C O M P Ξ Z on Warpcast

C O M P Ξ Z pfp

🧵 The Necessary Transparency on User Labeling in FC and Improvement Suggestions 🧵/10 - New Update for Jan, 30, 2025

4 replies

8 recasts

18 reactions

C O M P Ξ Z pfp

1/🧵 Introduction Farcaster is a growing decentralized platform aiming to build a high-quality community. However, recent statistical analysis reveals some concerns: - Only 9.03% of users are categorized as "Unlikely to engage in spammy behavior." - 50.09% are "Likely to engage in spammy behavior." - 40.88% "Might engage in spammy behavior." These stats reveal a significant challenge in distinguishing genuine users from potential spammers. Do we really only have 9.03% of users who are purposeful?! If so, we will need to create more guides for commcunity.

1 reply

2 recasts

7 reactions

C O M P Ξ Z pfp

2/🧵 Current User Base As of now, Farcaster has released 484,688 users label as open source and this resource has become attractive to project owners! But, with the current classification system, distinguishing between valuable users and spammers is far from perfect. It doesn't seem logical to use this data for filtering users in certain projects like $degen and others. It's better to use custom algorithms for this purpose. - cc @jacek

1 reply

1 recast

2 reactions

C O M P Ξ Z pfp

3/🧵 Analysis of Existing Issues Probabilistic, Not Absolute: The current classifications are probabilistic, not definitive. They’re evolving benchmarks based on Farcaster’s official data and still under testing. These labels guide development but aren’t final judgments on user behavior. So it cannot definitively confirm the quality of an account for now.

1 reply

1 recast

2 reactions

C O M P Ξ Z pfp

4/🧵 Inaccurate Account Evaluation New users often get wrongly tagged as spammers because they haven’t had enough time to engage meaningfully. This can be particularly harmful for users still getting familiar with the platform.

2 replies

0 recast

1 reaction

C O M P Ξ Z pfp

5/🧵 Need for Algorithm Updates The current algorithm may not accurately assess user behavior, leading to wrongful classification of legitimate users as spammers. There’s a pressing need for an update to fix these issues. @dwr.eth

2 replies

0 recast

1 reaction

C O M P Ξ Z pfp

6/🧵 Unsuitable Data for Filtering These statistics shouldn’t be used as a primary filter for projects or products. Many trusted users are mistakenly labeled as spammers, including those involved in legitimate activities like airdrops. We need more refined criteria.

1 reply

0 recast

1 reaction

C O M P Ξ Z pfp

7/🧵 Lack of Transparency in Algorithm Functionality We don’t yet fully understand how the algorithm works. Clearer, more transparent explanations of the criteria used to evaluate users are crucial. This will help the community align their activities with Farcaster’s expectations. cc @v While we have tried to publish some articles to guide the community, everything is really based on professional experience and not on the actual performance of the algorithm. @pichi

2 replies

0 recast

1 reaction

C O M P Ξ Z pfp

Frames should be informative by better UX 8/🧵 Frames that provide information about user account status should be designed to the highest possible level, since the data is open source, so all developers should use better UX for its results. Because part of the design of the frames will be related to education and culture in this regard. @mvr and others great developers.

1 reply

0 recast

1 reaction

C O M P Ξ Z pfp

9/🧵 Recommendations for Project Owners Since these labels are probabilistic, project owners could view them as concept indicators for future, not final decisions. Because the data is still raw and a large part of it has not yet been updated. - cc @jacek Given the evolving nature of the dataset, it’s best to wait for further updates and refinements from FC's team before drawing any conclusions.

1 reply

0 recast

1 reaction

C O M P Ξ Z pfp

10/🧵 Conclusion Farcaster is still evolving, and so are its user evaluations. We must continue refining the algorithm, improving transparency, and ensuring that users are not unfairly labeled based on incomplete data.

0 reply

0 recast

1 reaction