Backed by: @pithy-sh/rating. Scopes: rating:players:read · rating:players:adjust.
What it shows
Rating is the capability it reads, pools is what a pool is, and the reference has the routes behind every read and act here.
The pools your games rate players in, each pool’s players by skill, and one player’s rating with the outcomes behind it, read live from your Worker. It appears in the rail as Ratings, beside Game results, when your Worker composes rating, at whatever path you mounted it.
The index lists every pool, with a filter. Each row carries the pool’s algorithm and the games that rate it, or No game rates it.
The record is one pool. Its algorithm and the games that rate it, then its players, highest skill first, a page at a time. Each player row carries their place, their skill and how many games it counts.
Choosing a player reads their rating: the player, their skill, their experience, the games counted and when it last changed. Under it come their outcomes, newest first, a page at a time. Each outcome carries the skill they went in with, the experience it gained, and the result it came from when there was one. A player with no rating in the pool says so.
A voided outcome reads as voided, with the result that voided it, in place. A rating that moved without a new game is explained there: a voided result took the outcome back and recalculated the rating.
A player is a user id. It links to their record in Users when your Worker composes auth and your connection holds auth:users:read, and is plain text otherwise.
Recalculating and resetting a rating
Two acts, on the player the record shows: recalculate their rating, and reset it. Each one previews exactly what it will change, asks for a reason, and confirms once.
| Act | What it does |
|---|---|
| Recalculate the rating | Rebuilds it by replaying the player’s logged outcomes through the pool’s algorithm. Voided outcomes count toward nothing. The preview shows the skill it starts from |
| Reset the rating | Returns it to where the pool’s algorithm starts. Experience and games counted are kept, and the preview shows both |
A rating with no logged outcome cannot be recalculated, and the record says so before anyone presses it. It predates the outcome log, so there is nothing to replay. Reset it, or leave it.
What is shown after an act is your Worker’s answer: the skill before and after. The player’s rating and the pool’s players read again after every one.
Each opening of an act mints one idempotency key, and every press of that opening sends it. A press whose answer was lost and is pressed again gets Already done, and nothing changes a second time.
A refusal your Worker names stays in the question, with what you typed, so you can fix it and press again: the pool is gone, the player has no rating, or the key was already spent on another act.
Each act is written to the dashboard’s own audit record: who did it, the pool, the player, the act, the reason, and whether it changed anything. Never a skill.
On every plan, and on Team
The reads render on every plan. Seeing the pools, their players and a player’s outcomes costs nothing.
The two acts are Team. Below it they are drawn locked, saying what Team buys.
Two scopes, a read and a write
rating:players:read is every read here, and it is a read, so it is in the default grant. A connection without it sees the pane refuse by name and send nothing. Scopes has how to add one to a connection made before it existed.
rating:players:adjust is both acts, and it is a write, so it is never in the default grant. A connection without it shows the acts unavailable, with the scope they need.
It keeps nothing
Recording an outcome and choosing an algorithm are not buttons here. Those belong to your game and its config.
Nothing here is kept. Pools, ratings and outcomes are fetched from your Worker for this screen and held in no Pithy table.