Of all the social media platforms I've seen that use some sort of tag system, very few avoid this massive and obvious problem: variant tags.

Bandcamp, Youtube, and Devrant are some examples that come to mind of platforms that have just fallen flat on their face in this regard. For example, I've seen Bandcamp albums tagged "music, game music, game, video game, video game music, rpg, rpg music, role playing game, soundtrack, fantasy, fantasy music ..." you get the picture. I've seen Devrant posts tagged obscure things like "browser vendors suck", which is obviously not something someone would know to search for.

One possible countermeasure would be to either limit the number of tags on an item or have the weight of each tag be inversely proportional to how many you put (might need nonlinear scaling so they wouldn't be incentivized to only put one tag), to encourage users to decide on the best variation and use only that one. But this isn't a perfect solution, because different users could still reach different ideas of what the canonical variation should be.

dev.to allows arbitrary tags, but has a lot of the most important ones predefined, and when typing a tag you get suggestions of popular ones that start with what you've typed. This makes sure users will use whatever's become the de facto canonical version of a tag unless there just isn't one for what you want. dev.to also doesn't allow spaces, capital letters, or anything else other than alphanumeric characters, which greatly reduces the possible variations of a tag. Due to these countermeasures, even though it allows arbitrary tags, I've yet to see a dev.to post abusing the feature.



This page was last modified 2020 May 28, Thursday, 14:31 (UTC)