Understanding number diffs
Differences in numbers across clients and tools
When looking at numbers across clients e.g. Merkle’s Farcaster client vs Recaster vs Coinbase Wallet, you might notice differences in numbers. Specifically, number of followers, reactions, replies, etc. There can be a few different reasons for this.
1. Clients optionally apply filters on top of protocol data
Farcaster protocol remains open to all. However, this means that the protocol has a lot of spam activity. Similar to how most email is spam and you never see it in your inbox, a lot of Farcaster protocol activity is spam and clients don’t show it.
2. Some show raw numbers
If a client is showing raw protocol data, those numbers won’t match another client that is applying filters on top of raw data. Raw protocol data always has higher numbers.
3. Spam filters are not standardized
Each clients filters as they see fit. Clients that use Neynar can use Neynar’s filtering mechanisms will match each other.
Filter spam using Neynar
When using Neynar APIs, you have the choice of fetching raw data or filtered data. APIs return raw data by default, you can filter the data by turning on the experimental flag on any of the API requests.
E.g. the first request passes the flag as false.
while the second one passes it as true.
You can look at the differences in follower counts below in the screenshots.
With experimental flag set to false (raw data)
With experimental flag set to true (filtered data)
As you can see, the raw data shows 273,791 followers while the filtered data shows 146,993 followers - a significant difference due to spam filtering. This is different from what another client might show due to their own filtering logic. E.g. Merkle’s Farcaster client shows it as 145k which is close to Neynar’s 146k number but not exact. This is because each company uses their own spam filtering logic.