Page Comparison

Current status:
I don’t see any value going ahead with this until we have some fixes for locattion accuracy on Android at least, to ~+/-20m. Current wildly variable accuracy will result in false positives & false negatives.

Trying to “mask” the inaccuracy using 7 char geohashes (150m x 150m) will simply result in even more false positives.

Field trial will be good for testing false negatives - i.e. we can try to target positive scenarios & see how many we hit.

Targeting false positives will be harder: we’ll need to create contribved scenarios in which we expect false positives (A sits in restaurant, B sits in bar across the street). Then we’ll observe false positives. But with 7 digit geo-hashes we can be very confident we’ll get false positives on those circumstances, so there’s very little to be learned there.

Some illustrations of the problems with 7 char geohashes in here: Moving Safe Paths & Safe Places to use geohashes natively