Farcaster

Top interactors (by number of trending casts interacted with):
(trending in this cases means in the trending feed)

1. ndyy (FID 884066) — 168 casts (Likes: 168, Recasts: 0, Replies: 1)
2. vasiliyzibrov (FID 830668) — 136 casts (Likes: 136, Recasts: 0, Replies: 0)
3. noicebot (FID 1057944) — 97 casts (Likes: 0, Recasts: 0, Replies: 119)
4. pugson (FID 557) — 88 casts (Likes: 84, Recasts: 7, Replies: 22)
5. ozgurper (FID 877655) — 85 casts (Likes: 85, Recasts: 12, Replies: 5)
6. brixbounty (FID 18723) — 84 casts (Likes: 83, Recasts: 2, Replies: 11)
7. verge (FID 432754) — 82 casts (Likes: 82, Recasts: 0, Replies: 0)
8. jachian (FID 5494) — 80 casts (Likes: 80, Recasts: 3, Replies: 24)
9. nounishprof (FID 4167) — 80 casts (Likes: 78, Recasts: 5, Replies: 14)
10. yth (FID 326487) — 71 casts (Likes: 71, Recasts: 0, Replies: 1)
11. dylsteck.eth (FID 616) — 68 casts (Likes: 68, Recasts: 10, Replies: 8)
12. juliettemeon.eth (FID 685132) — 65 casts (Likes: 65, Recasts: 5, Replies: 2)
13. huy199506 (FID 374099) — 62 casts (Likes: 59, Recasts: 10, Replies: 13)
14. zinger (FID 1407) — 62 casts (Likes: 60, Recasts: 11, Replies: 15)
15. koolkheart.eth (FID 310531) — 60 casts (Likes: 59, Recasts: 3, Replies: 28)
16. samantha (FID 1606) — 60 casts (Likes: 59, Recasts: 4, Replies: 14)
17. ashmoney.eth (FID 12938) — 59 casts (Likes: 59, Recasts: 9, Replies: 5)
18. jagdeesh (FID 17491) — 59 casts (Likes: 59, Recasts: 1, Replies: 0)
19. ciberdelia (FID 422910) — 58 casts (Likes: 55, Recasts: 12, Replies: 49)
20. qt (FID 18570) — 56 casts (Likes: 56, Recasts: 0, Replies: 11)
21. akash2539 (FID 645888) — 55 casts (Likes: 55, Recasts: 4, Replies: 0)
22. robin (FID 369) — 55 casts (Likes: 55, Recasts: 19, Replies: 7)
23. mikadoe.eth (FID 477126) — 54 casts (Likes: 54, Recasts: 12, Replies: 7)
24. linda (FID 12) — 54 casts (Likes: 52, Recasts: 23, Replies: 16)
25. bertwurst.eth (FID 864405) — 51 casts (Likes: 49, Recasts: 10, Replies: 16)

So I've asked ChatGPT to analyse the data from the Rewards Leaderboard Top 10 I shared yesterday (thanks for the tip @dosir). This is what it came up with

Should we do this again next week?

📊 Key Insights
- High concentration of influence and interaction among top leaderboard creators and their superfans — with some appearing as top fans for 5+ creators.

- A potentially self-reinforcing system — those at the top engage with one another and remain highly visible.

- New reward logic could reduce “circular value exchange” — ideally diversifying who earns and who gets seen.

- Some users (like monadver, ejire5, gfam) may be acting like “engagement miners”, systematically liking/commenting on top creators.

✅ Suggestions (If You're Evaluating the System)
- Introduce diminishing returns on repeated engagement from the same user to the same creator.

- Reward diversity of fans — e.g. reward creators who attract engagement from new or varied FIDs.

- Visibility of top fans could itself become a reward metric, encouraging healthy discovery vs mutual reinforcement.

[artist and dev] <>building things. chilling in tub often. doing things I like</>

day job pays the bills, night job builds the future

Hamster

imo you should compare the amount of likes people get from people in their “inner circle” vs likes they get from others, at stack overflow one metric we used for voting rings was “these people are only ever liking each others content, not anyone else’s”

I mean it seems obvious the inner circle and repeat engagement from top accounts to each other has been driving this since the start hence literally the same users still in it paired with some users being able to single-handedly force trending

i dabble with ai using open identity data from crypto

prev startups & racecars.

building @ network school

Degen

noooo haha don’t read this about the top accounts please try the actual data and look at the clusters — top users interact with each other a lot but also with TONS of other people, proper “voting rings” only interact with each other

outside users also interact with a lot kinda nullifies that justification?

I mean I guess we just need the mean of the top 50 vs 100-150 slots for outbound to determine that

From what I’ve seen casually most of the reply guys that prop the network up don’t get a ton of interaction like okay I’m going to go back to saying it’s easier to reach the guy who was temp ceo of OpenAI on X than a high ego farcaster user 

Anyway you’re right I shouldn’t think about this gonna stop now

I’ll at least research it but why not obfuscate outside the ring + my argument isn’t that they are in like rings just that they like each other a lot and the self-reinforcement is so high (single like trending) that it’s not something to be defeated by a normal user without satisfying said people

Shaw for example trended because they want him here not because they are interested in the content which they don’t otherwise engage with

i dare u to do a deep research on how stack overflow and metafilter and something awful tackled this problem and do the same analysis on fc data

I think its pretty safe to always assume the easiest approach is what farcaster/merkle did to implement this