Turning GPS Coordinates Into Postal Codes

GPS gives you coordinates like 51.0447° N, 114.0719° W, which are precise but terrible for comparison. Two caches 50 metres apart look completely unrelated numerically. Geohashing fixes this by recursively subdividing the planet into a grid and encoding each step as a character.

Start with the whole Earth. Split it in half vertically (longitude): west gets 0, east gets 1. Split horizontally (latitude): south gets 0, north gets 1. Now you have four quadrants. Pick the one containing your point and repeat—subdivide again, append another bit. After 50 iterations you’ve narrowed it down to about a square metre, and you have a 50-bit binary string. Encode that in base-32 and you get something like c3nfb7q.

The beautiful part: nearby locations share prefixes. Two caches in the same neighbourhood might be c3nfb7q and c3nfb7w. Truncate to c3nfb and you’ve described a 5km × 5km area. The string length controls precision, and proximity is visible at a glance.

fun geohash(lat: Double, lon: Double, precision: Int = 6): String {
    val base32 = "0123456789bcdefghjkmnpqrstuvwxyz"
    var (minLat, maxLat) = -90.0 to 90.0
    var (minLon, maxLon) = -180.0 to 180.0
    var hash = ""
    var bits = 0
    var bit = 0
    var isLon = true
    while (hash.length < precision) {
        val mid = if (isLon) (minLon + maxLon) / 2 else (minLat + maxLat) / 2
        if (isLon) { if (lon > mid) { bit = bit or (1 shl (4 - bits)); minLon = mid } else maxLon = mid }
        else { if (lat > mid) { bit = bit or (1 shl (4 - bits)); minLat = mid } else maxLat = mid }
        isLon = !isLon
        if (++bits == 5) { hash += base32[bit]; bits = 0; bit = 0 }
    }
    return hash
}

: mid ( min max -- mid ) + 2 / ;
: split ( val min max -- min' max' bit )
    2dup mid >r over r> > if swap drop 1 else drop 0 then ;
51.0447 -90.0 90.0 split    \ lat: updates range, leaves bit on stack
-114.0719 -180.0 180.0 split \ lon: updates range, leaves bit on stack

Google Maps uses this internally for spatial indexing. The Geohashing.org algorithm (2008) by Gustavo Niemeyer made it popular, though Morton codes did similar interleaving decades earlier. When you’re searching for caches within 2 km, you query all hashes starting with your prefix instead of checking haversine distance to thousands of points.