In the unlikely event that you are not already bored stiff with all this chattering about Ghana’s digital addressing system (DAS), let’s play an empathy game.
Supposing you were some kind of Big Shot and your decisions mattered hugely as to where this DAS system goes from here.
You look left and then right and see that in the midst of all the chatter there is actually just one fork in the path to a policy decision, leading to diametrically opposed destinations.
Some folks want you to immediately integrate the DAS into every other major IT system in place or planned – national ID, NHIS, Pensions, IRS etc. Some others want you to wait for it to grow for a year or two and then to run a major, independent, statistical validation exercise, and only if it passes a 95% accuracy test should you go ahead. What would you do?
Firstly, if you have any capacity for judgement, you would immediately mute 99% of the chatter so that you can sieve out the different types of truly coherent arguments on each side.
No sooner have you done that and you would realise that the operational concerns are not of the same make as the conceptual ones, and that it is best not to mix them up.
Everyone knows that every computer program can be ‘iterated’ to stability. Bugs can be fixed quickly. User interface defects can be addressed relatively smoothly. And infrastructural gaps can be plugged. That’s exactly what programmers do for a living.
Conceptual and architectural flaws, on the other hand, requires a completely different attitude. They are NOT programming problems. They require subject matter expertise that normally lies outside the field of information technology properly speaking. But sometimes they even go beyond subject matter expertise and reside in a delicate grey zone between raw cognition and creativity.
Any policymaker would want to quickly test if they have got a “conceptual bug” on their hands.
The first order of day when making such a determination is to eliminate the defects that on first sight seem fundamental but on inspection can be shown to be operational rather than conceptual.
In the case of DAS, there are a lot of them. One such one is the concern, frequently spouted in the chatter that has greeted the launch of the DAS, that multiple grid-cells in an address would necessarily lead to multiple addresses.
Whether or not multiple grid cells lead to multiple addresses is actually a matter of user administration. So long as *an address maps to a user account*, multiple grid cells can, in fact, be allocated to the same address. What is important is how to fix the user registration and administration problem such that when two accounts “lay claim” to the same grid cell, that grid cell is simply removed from both accounts. To obtain an address, therefore, would require logging of a minimum number of unique, unclaimed, grid cells.
This is a mere bagatelle, per se, from a programming point of view.
In fact, had Ghana Post done any focus group testing ahead of their rushed launch of the platform and then commissioned a UX review of the application, all these minor details would have been sorted out without hassle.
Having thus eliminated the operational issues, your approach to the “problem” as a, to use a favoured Ghanaian adjective, “seasoned” policymaker would then be to take a very close look at the conceptual and architectural issues. Indeed, only the clear confirmation of these would warrant a suspension of plans to immediately integrate the DAS into multiple, critical, government services in the short term.
Yes, it is true that there has been a good deal of theorising why the DAS requires conceptual retooling. My big mouth and frisky fingers plead guilty. The big mouth of Maximus Ametorgoh pleads guilty. And the frisky fingers of Elvis Kumordzie plead guilty.
Is that enough though? Yes, it can be mathematically demonstrated with ease that based on the current construction an 80% accuracy rate is practically impossible. Even though we are on the equator, where grid cells approximate amazingly well to true squares, the problem is one of the underlying topography. With zero validation of the actual address information and two sources of GPS errors: satellite and the transmitters/receivers in the many China and high-end smartphones around, the prospect of mismatching grid-cells across neighbouring addresses cannot be mitigated by clever tweaking of UX alone. Fundamental error correction is required. But this is still theory.
A crack policymaker would want more. They would want to know: what are the actual, practical, experiences of other societies in implementing postal geolocation systems identical to Ghana’s DAS?
If perchance they have assistants with some time on their hands, they could immediately review the record of the following projects in order to gain a quick first feel of the “localisation” methods that have been used elsewhere to accommodate this, admittedly, alien technology to suit the addressing needs of host societies:
1. What3Words implementation of grid-cell addressing in Djibouti.
2. What3Words implementation of grid-cell addressing in Mongolia.
3. What3Words implementation of grid-cell addressing in Ivory Coast.
4. MapMyIndia implementation of grid-cell addressing in the National Capital Region of New Delhi and environs in India.
5. What3Words franchising of grid-cell naming toolkit to the Local Government Association of Poland.
6. The S42 postal project in Trinidad and Tobago.
7. Nigeria’s NIPOST-What3Words combo deployment of grid-cell based postal codes (MEHI).
8. The Lebanon’s implementation of Natgeo’s UAS grid-cell mapping technology for their postal codes revamping project.
The first thing that should leap up from these various assessments should be the phased rollout of these postal initiatives (all of them entirely identical to what we have started), the separation of concerns between the geolocation system and the municipal addressing layer, and, note carefully, NO short-term integration of these geocodes into the national ID programs and other critical services.
The next line of inquiry must then be focused on practical insights in respect of some of the shortfalls that prevent the use of grid-cell geolocation technology as an immediate source of truth in identity management (which is how we want to use the technology in Ghana).
The following quick mini case studies might be useful.
1. Ireland’s Eircode and how it spurred the Open Postcode project.
In 2015, Ireland decided to digitise its postal code system and introduced the Eircode project. As a result of what is widely accepted to be lackluster performance, Eircode has failed to replace the legacy system. (read an early assessment here: https://www.irishtimes.com/news/consumer/six-months-on-people-still-confounded-by-eircode-system-1.2476492). A major gap in the conceptual design is the use of arbitrary numbering.
Open Postcode has tried to make considerable architectural leaps such as “resolution elasticity” etc.
Studying Ireland to understand the organic improvement pathway for digital postcode technologies would seem an obvious thing for the policy maker in our empathy thought experiment.
2. US National Grid and Military Grid Reference System
These were some of the earliest implementations of the grid-cell concept approach to postal code generation (the idea that latitudes and longitudes based on WGS are hard to remember and must thus be replaced by square grid notation).
This system was supposed to provide a one-core, nationwide, emergency response coordination layer (based on the USNG-NAD83 standard). Yet, since being proposed for this role in 2001, accuracy and uptake issues have not allowed that degree of entrenchment to happen.
3. And my personal favourite: a framework paper evaluating a proposal to implement a grid-cell plotting system to improve agro-surveying statistics by the United States Department of Agriculture. This paper raises very interesting issues about resolution enhancement when using grid-cell technologies that converges closely with my own perceptions, albeit from a different angle.
Our hypothetical big shot policymaker having studied all this material should now recognise that there is indeed an architectural and conceptual gap in the current implementation of grid-cell postal code digitisation technology by Ghana Post. In fact, she would realise that the several dozens of such project worldwide continue to be phased in and iterated step by step to remove these conceptual defects.
Nowhere in the world has there been a rush to inject grid-cell data forcefully and aggressively into national ID systems and other critical services.
Thus, despite reassuring herself that grid-cell technology is clearly a very powerful approach to postal address modernisation that definitely has a major role to play in improving our current model, she can only prudently take one course:
Suspend plans to embed the Digital Addressing System data into the NIA’s database and other critical, low error tolerance, databases, and immediately instruct Ghana Post to begin preparations for large-scale, independent, statistical testing once uptake has reached 500,000 addresses.
Now, that wasn’t such a useless thought experiment after all, yes?