How the index works

Every campaign owns a full 100%. Its mindshare is not a slice of some global pie and is never contested by another campaign — it is divided among the supporters carrying that campaign's conversation, in proportion to what they scored. Rewards settle against that split, which is why it has to belong to the campaign alone.

The one figure measured across the whole index is the supporter board, which ranks accounts by total scored activity everywhere. That is a ranking of people, not of campaigns.

Every signal a campaign is tracked by is optional: a ticker, its hashtags, its token's contract address, posts tagging its X account, and replies in its X account's threads. A campaign declares whichever of those it actually has, and a post counts toward it if it carries any one of them.

Every matched post is worth a base credit plus compressed engagement, scaled by how credible its author's account is, and that worth is credited to each source the post hits. Campaign is the finest grain scored — network figures are sums of it.

Mindshare is that weighted score divided by the total scored inside the same campaign. Every campaign owns a full 100% — campaigns are never ranked against each other for a slice of one pie, because a small community holding its own conversation is not worth less than a large one holding part of a bigger one. Rank deltas compare to last week.

Boards are built from real posts. Once a week the scanner searches X for every tracked ticker and hashtag over a Monday-to-Monday UTC window, keeps the posts that genuinely carry the campaign, and folds them into one row per author. Retweets are excluded — they carry no authored content. The week is stored as a snapshot, so a board that has shipped does not silently change when a handle or a campaign is renamed later.

Mentions is how many qualifying posts you wrote. Originality counts the ones that were posts rather than replies. Smart engagement is the reactions those posts earned — likes, replies, reposts, quotes and bookmarks, weighted by how much each costs to earn. It is square-root compressed so one viral post lifts a standing without deciding the week, and scaled by a credibility factor that discounts brand-new accounts and follow-back farms. It is the heaviest of the three because it is the hardest to manufacture.

There is no LLM originality judge and no smart-follower graph yet: originality is a structural test, not a judgement of the writing, and credibility is inferred from public profile signals rather than from who actually follows you. A post is credited to a campaign by ticker or hashtag alone, so an unrelated post using a colliding cashtag can land on a board.

Reward pools are not funded and payouts are not running. Every board you can read today is the scoring model working; none of it has moved money yet.

Supporter accounts register through X. The handle a board ranks has to be one X vouches for, so there is no way to type one into a form and claim it. Once connected you can add an email and password as a second way in; the address is confirmed by a six-digit code from your inbox before it works as a login.

Project accounts are separate — a different database, a different sign-up, a different session — and use an email and password only, confirmed before the developer console opens. Holding both means holding two accounts, on purpose: the people being ranked and the people funding the pools should never share a login.

Passwords are hashed with scrypt and never stored in plain text. Sessions are signed, HttpOnly cookies lasting seven days. An account does not change your score — scoring reads public posts only, whether or not you have ever signed in.