SCOOP: on #WSocial's website, comment counts for posts by prominent users are incorrectly displayed, showing artificially elevated numbers.
An ATproto user came up with an interesting theory for it.
My article: "W Social, Fictional Metrics and the Beauty of Open Data"
🔗 : https://blog.elenarossini.com/w-social-fictional-metrics-and-the-beauty-of-open-data/
#blog #BigTech #EUBigTech #TEP #TrustedEuropeanPlatforms #TrustYourFeed
@_elena I assume this is not similar to the issue we have on the fediverse. Where one server has different (lower) stats as it only sees a subset of the like, boosts, etc compared to the post's host server?
I don't know enough about the AT protocol
@daj no, these are fake numbers. the cards displayed on the landing page are made by their website developer, they aren't the real posts.
You can check on mu.social as well, there is a DISCONNECT only on the number of comments, but the likes and shares check out.
On atproto you don't have this kind of problem
You can easily check and double-check the engagement numbers across the whole network using community-made APIs to check what we call "backlinks"
Or check inside the microblog app itself, if it's set up correctly with the relay and app server, you have a correct number of engagement too (Bluesky's app has the correct number always)
Here is one example, the W Social landing page said it had 85 replies
Elena Rossini 🌈
David - Forking Mad
Victoria