Monday, December 12, 2022

Claim inflation


I’ve specialised in international data and its management for 30 years now, and everything I learn (which I still do daily) enforces the fact that the foundation anybody needs for successful international data management is knowledge. Knowledge, that is, of the world, its systems, conventions, languages, cultures and ways. Knowing how to code brilliantly is unhelpful unless you know what you need to achieve, and knowledge is essential when choosing a partner to help you with your international data quality.

I recently came across a provider of international address validation which claimed to support “250+ countries”. Defining what a country is is not as straightforward as you might suppose. It depends on who you are, where you are, and your political background. There are unrecognized de facto countries and non-existent de jure countries. Even so, however liberal your definition, you would not get anywhere near 250. If you’re counting the more accurate “countries and territories” then you’d get closer, but 250 remains claim inflation. There was a time when every address validation company was trying to outdo the others with country support number inflation. One supported 240 so the next claimed 250 and one even went for 300 plus, which is just ludicrous. This had calmed down, so I rather hope that this new claim is not the start of a new round of unsupportable claims. The company claiming 250+ includes uninhabited rocks (they may have an ISO code, but there are no addresses to validate) and non-existent political entities such as Antarctica. Check the claims in more detail, and they become more preposterous – they claim validation to postal code level even for countries and territories which do not have postal codes.

I would feel better about seeing claims like this if I thought that most people dealing with international data were well enough informed to be able to go to this company and say “you claim to support more countries than there are, how can we be expected to trust you with our data?” This wouldn’t have to happen often for providers of these services to sober up and start telling the truth. The company concerned claims 2800+ customers, including many large companies which should understand addresses. I understand the pressures that companies put themselves under to market and sell their products, but claims need to be based on truth. I did contact the company to ask about this – I received no response. If more people working with international data would educate themselves better in … international data… then that data would be better managed, cleaner and better governed. Let’s hope that things improve in the next 30 years.

Sunday, October 23, 2022

Yes, but Google ...


Yes, but Google …..

That’s the start of so many sentences that I hear and read, and the prologue to having to explain, over and over, that the mighty Google are as prone to errors and have to follow the same data paradigms as other companies. Google is very good at some of what it does. In other fields it is average or, if I dare blaspheme, it is poor.  Yet Google is constantly being held up as the arbiter of everything that is correct. If Google says it, it must be so. It is the law, even in aspects as esoteric as language translation. In some cases this is just ignorance. In other cases organisations know that Google is wrong, but follow anyway because they make a commercial decision that they cannot go against the direction of the unstoppable machine that is Google.

This is a worrying trend which goes against the dictates of data quality.

Every database contains errors. Every database contains duplicates. Every database. Including Google’s. Google also lack knowledge, or lack the ability or desire to apply knowledge, in many areas. Problems may be the result of poor data management practices, of which Google is the victim just as much as anybody else; and of the perennial and ubiquitous problem of lack of knowledge or lack of motivation to acquire the required knowledge.

Thinking very specifically now of Google Maps, at the time of writing you may see a lot of duplicate information where they have merged sources and been loose with their de-duplication.  That single electric vehicle charge point at my local railway station? Google shows three. Those multiple building numbers on Hawaiian buildings on their maps? Duplication, because Google doesn’t have or apply the available knowledge about their format so doesn’t realise that 91-123, 123 and 91123 are all the same building. The failure of Google to find addresses in the borough of Queens in New York? Again, a failure of knowledge about local variations in address systems. And, more often than not these days, the format of addresses displayed in Google Maps for many countries is demonstrably incorrect for that country.

That’s how things are now, and Google does change things around a lot so these aspects may no longer be an issue as you read this. Instead, other problems will pop up. Because Google makes mistakes, just like anybody else. What really worries me, though, is how people can’t see, or can’t accept, that Google is anything but perfect. Will Google’s errors cause institutions to start formatting addresses the wrong way, because “Google”? I hope not. In the meantime, I shall keep plugging away and explaining, every time I hear “but Google …”, that Google has a long way to go before they reach omnipotence in knowledge and its application. It’s not even close. So, please spare me the “Yes, but Google …” 

Tuesday, October 12, 2021

The unaddressed ... and, inevitably, What3Words ...


I received an e-mail the other day from one of the millions of “unaddressed” people of the world, living where there are no street names or building numbers. He lives in Ghana, and does have a postal code, a code which resolves to a GPS location so that Ghana Post can deliver to him.  But that’s his problem – it’s only used currently by Ghana Post. He would like to order from companies outside Ghana, but they all require a street address and none will accept the Ghana GPS code, nor a latitude/longitude. What to do? I wish I’d had a short-term solution for him.

There are around 30 global code systems that are eager to fill the unaddressed gap, and a further 20 or so which work at a national level. No organisation would be keen on implementing all 50 systems in their online retail portals – in fact, few organisations seem keen to implement any at all, despite companies such as What3Words throwing ridiculous amounts of money around to try to be the default choice for adoption. Adoption by a one organisation wouldn’t be sufficient – the whole chain, including all delivery companies, would need to adopt the same code system too.  Would Ghana Post be willing to deliver mail using another company’s code system?

Regardless, some of these code systems have been around for a number of years, and their adoption rates, despite their best efforts, remains low.  There are good reasons for this. Postal address systems are very varied, both within and between countries, but most consist of similar sets of information and all, to a greater or lesser extent, can be interpreted by using something we all have with us at all times – our brains. What3Words likes to market itself as new and edgy, a start-up; but it was founded in 2013 – almost middle aged, in my book. They’re haemorrhaging money at an alarming rate. In the good old days, questions would be asked about a company that wasn’t even close to even covering its costs after 8 years. But it appears that investors will continue to throw their money into this pit despite increasing rates of negative publicity about its many flaws. What3Words, in their overweening conceit, simply will not accept that their system is anything other than perfect, despite obvious proof otherwise. This will be to their cost – there’s only so much their marketing can do to hide the facts. At what point should it become clear that What3Words and other, similar, systems are not what people are looking for? The amount What3Words spends on marketing and legal procedures each year could provide a lot of Ghanaians with the infrastructure required to give them the addresses they sorely need.  I know where I would prefer to see this money spent.

Wednesday, September 29, 2021

How useful are telephone numbers in addresses?

Some people include a telephone number in their street address information in order to improve the chances of a package being delivered. In fact, some countries include a telephone number in their official address block format. But in a world where parcels are increasingly chucked over fences or launched from moving vehicles in order to achieve faster delivery times as margins become ever smaller, what chances are there that a courier would take the time to call a number to try to improve the chances of a package being delivered? Even the better courier companies have, thanks to Covid-19, abandoned getting signatures acknowledging package receipt. The way things are going, I foresee package delivery going down a tariff path similar to that followed by airlines and health services – the standard price you pay just gets your parcel into the system. If you want it treated well and delivered to the intended recipient, you would have to pay the premium.

But even if it were practical for the courier to use their time to call the intended recipient to help them get to the delivery point, how useful would it be? We all know that it’s not easy giving accurate directions to somebody even when they’re standing next to you and you’re both facing in the same direction. What are the chances of providing enough useful information when you don’t know where the courier is, in which direction they’re pointing or which positional coding app they happen to have on their phones?  If they cannot find you with the address information already provided, would a telephone call provide enough information to help? Apart from the occasional “I’m outside the front entrance, which floor are you on?” type of request, I wonder about the usefulness.

What do you think? Have you experienced telephone numbers in addresses providing a useful addition? Are they actually being used? Any anecdotes? I’d love to know.

Monday, November 13, 2017

When is an implemented code not an implemented code?

Regular readers know that I am not a fan of locational coding systems as a replacement for postal (humanly readable) addresses. I do not believe they can replace humanly readable addressing, and, despite a lot of hot air coming from various companies, I have yet to see a system in full working order.  

Take What3Words, for example.  OK, so I know I seem to bang on about them a lot, but I have a strong aversion to hype, and a stronger aversion to any organisation that sells themselves through clever marketing shored up by – well, very little else that is apparent to me.  Anyway, if you live on the oxygen of publicity, and you keep sticking your head above the parapet, you have to expect to be shot at.

So, What3Words.  They have announced in the past couple of years tie ups with various national postal services – Mongolia, Sint Maarten, Côte d’Ivoire, Djibouti, Tonga, Nigeria, Solomon Islands and Kiribati, in that order.  What3Words is an off the shelf solution – it should be fast and simple for any organisation to implement.  So, where are the implementations? I look at a lot of addresses in my job – I data gaze millions of addresses – and I still haven’t seen a single locational code actually being used.

So, I set myself a task – check these countries’ websites for progress on implementation. 

Mongolia has a page of information about What3Words. Sint Maarten has nothing (that I could find) but there is a video on their Facebook page.  Côte d’Ivoire, Djibouti, Nigeria … nothing. Not a word. Tonga’s website had been hacked when I was checking…. Only on the Solomon Islands’ website is What3Words given the place it should have if it is a replacement for, or supplement to, the existing humanly readable addressing system. “Introducing Solomon Islands [sic] New Addressing System” it trumpets. A sound I would have expected from other websites.  But it is not to be. In fact, not a single one of these websites, even that of the Solomon Islands, has the contact address for the postal service concerned given in anything other than a traditional postal address format. Not one contains its What3Words' address. 

Leading by example? Apparently not.

So, what’s the progress on implementation in those six countries outside Solomon Islands?  Is it to be kept secret from the users?  Will it be quietly dropped? Or am I over estimating the speed at which these organisation work? (Though if Solomon Islands can do it, this should not be a valid excuse for the others). If it’s the latter, I notice that both Lebanon and Mongolia (again!) announced a partnership with NAC to use their codes in 2013. Four years later and nothing (visible) has happened.
This is not to say that code systems aren’t being introduced, and implemented.  Look at Ghana, for example, happy to publicise and implement its sparkling new home-grown system, and to publicise its own address in traditonal human-readable form, and as a locational code. I am curious to see how the uptake for that system is, and how well this implementation sticks.

So what’s going on here?  The emperor’s new clothes?  Crying wolf? Let’s see some implementation, and measurements of the success of new systems.  All this announcing without follow-up is unhelpful in the extreme.

Tuesday, February 2, 2016

Eircode, six months on

The Irish Times took a look at the uptake of the Eircode six months on from its launch and concluded that uptake amongst the ordinary citizens of Ireland was sparse – about 2% according to one post office worker. Highly unscientific, but pretty interesting nonetheless.

Would it be inappropriate if, at this point, I got up on my desk and jigged around, shouting “I told you so!”?

Take up of postal code systems is always slow – I remember how long it took all the members of my family to start using their UK postal codes – but codes which are not designed with people in mind will, I think, never fully succeed.

We are not computers and do not think like them. Our geographical psyche works on an ability to associate with place and to be able to connect with other places nearby. In other words, if my land line telephone area code is 0495 I know that that’s the code my neighbour also uses. If my postal code commences NR14 1 then I know that that’s the same for everybody in my street. If my house number is 7 I know that my neighbour’s number must be 5, 6, 8 or 9. When my address is “High Street” then that’s the address for everybody else in the same thoroughfare.

The Eircode, and many other codes which for profit companies are launching, such as What3Words, Geotudes and Posttudes, fail to take into account the way real people think. Codes can be a boon for businesses with the infrastructure and skills to manage and decipher them (and that, after all, is where any profit for the code companies is going to come from), and the use of codes to provide temporary addressing in areas of world without an address infrastructure has merit; but without taking account of us, the people, I think their time has yet to come.

I can’t remember the What3Words of my address, partially because I don’t need to know or use it, but also because it has no connection with that place and is not part of my mental map. My mental map, and those of most people, is composed of significant (usually, but by no means always, named or numbered) features – buildings, streets, hills, trees and so on. Information about my residence which doesn’t fit into that structure is easily forgotten. When I can’t remember my own codes, there no chance I’ll ever remember those of my neighbours, which have no connection at all with my own. Though one gets the impression these days that few people can get by without a smart phone glued to their palms, I don’t believe that people really want to have to use technology to decode their environment. If somebody asks me where the station is, or how to get to the car park, or where the nearest supermarket is, the chance that I will ever resort to a code is zero. And when I have to call the emergency services because my neighbour’s house is on fire? They’ll have to get the required information the old fashioned way.

Codes such as the Eircode will eventually become more used, but I feel much of that progress will have to do with a certain level of coercion rather than a natural increase of uptake. Similar location codes launched in Middle Eastern countries only gained even a minimum of traction when the population were required to use them for essential services, such as their utilities. But will these codes ever become an integral part of people’s daily lives? 

I remain sceptical. 

How Google could improve Open Location Codes

A blog post, in collaboration with PCAPredict, about how location codes, specifically Google’s, can be made more relevant to human users. January 2016. Read it here.