StayOnMap puts every home where it really is. That sounds like a solved problem until you look at how homes are actually described. An owner types an area name that covers several neighbourhoods. A landmark is used instead of a street. A pin is dropped from wherever the owner happened to be standing when they posted.
Each of these is a claim about location, and they do not always agree. This note is about what we do when they don't.
Two claims, not one
Our first instinct was to pick the most precise source and discard the rest. We now think that loses something. When a written address and a pin disagree, the disagreement itself is useful: sometimes it is a typing mistake, sometimes it means the home is on the edge of an area whose name carries a different price, and occasionally it means the listing is not what it appears to be.
So instead of resolving location once, we keep both claims and record how far apart they are and in what way. A pin inside the named area is a different situation from a pin two areas away.
Why surroundings change the question
People do not rent a coordinate. They rent a walk to the station, a street that is quiet or loud at night, a shop at the corner. The same home described with the wrong area name can look like a bargain or a mistake depending on what the reader assumes is nearby.
That is why location sits underneath everything else we build. If the place is wrong, the surroundings are wrong, and any answer a system gives about the home inherits the error.
What we are still working out
We have not settled how a disagreement should be shown to someone searching. Flagging every mismatch would bury people in warnings about typing errors. Hiding them would remove one of the clearest signals we have. Our current direction is to let the gap feed into Trust Score quietly, and only surface it when it is large enough to change a decision.
We will publish what we learn as the work produces it, including the parts that do not hold up.
