bradq
@bradq
Suggestion To Improve Rewards Algo *** Build Up Farcaster Upper-Middle Class Not a complaint--true suggestion. We are seeing casts about 'points freezing.' I think there's an issue. (more in thread) But I think there's a connected & bigger issue: The algo doesn't love the upper-middle class of FCers Algo rightly weights OGs/auto-follows (Maybe too much, but np) Algo also is great for newcomers as we've seen 10-12 times in the top10 (Love it!) BUT, a lot of great FCers are in the 2K-15K & not OG/auto-follow. They are, however, some of the most thoughtful, discerning, engaging FCers. You can see the algo doesn't love or often like them. I can get numerous r/c, comments, likes from them & get no points from them on a 'good' cast. On others I can get similar Middle class action & get a couple interactions from OG/AF types & get numerous points all from the OG/AFs. I love Rewards & benefit from it. Think this would improve it. 👊 ( @dwr.eth @v ) see freeze examples below
9 replies
1 recast
13 reactions
Dan Romero
@dwr.eth
Thanks for the feedback. Varun already shared the stuck score issue which we can iterate on. 1. What does it mean the "algo doesn't love"? They aren't in the top 500, i.e. scoring, or they are just further down the scoreboard? 2. Do you have a list of users you think are being penalized (explicitly or implicitly)? 3. Do you have examples of casts?
5 replies
0 recast
1 reaction
bradq
@bradq
Turns out the cast we are talking on ends up being a great example---0 points. https://warpcast.com/bradq/0x88967ca8
0 reply
0 recast
0 reaction
