t add an attachment to your email, it reminds you before you send it out that you may have intended to attach something there.

I think the biggest challenges here, since it is possible to add-on to Discourse from what I’ve heard, would be setting up what exactly determines needing the reminder. How much of the post needs to be the “Nice!”, so that there’s no false flags?

There’d also be the harvesting of terms to use, but considering they seem to be coming in more often as of recent, I don’t think there’d be trouble gathering a list of keywords.

1 Like

The idea is that this method is rather unobtrusive; while it does have the chance of popping up when it does not necessarily have to, when it pops up the user is simply given a reminder. I doubt there’s much harm done in the case of false positives.

1 Like

That is a good point there, guess human judgement would be able to handle everything that isn’t a generic “Nice!”, same as the “There may be a similar thread!” reminders when making a new thread with a similar topic.

I know that for the first two posts you’re automatically given a prompt on the right side (where your post preview is,) but maybe it needs to stick around a little bit longer, maybe for your first five to ten posts on the devforum…? I can also see new users seeing the “big orange warning box” and getting a little bit scared. Maybe there’s some way to slim down its contents to get across the gist of what it’s for, but in a less verbose manner? I don’t even remember its contents now, actually… all I know is that it was top-to-bottom filled with text.

Additionally, it should allow for compliments, but only where there is substance to the body of the post. That might be harder to detect, a kludgy solution might be to go “hey, maybe you should add more to this” if it’s under, say, 30 characters in length.

1 Like