wavebeem

world wide weirdo

ย 

๐Ÿ’• @hauntedlatte

๐Ÿ  portland, or, usa

๐Ÿ“† mid-30s

๐Ÿ’ฌ here to make friends and chat

ย 


ย 

๐ŸŽฎ video games
๐Ÿ‘ฉ๐Ÿปโ€๐Ÿ’ป web development
๐Ÿ‘ฉ๐Ÿปโ€๐Ÿซ teaching others

ย 


ย 

๐ŸŽจ digital aesthetics
๐Ÿ’…๐Ÿป makeup & jewelry
๐Ÿ‘— gothic fashion
๐Ÿ‘ฉ๐Ÿปโ€๐ŸŽจ making pixel art

ย 


ย 

๐Ÿค˜๐Ÿป progressive metal
๐ŸŽธ video game music

ย 


ย 

๐ŸŸข everything green
๐ŸŒŸ neon colors and transparent plastic

ย 


blog + rss
wavebeem.com/
discord
@wavebeem

tef
@tef

Over in C++ land, there's a useful adage:

With a sufficient number of users of an API, it does not matter what you promise in the contract: all observable behaviors of your system will be depended on by somebody.

It's a very similar vibe to "the purpose of a system is what it does", or even "the street finds it's own uses for things."

My favourite example of this is in Golang. To discourage people from relying on the rather accidental order of entries in a hashtable, Golang randomizes the output. Now? People use the hashtable as a source of randomness.

I feel this is worth bearing in mind as people talk about fixing cohost.

For example: A number of people have suggested having "trending tags" or "active tags" as a means of discovery. Listing the most used tags within a given time window. In theory, people will tag their posts as normal, and users will browse the active list to find content outside their feed.

In practice? People will tag their posts to drive views to their pages. People will post more to drive up the tag numbers. It's like sticking a high score table on a game and asking people not to compete. It's like setting up a popularity contest and expecting people not to stuff the ballots.

It doesn't matter how well behaved the userbase is right now, either. Implementing a popularity contest will attract the users who like to game it. A table of active tags will quickly stop being a way to find what people are talking about, and quickly become a global competition for which tags people want to see on the site.

Or if you like, "when a measure becomes a metric, it ceases to be a good measure."

That doesn't mean a "active tags" list is a bad ideaโ€”It just means that implementing it well is a difficult and treacherous task. Features need to be designed around what sort of behaviours they want to encourage, and designed around what behaviours need to be discouraged, too.

You don't need to look any further than twitter to see how features play out in practice. Trending topics devolve into a list of potentially dead celebrities, football scores, and whatever hate campaign mouthpiece posted engagement bait that day.

As much as I want to be able to find good posts on cohost, trending topics is too high of a price to pay.


You must log in to comment.

in reply to @tef's post:

Firmly agreed. I personally think the desire to "find good posts on cohost" is something that should be discouraged. The internet has fostered (if not outright created) a need to find "good content", as if there's some metric and we're all falling short. Maybe I'm "old man shouting at clouds" but I don't think we should be incentivizing folks to consume endlessly.

so what your saying is, my hashtable needs an entropy source that's noisy enough to stop accidental order dependencies and intentional collision attacks, but not so noisy that it's a useful source of randomness for the rest of the program. got it

I don't think "trending tags" is even a good discoverability method. Even aside from the spamming and other bad behavior, it's a quantity-over-quality metric that surfaces the subjects people post about the most rather than the ones with posts people actually want to see

On Twitter, the trending section is basically a dynamic clickbait generator that's constantly trying to lure me into the latest celebrity or politics news. I straight-up adblock it so I don't have to see whatever tempting ragebait headlines it's trying to lure me with