Match where it is rather than how it's spelled
Here are two addresses. Tell me if they're the same place.
The autocomplete that couldn't finish a word
We turned the demo into a real geocoder. You type an address and get a rooftop coordinate, computed in your browser without a server. The last feature was autocomplete. It completes the city name while you type. We had shipped autocomplete as a command-line tool days earlier and seen it rank San Francisco above San Diego, so we considered it done. We connected the same function to the search box, typed New Yor, and it suggested Denver.
Can you fix order-blindness by turning up the volume?
Two days ago we retired a piece of decode-time implementation we'd been leaning on,
One row crashed our corpus build. Twice, on the same character.
Our training corpus is built from source. The build combines hundreds of millions of address
The model said P.O. Box all along. We weren't listening.
Last night our release check failed twice before it passed, and both failures had the
Four numbers tried to lie to us in 24 hours. We shipped anyway.
Yesterday we wrote about a lookup table that scored a perfect 100 and nearly talked us
A lookup table scored 100%. We shipped the model anyway.
This morning we published a post that ended with a tidy rule: some address tags don't
The right name in the wrong state
Our resolver scored 93.7% on the metric we'd been quoting for months. On the same addresses, its median answer was 326 kilometers from the truth. Both numbers were correct. Only one of them told the truth.
We spent three retrains fixing a German bug that didn't exist
Our parser's international-order German 'collapsed' to 44%, so we retrained it three times. Then we measured the thing that can't be gamed and found out the model had been right all along. The metric was lying. Native German was sitting at 96%.
Which Berlin? When your metric grades the wrong thing
Our resolver kept dropping German addresses in New Hampshire, and our scorecard handed it a gold star every time. This post shows how a name-match metric lied to us for months, and how much a postcode hint improved placement once we measured by distance instead.