Proposal X: LobsterNFT — fair user impact score (FUIS)
This article will show some data we have from the LobstersDAO DeFi & NFT tg community and how we can measure and do an interpretation of an “impact” from each community member.
Why?
https://t.me/lobsters_chat/265916
So, basically: more contributions/impact members have => more LobsterNFTs must be dropped for them as a reward for their contribution.
Data
We have a table with stat about: messages, active days, digest appearances:
The idea is to make a score based on those data with some modifications to decrease inequality.
FUIS(data) = a*Log(messages) + b*Sqrt(days) + c*Sqrt(digest)
Messages
Let’s look at the sorted chart of messages:
It shows that vast majority will not receive anything based on this stat. They are highly diluted by small set of members who are doing most of communication efforts. Also, many messages are replays/thanks/reactions which is hard to consider as a best contribution here.
If we’ll use Log(messages) the situation will become different and more “fair”, since even small but productive discussions/comments will be countable.
Days
Let’s look at the sorted chart of active days:
This factor is much less inequality than “messages”, but still vast majority in a huge favor here, so here we can propose to use Sqrt(days).
Digest
Since Feb-21 we’re doing a small daily job to highlight some (5–7) mostly discussed topics from the LobsterDAO and publish summaries in a form of LobsterDaily Digest. It’s done by the people experienced in DeFi.
Let’s look at the sorted chart of digest appearances:
Seems to be reasonable to use Sqrt here also.
Fair User Impact Score (FUIS)
FUIS(data) = a*Log(messages) + b*Sqrt(days) + c*Sqrt(digest)
So, the proposal includes having 3 parameters a+b+c = 1, which represents community opinion about what they think is more important. My personal opinion is: a = 0.2, b = 0.5, c = 0.3.
P.S.: all data factors will be normalized by dividing to the MAX number, to keep everything simple.
Best,
Alex P
Twitter: https://twitter.com/AlekseiPupyshev
