Hacker News Reader

Comment by couchand | Hacker News Reader

This reminded me of a little suggestion by Bret Victor (http://worrydream.com/MagicInk/#reducing_interaction).

Imagine honing in on familiar areas simply by typing the first few digits of a zip code—type “9” to immediately zoom into the US west coast, followed by “4” to zoom into the SF bay area and then “5” for the east bay. Because of the immediate feedback, the user can stop typing when she gets close enough, and use relative navigation from there.

I coded up an example a little while ago (http://andrewcou.ch/zip-to-it/).

View on HN

Are you keeping stats on your request? I'm curious how high 90210 stands out.

That's a great idea. (Un)fortunately it's all client-side.

Guessing you could probably put everything into a redis instance and just hincrby/incr for either the individual queries ie "902", "9021", "90210."

Not sure what would be more performant, REST or a websocket. Not doing anything would be the fastest, though!

http://redis.io/commands/HINCRBY

A Google Analytics tracking of the events would be just fine for this goal.

Oh, great idea! I've never dived (dove??) into event tracking. Is something like this what I want?

    ga('send', 'event', 'input', 'search', 'zip', zipCode);

That will do the trick!

This is really cool, but ultimately not that practical because it doesn't display the zipcode boundaries.

If it were integrated with zipcode boundary information as this [ http://www.usnaviguide.com/zip.htm ] site does with google maps, it would be pretty awesome.

Interesting idea. I didn't really try too hard making it a complete product, since I was only interested in proving the navigation concept that Bret Victor had described. But that's a good idea, and if you're interested feel free to fork it and add that capability!

Another example of the same zip code interaction idea, with some pretty UI: http://benfry.com/zipdecode/

I think this zooms in too far on the last level. Very nice, though!

I had the same concern when I built it, but the data source I was using only provides centroid information, not boundaries. I thought for a while trying to come up with a solution and ended up ignoring it, since I was mainly interested in a proof of concept. If you'd like to fork it and improve you're more than welcome to!

Reply on news.ycombinator.com