Hacker News Reader

Comment by remh | Hacker News Reader

Yeah this is a misleading title. Title should be: "Should Mapcodes Replace Addresses"

After working with address data for the past year or so, my answer is yes please.

Address data is so incredibly difficult to work with and it's pretty clear that it's difficult for everyone involved. Census data is incorrect, OSM is often incorrect, other free sources are filled with typos, post office is consistently bad (No, my last name is not Nicholson...), shapefiles are often haphazardly done. It's so difficult to work with. That said, the amount of quality work to even get it to this point is amazing. Serious kudos to everyone who helps out here.

I've spent an asinine amount of time recently trying to get Chicago's parking ticket data cleaned up. Lake Shore drive is probably the #1 culprit:

Unique lake shore drive instances:

  $ grep "LAKE SH" tickets.csv | awk -F';' '{print $7}' | sed 's/.*LAKE/LAKE/' | sort | uniq | wc -l

128

Example of the awfulness:

  LAKE SHORE DDRIVE
  LAKE SHORE DER
  LAKE SHORE DE WEST
  LAKE SHORE DIE
  LAKE SHOR EDR
  LAKE SHORE DR ''
  LAKE SHORE DR AV
  LAKE SHORE DR AVE
  LAKE SHORE DR EAST
  LAKE SHORE DREIVE

Given how many of those are typos I'm not sure how much advantage there is in replacing them with compressed incomprehensible strings. Typo one of these map codes and you'll end up in a different zip code and at first glance nothing will look wrong because they are designed for machines, not people.

To some extent, this could be addressed with standard error detection/correction techniques (e.g. error-correcting codes). But that would make the strings longer and often wouldn't help until they made it to a computer anyway.

The units have GPS in them. The number of possible streets near them is pretty small, so an autocomplete or similar might work really well.

Or just a dropdown.

Well, more it would emphasize (and pretty much require) a better organized system. The lack of emphasis on proper addressing from everyone makes life pretty hard. :)

I'm not sure I understand how that would even be possible. I'm in an unfamiliar neighborhood without my phone, looking for a business, and all I have to go by is: WQ.TY. How the hell am I going to find it? That's not written anywhere. Even if I know I'm currently at YW.SRN, how does that help me? With number, street, city, I can get to most places by just looking around at street signs and asking people.

Yes, as the article points out, there are places with unnamed streets. That's a bummer. But these map codes don't help unless the proposal is also to put signs up everywhere telling people where their map codes are and how to get to the next ones over.

"I'm not sure I understand how living in cold climates would even be possible. I'm in an unfamiliar neighborhood without my clothes. How the hell am I going to stay warm?"

In practice, you could either (1) not go to unfamiliar places without your phone, or (2) ask someone who does have a phone for directions.

Technology is great, but there should be graceful fallbacks.

(The idea of mapcodes is great though. I've been to a few countries where even cab drivers don't know the proper street names)

Reinvention of DNS. Speaking of money for imaginary rented identifiers, there is/was a LOC type for DNS. So meet me at pizzahut.com isn't going to work with thousands of locations (but... anycast?) although meet me at momandpopsfamilydeli.com will work.

If every Pizza Hut had its own DNS server (kinda expensive and impractical), and each location was guaranteed to respond fastest based on your location (which it won't today), then it would work.

Decentralized Pizza Resolution.

Like if you were at the Combination Pizza Hut and Taco Bell on Jamaica Avenue, and you're trying to meet up with your friend who's at the Combination Pizza Hut and Taco Bell on Jamaica Avenue, you might never actually find each other.

Reply on news.ycombinator.com