While it's great that the auto-follow list is no more, the damage has already been done in that any (off-the-shelf) influence-driven algorithm will heavily skew to favor accounts that were on that list.. unless proper adjustments have been made

And if you're developing your own proprietary influence-driven algorithm, it's imperative to take this into account cc: @betashop.eth

Some food for thought for the @openrank & @airstack.eth teams..

Specific to OpenRank's Following Rank + Airstack's Far Score: excluding accounts that have never casted (there are 200K+!) from your respective algorithms (when looking to the followers of an account) will help alleviate the skew which heavily favors accounts that were on the auto-follow list

The official account for Airstack, developer of @moxie-protocol. airstack.xyz. On a mission to grow the Farcaster GDP.

Some food for thought for the @openrank & @airstack.eth teams..

Specific to OpenRank's Following Rank + Airstack's Far Score: excluding accounts that have never casted (there are 200K+!) from your respective algorithms (when looking to the followers of an account) will help alleviate the skew which heavily favors accounts that were on the auto-follow list https://warpcast.com/geoffgolberg/0x82ad8678

I like running, poker and oysters | Founder, Social Forensics (SocialForensics.com) | Berkeley & UPenn | Learning to listen

we'll study it, it's an interesting suggestion

some questions i have: do you feel like some casts are earning too much? or that some casts don't earn enough? flattening the curve at the top will just mean less rewards for gaining influential distribution. are you also recommending we add more weightage in the middle? if so, what would be the rationale as to why?

Engagement from these 30 accounts are most valuable re: making it rain Moxie..

For comparison, here are Far Scores (and Far Ranks!) from a sampling of other users to benchmark against the top 30:

- @adrienne (81.87; Far Rank = 103)
- @christin (29.84; Far Rank = 188)
- @proxystudio.eth (10.11; Far Rank = 325)
- @maretus.eth (7.94; Far Rank = 341)
- @mazmhussain (2.02; Far Rank = 559)
- @geoffgolberg (0.83; Far Rank = 911)

What does this mean?

It means engagement from @jessepollak is 357 times more valuable than engagement from @mazmhussain

It means engagement from @linda is 55 times more valuable than engagement from @proxystudio.eth 

There are fewer than 1,000 accounts that have a Far Score greater than 1

Make of it what you will..

Thanks, Jason

It's not that I feel some casts are earning too much; rather, my point is that accounts with Far Scores above 100 effectively determine how Moxie Rewards are distributed across the protocol (there are fewer than 100 of said accounts!)

I am suggesting smoothing out the distribution curve – and yes, that would involve increasing the Far Scores of accounts outside the top ~100 (as well as reducing the Far Scores of the top ~100 accounts)

This can be done in such a fashion that the same amount of Moxie that's being earned daily via Rewards currently doesn't materially change

chef at moxie.xyz | Moxie is your onchain AI assistant—analyzing transactions, unlocking social alpha, and executing trades on your behalf.

Thanks, Jason

It's not that I feel some casts are earning too much; rather, my point is that accounts with Far Scores above 100 effectively determine how Moxie Rewards are distributed across the protocol (there are fewer than 100 of said accounts!)

I am suggesting smoothing out the distribution curve – and yes, that would involve increasing the Far Scores of accounts outside the top ~100 (as well as reducing the Far Scores of the top ~100 accounts)

This can be done in such a fashion that the same amount of Moxie that's being earned daily via Rewards currently doesn't materially change https://warpcast.com/geoffgolberg/0x4781f54f

Not suggesting artificially increasing/reducing Far Scores (I should have worded things more clearly!)

That would (very likely) happen naturally as a result of the suggestion I made, i.e. excluding accounts that have never casted from your algo (when looking to the followers of an account)

You had me listening on the first part about evaluating whether some users are over weighted 

…however you lost me on artificially boosting the scores or other people … that’s like manipulating CPM rates on advertising 

Our goal here is not to distribute as much moxie as possible each day. It’s to reward content that earns attention and distribution 

We’re also intent on not just building a system that gives stuff away for free forever. When 3rd parties come in and buy moxie for rewards, they have to have faith that the system is designed for them to pay for the attention they get