Friday, 5 October 2012

Politics versus Plan B

There is no more important place to get IT projects right than central government. Unfortunately, internal politics encourages the opposite.


This is a visual metaphor, with little or no relevance to the actual article.

Who’d be a prime minister? On one day it’s all “We Love You Dave/Gordon/Tony”, then the moment you’re under 35% in the opinion polls it’s a catalogue of everything your government’s doing wrong. This year it’s been the granny tax, pastygate, the petrol non-strike, G4S and the West Coast rail franchise to name a few. All that’s missing is a good old IT shambles. After all, the last government kept us busy with the lost child benefit discs and the ill-fated NHS system. Well, for all you restless journalists itching for a story, I recommend you keep an eye on the upcoming Universal Credit benefit system.

In case you’re not following UK politics on an hourly basis, Universal credit is a plan to merge a number of key benefits such as jobseekers’ allowance and tax credits into a single system. Benefits is a controversial issue right now, but this is a software testing blog, and the issue of interest is that this is all dependent on a new IT system being developed. Now, before I go any further, I must stress I don’t know anything about how this project is going. For all I know, it could be all going swimmingly. But what if it isn’t? There doesn’t seem to be any kind of Plan B ready if the project goes behind schedule. And if this happens, it won’t be the first, because I worked on the last IT project where that happened.

Which IT project, I hear you ask? Well, please don’t be too harsh, it wasn’t my idea, they made me do it, but – I did software testing for ID cards. Yes, those ID cards. Remember them?

Now, I could give you a blow-by-blow account of ID cards testing, and at some point I probably will. However, the one overriding lesson I learnt from that project is the dangers of mixing politics and software testing, especially when it’s such a controversial scheme. And I’m not just talking about party politics, but also internal politics. The result of this, the thing that made the project such a nightmare, was the Identity and Passport Service committing itself to a deadline it couldn’t meet.

I don't know why IPS signed up to such an unrealistic deadline, but I can guess. When you have a project as politically controversial as ID cards, no government wants to show any sign of weakness; they say it’s going ahead and say when it’s going ahead (preferably far enough ahead of the next election to make it difficult for a future government to reverse it). Senior civil servants, meanwhile, are eager to demonstrate a “can do” attitude and make promises on when a project can be delivered. Once a date is set, any slippage is politically toxic for both. The press and opposition will see it is a U-turn and hammer the government. The civil servants who promised delivery on time get it in the neck for failing to deliver. Politically, it’s far better to stick to the date you set. Plan B? There is no plan B.

But whilst this might have been a prudent political decision, it was a terrible IT decision.  For a system as complex as IT cards that you are making from scratch, there was no way of predicting when it will be ready. And, worse, I can only assume the people who set these deadlines didn’t properly understand IT projects, because the moment I saw the timescale intended for testing I could tell it wasn’t realistic. Inevitably, the project descended into what is affectionately known as a Death March. Slippages in programming were compensated with cuts in testing. With the delivery date set in stone, the only way to meet to deadline was to declare the system ready to go, when the reality was quite different.

Fast forward to 2012, and I fear that not everyone has learnt lessons. In fairness, when Iain Duncan-Smith was questioned by a Select Committee last month,[1] he did try to explain this process wasn’t an October 2013 “big bang” and it would instead be introduced in stages. But when a Downing Street spokesman was asked about whether there’s any possibility of the October 2013 start date could be allowed to slip, the response was simply, I quote, “It's on track to be implemented in that timetable.” Which, at the risk of bringing up the same analogy again, is like responding to the question on lifeboat capacity with “The Titanic is unsinkable.”

In IPS’s defence, when the beta-quality system ID cards system went live, they did at least have the sense to keep the flow of customers manageable. Rather than open the floodgates on day one with the first 100 enrolments, they took in a few at a time, allowed for lost time with the inevitable bugs, and only ramped up the intake as and when the system was stable enough to take it. I’m not sure whether such a safeguard will exist for universal credit. The ID cards system was a pilot system designed for, at the most, tens of thousands of records. Benefits records, however, run into millions. Are we looking at transferring, say, a million cases on the new system by December 2013, ready or not? Even if well-intentioned managers at the DWP are trying to learn lessons and keep the timescale sane, will they be leaned on by minsters or permanent secretaries to hurry up? At the moment, I can’t rule this out.

There is no greater enemy to software testing than politics, be it party politics, management politics or just plain office politics. If you want testers to do their jobs properly, you must be prepared for them to tell it as it is, even if it’s not what you want to hear. You won’t get this in a culture where everyone is expected to be “positive” (i.e. anyone who expresses concerns is ignored or told to shut up). The opposition and press would do well to keep their eye on the Universal Credit system and keep their laptops poised. The government and civil service would do well to realise that it always pays off to have a Plan B.

[1] Actually, I don’t think it should have been Iain Duncan-Smith answering IT queries at all. Government ministers can only be as accurate on technical matters as the information they’ve already been briefed on. I’d much rather that on important IT matters, Select Committees directly questioned the people responsible for the IT, rather than use the minister as a go-between. This, of course, relies on the IT people answering questions honestly, which won’t happen if they’re worried about speaking out of line with their department. It could require a big culture change if politicians ever get a chance of knowing what’s really going on in flagship government projects.

No comments:

Post a comment