Conversation

Replying to and
I used an "author weight" = 1+1/(num_posts) where only posts with at least one internal link count. So people with fewer posts get bumped up. I'm obviously lowest author weight at 1.0037 and is next lowest at 1.03. People with just 1 post have author rank 2
1
3
I also divided by the age of the post (so a linear reciprocal discounting rather than exponential). A good post should accumulate more backlinks over time, right?
2
4
I'm kinda pleased that the top is NOT dominated by me via this reasonable ranking function. I think I need to tweak more, and use an age discount function that also discounts too-young articles maybe 🤔... or is even 1 link a good enough proxy for age?
1
1
Hmm... would be nice to add a knob that recovers the raw link-count at an extreme. Maybe an (1+lambda) exponent on the raw link count
1
1
Also, since the archives have just about 715 posts, almost exactly half of the posts have at least 1 link, and the rest are orphans. This is fascinating, a full 50% of my writing goes literally nowhere
2
7
If I stick with this serial thread-blogging model I'm starting now, then the ranking algorithm post 2019 would need to be different, a sort of weighting by length of chain or something.
1
3
Hmm... a simple way to add that to the formula would be to simply multiply by the "series depth" of a post. So a part-IV post with 1 link would end up with a raw weight of 4.
1
1
Show replies