A score whose origin you don't know is worth zero. So let's say it right away: MushScout's forecast comes from three sources, and they work together from day one, even when your diary is still empty. To keep them honest there is only one rule: every number must be disproven.
The three sources
1. What is known about mushrooms
Each species starts with its own parameters: how much water it needs, in what period, within which temperature range, under which trees. It is the basis that allows the app to say something sensible the same day you install it, when it still knows nothing about you.
2. Everyone's findings, anonymously
The dates and conditions of the users' findings are aggregated over large areas: they serve to correct the starting parameters with what really happens in the Italian woods, which does not always coincide with the books.
Aggregate really means aggregate: no one sees anyone else's place, and the coordinates do not leave the phone of whoever saved them. How place privacy works we explain it in full.
3. Your diary
Your exits, in your places. It is the source that makes the prediction your instead of generic: as you record, the comparison stops being "September in the Apennines" and becomes "your beech forest, in the years in which it gave something".
There is no level to unlock and no minimum exit threshold: the three sources weigh together, and the weight of the third grows alone with the diary.
Because the empty exits count as much as the finds
This is where MushScout behaves differently from all album-like apps.
Knowing that in a place, in that weather, in that week, there was nothing, is as good as knowing that there was a full basket. Without the "no" there is no probability: there is only a list of lucky days, which is exactly how we remember the woods by heart - the good years remain, the empty-handed Sundays disappear.
That's why it's worth it also note the outings that ended badly. It costs ten seconds and teaches the app when it is best to stay at home, which is half the service.
A prediction that can be proven wrong
Under each low score it is written what is holding it back, and by how much: "12 mm dropped, 28 more needed".
That 28mm is not a conservative figure of speech. They come out questioning the model itself: they are the rain which, if it fell, would bring the score above the threshold. It means that the forecast is exposed — if it rains 30 mm and nothing comes, it's wrong, and you'll see it first.
It's the opposite of a bulletin that says "fair conditions": that cannot be denied because it didn't say anything.
Two things that we measured and that proved us wrong
It would seem logical that the personal story of a forest — "here, on September 12, I always find" — is the best way to order the releases. We tried it: orders worse than the weather. The past of a place says how it went, not how it will go, because every year has its water and its heat.
For this reason the notebook of each forest, year by year, is declared for what it is: diary, not prediction. It's there to understand if you're early or late, not to decide for you.
The second: four mushrooms growing on the wood remain unscored, on purpose. Measured against real observations, the weather predicts nothing useful to them — and a number worth less than a flip of a coin is worse than no number at all.
Yesterday's weather, not yesterday's forecasted weather
The app takes the rain of past days from the weather archive, that is, from what actually fell, not from the forecasts made then. The difference is not theoretical: in the same forest, in the same period, the forecasts gave 55 mm and the archive counted 148.
With 55 mm that forest remains closed; with 148 he is ready. An exit is decided on this data.
The weather data comes from Open-Weather, licensed under CC BY 4.0.
When you know more than the model
A weather cell is kilometers wide and in the Apennines it fits both the valley floor and the ridge. If you were there and you know that it rained differently, you tell him: you choose the day, you indicate how much water has fallen, and for that place your word beats the model.
It is the piece that closes the circle: the app learns from the data, but those who walk there always know more about a single forest.
Frequently asked questions
Do you need to record a lot of releases before it works?
No. It works from day one with the scientific part and the aggregated data, and it becomes yours as you record. If you want to shorten the time you can write down the trips you have already made from memory: the app finds the weather for those days on its own.
Why should I mark a bad exit?
Because it tells the app when Not it was time. It is the data that makes the forecast reliable, and it is what almost no one collects.
Do my finds end up in a shared archive?
Only aggregated data on large areas and in anonymous form enters the common part. Your points remain yours: no other user sees them, not even your friends.
Does the app also learn from the photos I upload?
Recognition from the photo is a separate function and serves to offer you a name to verify. It is not an identification, and should never be used to decide whether a mushroom should be eaten: for that you need a mycologist from the ASL. We'll talk about it in cards and security.
If I change area, does the forecast start from scratch?
No. The scientific and aggregate parts are valid everywhere; what's missing is only your history of those new woods, which is built up exit after exit.
