Hi @ned thanks for responding.

• the mathematical formulae and implementation mechanisms were already posted in my first article here: https://steemit.com/utopian-io/@scipio/how-to-solve-spam-on-steem-introducing-userauthority
  Instead of Gauss-Jordan row reduction I've used (low-CPU / low-RAM) multi-iterative substitution, which is presented in the spreadsheet / Excel found there as well.

• the implementation plan is as follows:

1. read all follower transactions, chronologically, from the blockchain
2. store follower transactions in a temp MongoDB collection: followDB
3. update followDB per blockchain block
4. include last block id
5. approximate all user UA via multi-iterational substitution, begin with 1 per account, store in temp MongoDB collection: uaDB
6. equilibrium state is reached when abs[(it(n) - it(n-1)] < x % (confidence interval to be decided, can be very high)
7. store, as binary, the equilibrium per user as [uid: ua] : ua.db (binary file)
8. hash-encrypt ua.db , store hash in latest blockchain block for consensus
9. witnesses read ua.db per user (let's say 8 bytes per user: UA(user 100,000) is found at byte 800,000