Content
@
0 reply
20 recasts
20 reactions
Dan Romero
@dwr.eth
Why don't you clean up "the bots"? Useful to be precise here: "bots" is a catch all term that people use to describe low effort, uninteresting or otherwise off-topic content. However, a "bot" is traditionally thought of as a programmatic account, i.e. controlled by a script or software. e.g. bot net When people on Farcaster refer to "bots" they are more often referring to actual people using a mobile device (with the aid of translation tools and ChatGPT) who post uninteresting content (from their pov). They are casting and replying in hopes of appearing to be a "real account" because they have learned that turns into a future economic reward in crypto. Recent tipping meta has increased the expected value of these accounts. It's possible to algorithmically label these accounts, but then you create a "shadowban" meta. So you massively increase the number of support messages and complaints. This was the state of the world before priority mode.
84 replies
126 recasts
832 reactions
Dan Romero
@dwr.eth
We're going to keep iterating here, but our pov is: 1. Priority mode works pretty well out of the box 2. We need to offer ways for actual interesting accounts to break through 3. We'll likely bring back some version of spam labeling since priority mode has created it's own set of customer support complaints. "Why don't you just use onchain signals?" a. Based on our waitlist last year of 300K sign ups, "bots" have plenty of Ethereum addresses with assets. b. Most new users don't want to connect an address during sign up (friction, doesn't work well on mobile). So you'll miss out of on good accounts. c. Active onchain presence != you regularly post interesting content. There are some people of course, but plenty of examples otherwise.
30 replies
5 recasts
306 reactions
CryptoPlaza🎩Ⓜ️💜 OG449
@especulacion
"I believe it is a huge challenge, which is why I think it would be interesting to have a sandbox or a marketplace of potential algorithms that the community could propose. Even a hackathon with this objective would be interesting. This would allow us to move from debating to comparing
3 replies
1 recast
1 reaction
CryptoPlaza🎩Ⓜ️💜 OG449
@especulacion
Indeed, I don't believe that a single algorithm can truly provide the solution; it is likely to be a set of algorithms, each optimized according to different criteria.
0 reply
0 recast
0 reaction
