They changed the algorithm on Dec 7, 2016.
Before: https://web.archive.org/web/20161206000344/https://www.reddit.com/
After: https://web.archive.org/web/20161208000333/https://www.reddit.com/
Best guess, based off the vote count differences of those two days, is that if you see a vote with 100,000 count, it’s likely closer to 20,000 count (divide by roughly 5 or so).
There was a post about it by KeyserSosa on Dec 6, so it’s not some conspiracy or anything.
I figured YSK.
“Completely” made up is a bit hyperbole. They’re not random, they are usually in some way indicative of the number of upvotes/downvotes a post is getting.
The reality is that any naive upvote system can be gamed. The more popular Lemmy gets, the more lucrative it will be to systematically manipulate which posts/comments are promoted and which get buried. And a naive voting system that just reflects the raw number of up/down votes is trivial to manipulate. It’s harder than a signal in the noise issue, it’s more of a signal in a deliberately crafted manipulated signal issue.
If Lemmy gets more popular, it’s only a matter of time before it’s forced to come up with alternative methods for deriving post sentiment.