-
proof of concept for a prediction market to moderate comments by side loading a script through custom google tag manager
-
gets the substack user id + email + subscription id of current logged in user through window._preloads to create a "mantax id" which holds the predictions the logged in user makes
-
we can then give each user a calibration score when looking at comments
routes
- GET /user_scores?article={article-id}
- Gets all the scores for all the users who comment on this article, can use cloudflare workers to cache this json
- And then for any users that have commented since the scores were cached, we can get the delta and re-cache it
- And then update the cache anytime the predictions are reconciled
- POST /predict
- Takes either the mantax id of the logged in user, or enough identifying information to make them a mantax account
- Makes a prediction abount whether a comment will be banned
- POST /reconcile
- Will be a button on scotts dashboard
- Gives the emails and userids of the subscription list + a list of bans to mantax server, which can then reconcile all predictions and update scores
- GET /user
- gives back their own history, along with current amount of mantax bucks they have to spend on predictions
table
users
mantaxid
substack user id
substack subscription id
user email
calibration_score
mantax_bucks
predictions
comment id
mantax id
direction "ban / allow"