Do we still need addresses?

When Ireland introduces its postal code system this year, it will be unusual for a number of reasons.  Each residence will receive its own code (Eircode) – not unusual – but these codes will not be related directly to place, nor be apparently related to those codes assigned to neighbouring residences.  It’s the first time (that I can think of) that codes are being created by a business with a profit motive (Capita) at the time of implementation, and by virtue of its design most organisations wishing to make use of the codes, be they businesses, utilities, local authorities, pizza deliverers, couriers or the emergency services, will have to purchase files and services from Capita to decode them.

Creating codes (other than postal codes) to replace or enrich addresses is nothing new. There have been barcodes, phosphate stripes, QR codes, GIS co-ordinates, what3words, Geotudes,  Posttudes,  and more. Because they are codes, unless they are used with information which is readable for humans, such as a postal address, they all need de-coding, and generally they decode to either a latitude/longitude or map, or the postal address itself.

The impression I get is that these are often solutions looking for a problem. Postal addresses are logical and useful and therefore almost universally applied over the whole planet. They form part of our daily experience, contain a set or instructions in a hierarchy which allow us to find almost any point, and are an integral part of our mind maps. Details can be added or dropped as required. Addresses often contain far more information than any code system.  No code system, including any postal code system, contains information on how a package is to be routed beyond the delivery point – floor numbers, room numbers, departments and so on. Addresses contain meaning for the humans who need to understand them, which no coding system can ever have.  They are linked to many opening available information resources – maps, signs containing the place name as you enter the city, street name signs and numbers on buildings. When I speak to a friend about a building on a street, it means something even if they don’t know exactly which street, because we share knowledge of our district and can present information in context with other information.  No code can replicate this.

What words?

Take what3words as an example. They have split the earth into 9 square metre chunks, each of which can be identified using just three words, following the logic that words will be easier to remember than strings of numbers. As these words have no relationship to each other or to the piece of Earth they are describing, they are useless unless decoded. Gingen.wieso.aussichten is next door to nachbarland.miteinander.abzuwarten and opposite sünde.frühen.gespräch. No use at all to the man or woman on the street, sensitive to error and dependent upon a website being up and a company surviving to be of any practical use.

Codes, by their very nature, may be restrictive and inflexible. The 9 square metres covered by each what3words code, for example, might cover only a small part of a building, several dwellings, or a tower block with numerous inhabitants, businesses and delivery points.

Any notion that we can remember a code, such as three words, more easily than we can remember addresses, is spurious, because the codes are apparently jumbled and random. Addresses are not, and they are written in a natural language. Furthermore, any addressing and coding system needs to be universal, at least on a country basis. Efforts to win us over to one of the many competing systems can’t work whilst others choose to use a different system. They just lead towards the nightmare of having to provide 17 alternative addresses/codes on one’s website instead of the one we use and understand currently. Every addressable object should have one address only, and not a multitude of codes.

Code systems are prone to errors. In the Dutch postal address system, a postal code and building number is sufficient to point to a distinct delivery point.  1018 VV 80, for example, points to Nieuwe Prinsengracht 80 in Amsterdam.  Yet any typo, for example 1081 VV 80, will make this “code” useless – the supporting information of a thoroughfare and place name is then required to locate the error in the code. What3words’ shorthand.marathon.rail is in Alaska, whereas shorthand.marathons.rail, just one letter off, is in Kenya. This lack of human interpretability is causing consternation with the new Irish postal code system. The emergency services, for example, point to being unable to use an Eircode to locate even a small area without decrypting the code, whereas their colleagues in Northern Ireland can interpret an area from their postal codes, which contain geographic pointers. This is the reason that the Eircode is expected be used as part of an address and not as a replacement for the address.

Furthermore, the management of code data files is as prone to the possibility of errors and data quality failings as any other data file. Without an ability to humanly interpret a code, any errors within the master records could have serious consequences on the ground. Codes add a layer of complexity (and, potentially, error) into a system.

Whilst coding systems can improve and clarify addresses, and can provide a delivery system in areas where no street address system yet exists, they can never replace postal addresses as we know them. The future of postal addressing is secure.


  • Gary Allemann

    Interesting post Ken. This relates to a couple of posts I wrote recently in which I responded to a hypothesis that address data was no longer relevant due to the desire by corporates to shift to cheaper media such as email.

    This post drove quite a debate on LinkedIn which is summarised here http://blog.masterdata.co.za/2015/01/15/quality-address-data-business-drivers/

    Quality address data is still very much relevant and unlikely to go anywhere soon

  • cprescott7

    As a way to organize a systematic set of locational designations which are basically “interlinked and cross-referential”, it’s hard to beat street addressing. Where I find what3words most useful for is in the interstices, the blanks in the “usual” address location systems. This became very personally clear to me when I heard what3words’ founder Chris Sheldrick speak about his system at the recent PostalVision 20/20 conference in Washington, DC. He spent 10 years booking bands and getting them hither and yon to their performances. Many of these places were enormous conference centers or theaters for which he was given a street address, but no indicator of the location of the “relevant” entrance. Of course the street address was the front entrance, whereas the entrance to the performance venue or back stage might be hundreds of yards away, and not blessed with an address designation in the street addressing system! This struck home because I once walked two miles or so in Dusseldorf to get to the “correct” service entrance for the stage where I was speaking!

    As a locator, provided you have access to the secret sauce, then what3words is a handy and accurate tool. As a system for public use where locations need to be “ordered” for many public uses and locatable by someone without access to the secret sauce, Graham is right.
    Charles Prescott, Global Address Data Association

    • GRBB

      Charles, aren’t your examples more a case of poor information provision? A traditional street address designation can always be enriched. If the organisers had told you/Chris Sheldrick that “the artists’/speakers’ entrance is 200 metres from the main entrance to the left – please follow the yellow signs”, wouldn’t that have been more useful, using pointers on the ground, than a code which requires a phone, an app, an internet connection ….and more? Not that street addressing couldn’t be improved – Germany’s system is heavily burdened with historical weirdness, as you found out, but adding a pretty opaque layer on top of it seems to me to be, in most cases, overkill.

  • rshibley

    Very interesting post. You should check eddress, http://www.myeddress.com. They take another approach at it. The code belongs to the user. So one code for Home, no matter where home is. The code never changes you just point it somewhere else.

  • Pingback: Open Location Codes | Blogs and Resources from PCA Predict (formerly Postcode Anywhere)()