How the Hive Mind works

What we log

When you submit a burn test from My Maker's Log, we store the recipe shape (wax family, wax SKU, wick family and size, vessel diameter, fragrance load, fragrance IDs, additives) and the outcome (total burn hours, melt-pool measurements, symptoms you flagged, and whether the wick was right, too small, or too big).

We do not share your name, your saved recipe contents, or any free-text notes. The Hive Mind aggregate strips everything except the counts.

How we bucket

Two burns land in the same bucket when they share:

• Wax family (e.g. soy, paraffin, coconut blend).
• Wax SKU when known. We fall back to wax family when the SKU-level sample is thin.
• Fragrance set, normalised by sorting and de-duplicating the IDs so blend order does not fork the bucket.
• Vessel diameter rounded to the nearest 5 mm.
• Fragrance load rounded to the nearest whole percent.

The landed wick

For each maker and each recipe, we record one "landed" wick: the family and size on their most recent burn marked "correct". If they have not marked a burn correct yet, we use their most recent log so first-time burners still contribute a signal. Each maker counts once per recipe, no matter how many burns they have logged.

Sample-size floors

• Below 5 landed wicks in a bucket, we show nothing. The engine pick is on its own.
• Between 5 and 9, the headline is marked "tentative". Treat it as a hint, not a verdict.
• From 10 to 24, the headline is "solid".
• 25 or more, "strong".

Every card also shows the alternates and their share so you can see how split the bucket actually is.

Your logs override

Once you have three of your own burns logged for a recipe, the refinement panel switches from "Makers like you" to "Your batches say". Your hands-on data is always treated as more reliable than the aggregate.