One summer, in college, my friend received a postcard. It was addressed: (his surname), Offaly, Ireland.

This is a classic example of where humans deal with ambiguity a lot better than computers. Addresses are tricky things for non-humans to understand. The London Road is unlikely to be in London but is likely to be in any number of places on the way to London.

If we want to look for something these days we are likely to type it into a search engine. If what we want to look for is a place, then that search will be interpreted against a mapping services and in case you hadn’t noticed, most of these are awful. Well, I mean, they are great –  mostly, but you would not want to rely on them.

The problem runs something like this:

Rathfarmham is a place in Dublin, Ireland. It is a place which people quite happily put in their addresses and then receive replies to.

The Google Maps product has a geocoder. If you send it an address, it sends back where it thinks that is. If you type in ‘Grange Road, Rathfarnham’, it will tell you where that is. However, it doesn’t really know a Grange Road in Rathfarnham, only a Grange Road, and because of the crazy old world country I live in there is another Grange Road a few clicks away.

Even worse, it doesn’t really know Rathfarnham is a locality. If I try ‘Silverwood Drive, Rathfarnham’, then I get back the correct ‘Silverwood Drive’, but if you look deep inside the response it thinks the real address is Silverwood Drive, Ballyboden.

Now until I used the Google geocoder I did not know where Ballyboden was, despite having lived close to it throughout my childhood years. Somebody, somewhere down the line has made the seemingly unimportant decision that Ballyboden is the main locality and that Rathfarnham is a place and not a locality.

Ok, so this stuff is just annoying, nothing more, right? Well not really. There’s money in search results. How many people who have a business think about the appearance of their physical address to search engines? I’m guessing very few. For instance it is common for people from the same business to write their business address in quite different ways. Even if you have a standard address and never deviate from it, is it one that is mapping data friendly? You might be missing out to a competitor with a less ambiguous address (machine-interpretable).

Ambiguity and commonality are the enemies of correct identification. Our ‘Grange Road’ issue becomes a more common problem if our address is Main Street or High Street. Now we definitely need another matchable identifier within the address to have any chance of finding our High Street and a different one on the other side of the county, state, or country.

The challenge is to provide a solution that is both machine-interpretable and human-readable. This has been solved by several countries such as Singapore where the postcodes have been honed to building level accuracy, and from there a standardised floor/suite syntax completes the address.

As we use more machine based location services it makes sense to use increasingly machine readable addresses. This is not to say we should give up the addresses of our forefathers to an alpha-numeric string or abstract machine interpretable symbol but there is a good case for including a unique identifier in physical addresses as standard.

At one stage ad hoc delivery/location solutions were fine: a package would cross an international border and a large purpose built organization would take responsibility for its correct delivery. However this approach is unreliable and archaic when applied to divererent automated systems controled by an expanding number of companies. The internationa situation makes life even more complicated as because of a network of varying standards and methodologies across the world. In the absence of international standards it is up to individual governments to assemble solutions, each requiring a separate approach.

Zip codes and other unique identifiers are one solution. However, these were often designed to handle a different problem and may not be sufficent accurate, relying finally on a level of human interpretation. Some of these are further tied to commercial organizations making their availability unreliable. For many applications any charges  will create an unacceptable barrier to entry. Location based services aren’t just going to be “find a pizza joint”; public information from health screening services to disaster emergency updates are part of the story too.

Some countries, notably for me Ireland, are falling worryingly behind. Systems are  being developed for public services that rely on commercial organizations with varying levels of commitment and skill. Where a government should see a responsibility to provide equal service support across its territory, current mapping and geocoding vendors will naturally apply their efforts to high return sectors first.

Many countries recognized early on the importance of a mail system to their societies, enacting legislation to ensure universal accessibility and harsh penalties for interference in its operation. As the transference of information and its type changes, the willingness to approach the associated opportunities at a societal or governmental level appears particularly moribund.