Most simple DAU metrics have been compromised. Filtering them is not easy because sybil users are still users, there are users who use blockchain very few times just for simple fund transfers, and there are also degens who are active on blockchain all day long.

Nevertheless, it seems like we could try some interesting filtering. Simple filters based on a specific time period would be (1) "# of txs (count *)" and (2) "# of interactive addresses (count distinct to)".

While these numbers could be applied differently depending on the characteristics of each protocol, they could provide interesting interpretations for existing DAU and Retention analysis