Just like many tech enthusiasts, the announcement of a
digital solution brings life, hope and joy. Also having a new government
in place meant that we could afford to give our policy leaders the
benefit of the doubt and live for a brighter day (you'll get this joke at the end, I promise). Yesterday, Ghana launched the Digital Addressing System in a well attended event and this is my two cents on it.
Just as with many products in Ghana, you really don't know
what you're getting until it arrives. From the start, we were furnished
with little information about what the specific goals of the digital
addressing system intervention are. But judging from the president's
address at the launch yesterday, the Ghana post gps system was expected
to give every house a digital address which would replace the "behind the
Koko seller" and "just after the big billboard" style of giving
directions, which has a large place in our current addressing system. On many fronts,
the Ghana post gps system has come really short of realistic
expectations. A complete overhaul in system design will be needed, if
we want it to deliver in the right way. The application was developed by a
third party (asaaasegps) and from what I gather, the Ghana post gps
application is just a fork of the asaaasegps application with some few
modifications. I will keep my sentiments about the bidding process and
selection criteria out of this blog post. I want to analyze the thought
process of those who were given the responsibility to solve Ghana's
addressing system problem.
Digital: Ghana is experiencing a digital bubble, it
didn't start today and it has surely been on for at least 5 years. Any
proposed solution with the keywords "app","digital","bio-metric","innovative" "easy to use" among others seem to be the automatic
answer to the problem at hand. From the president's speech yesterday,
the last time Ghana's addressing system was revised was somewhere in the
year 1974. Having grown up with the "plot" system which has a standard
plot size of (100mX100m) with an allowable range of 7,000 - 10,000
square metres, you would assume that any system supposedly designed to
uniquely identify property would use a ratio similar to this, since we have been building our houses from the year 1974 and maybe even before 1974 based on this.
Unless of
course asaasegps has outdated and incorrect information on their
website, they use the grid system to give every 5mX5m space a unique
digital address, with Ghana having around 16.1 billion digital addresses.
My primary suspicion after reading the 25 square metres declaration
from their website was: looks like my house will have more than one
address! (This makes the whole system entirely flawed and quite similar
to having more than one travel passport). I tested my hypothesis and even
though feedback from the app told me it's best if I stand outside, I was
able to generate two valid addresses from one house; one from my brother's
room and one from my room. The problem with two valid addresses for one logical location comes in this scenario, if I go to present this information
to a third-party, I really don't see how they can detect that I have
generated two addresses from one location, with which I can get
clearance to do anything I want. I can decide to use the address from my brother's room when I am going to take a loan from the bank and use the address from my room as my address when applying for a passport to travel out of the country and avoid having to pay the loan i took from the bank. Since it is an issue of identity,
having two valid identities is worse than having no acceptable identity
at all, since you will have to provide a verifiable and reliable identity
document with the latter.
OK so maybe saying a complete overhaul is needed is a bit radical, let us try and look at what can be done to fix the problem, since the project supposedly cost GHS2.5m.
A back-end algorithm to group my brother's room address, my room address and all open space within my house under one logical address is not implemented and cannot be implemented. There's just no way you can tell from an algorithm running on a server that they are all in the same house and not in different houses. [Unless of course they zoom in with Google satellites and using an image recognition algori...... which is another thing ]. We have many dispersed and unplanned settlements in Ghana. We cannot define structure to our housing that easily. [If it is available, shame on the development team for not prompting me after I generated the second address]. I don't think they have it set up simply because: a back-end clustering algorithm will fall short in
predicting that my brother's room and my room are in the same house.
What metric will be used? What data will the algorithm be trained on?
The algorithm will be a total mess in densely populated settlements! (I apologize for my zero effort in assembling my rhetorical questions and exclamations into an easily digestible argument).
Applying a different dimension of thought will suggest that, we can get information on locations which are roads, water bodies etc. [basically places we cannot logically reside in], and then overlay that data with the data we are getting from active Ghanaians , who are doing their patriotic duty of mapping out all inhabited locations, without any clear incentive to do so.
Asaasegps sorry, Ghana post gps will then rely on the data submitted to build up a map that will help us do the clustering. Going with this approach will mean that we're making the bold assumption that housing, road construction and project siting in Ghana has structure. And even worse, we can trust people with no clear incentive for them to do the right thing.
well said.
ReplyDeleteThanks! 👊🏽
Delete