Algorithms, pensions and desperation
- Steven Boykey Sidley
- 06 Mar 2017 01:20 (South Africa)
Once upon a time I was very closely associated with a company that developed advanced technologies in the cryptographic-payments space. I was a shareholder and director of this company. Our major competitor at that time was Net1 (then called Aplitec), now at the centre of the pension grant debacle. Net1 eventually purchased our company.
So I know something about the technologies that underlie the payment of pensions, all the way down to algorithms that secure the authentication of recipients and the transactions that ping-pong through the various system modules.
So I say this with complete confidence – there is no chance that the department will find an alternative equivalent solution by April 1, 2017.
Or, more alarmingly, 2018, and probably even 2019.
No one in government seems to have understood that the technological and logistical system than enables this payout system is about as complex a project undertaking as could be imagined, a very long time in the making and maturing. Not only in the maths underpinning the cryptography, but in the authentication of recipients in rural and telecoms-poor environments, the non repudiation of the transactions, the confidentiality of the databases, the logistics of cash management, the biometric technologies, and, and, and...
This is a massive, audacious and labyrinthine machine that has been built over decades and tens of thousands of man-development hours, and it works. There are no other South African organisations (and most certainly no government departments) who have the resources, skills, technologies and experience to do this, at least not for a long, long time. And even if a new system is designed now, it would certainly have to deploy technologies and skills from the US, Europe or Asia (unlike CPS’s UEPS technology, which was developed here).
Anyone in the chain of government who imagines that they can simply go out and hire another technology company to build this quickly is mistaken. Those who think that cash trucks can solve the problem display a shocking ignorance of even the most basic system flow.
Who are they going to pay?
How are those people authenticated as valid recipients?
How are they going to move from biometric to PIN, as has been suggested?
How long with it take?
Is PIN appropriate for people unfamiliar with either PINs or bank accounts? (There are millions of recipients without bank accounts.)
There is simply no Plan B on the table, not in the short or medium term.
I know Serge Belamant, the CEO of Net1, at least a little bit. I was as Wits with him, and have engaged with him a few times in a previous life. He is incredibly smart (he started as a technical guy – he understands the algorithms and their hardware means of expression at a bits and bytes level). He is a wily businessman – he managed to legally list his company in the US when everyone thought it was impossible. He is tough and uncompromising and occasionally abrasive. But the history of CPS and their BEE partners and Sassa is murky and largely inscrutable to me, and even after reading though the history (especially the fine investigative work of amaBhungane) I still do not understand the whys and wherefores and who-did-whats. But we all know that the courts pulled the contract from under this deal in 2014. CPS was bounced and, somewhat unsurprisingly, the government failed to plan (let alone execute) a replacement system.
But the moral culpability (or not) of CPS and other stakeholders is another matter. The real question is what is the right thing for CPS to do now. Belamant is faced with is a situation in which the government has come crawling back to to say – help. And he will. But given that the SA government is no longer his partner (and this would probably be a one-time thing), and he is not a charitable organisation, and he is responsible to his public shareholders, he is going to squeeze a flooding amount of blood from this dull government rock.
And it is not hard to understand why he would do so. DM