The stable and chaotic eras of UK Government IT

In the mid-2000's, I started my career in Accenture at their offices near the Old Bailey. There was a gold rush for government contracts at the time.

The NHS was working on a major project with Accenture called "Connecting for Health". There were many new IT projects kicking off, the labor government was investing heavily in transformation.

My first experience was with the Rural Payment Agency (a system to pay farmers). I came in to work as a test analyst with a bunch of other wet behind the ear graduates.

Technology as a second class citizen

Pretty soon after joining, it became apparent that Accenture viewed engineers as some kind of lower employee.

On the one hand, you had the consultants. They were on promotion rails. Up and out, cream rising to the top, etc.

Then you had engineers who were given a murky path. The only clear way to progress up the pay scale was to move into consulting and get on the more established path. Engineers who stuck around were given token management jobs.

Want to be a senior engineer? Manage a bunch of junior engineers. Often in completely different projects.

The pay was average but the pressure was often intense to get projects finished. The consultants needed their promotions after all.

There was also a huge cohort of engineers from India. We got to visit them in Bangalore which was quite an experience.

Heavy on the Oracle

Oracle products were everywhere. Oracle databases, workflow, order management. Stored procedures ruled everything. Understanding the ins and outs of PL/SQL became a must-have skill. A software stack at the time was Oracle UIs plugged into PL/SQL.

Every technical question had an Oracle answer. All the architects were Oracle consultants and the more Oracle you had, the better your system would be.

Some things went wrong

The NHS project became a 10 billion fiasco. Plenty of projects when up over budget. Communications between Indian development teams and consultants often broke down. Who thought that paying a farmer was so complicated? Did he grow nuts on his hill so he claims hill farm allowance and avail of the nuts scheme? Were they the right type of nuts? How big was this hill exactly?

Yeap, the rules were very complex. I'm not sure if it can be ascribed to an agile vs waterfall argument but the projects dragged on and over budget. The government became disillusioned. There was little in the way of software quality. Projects were stuck in the Oracle upgrade cycles.

I skipped off to investment banking for a while and left the government sector to try to make some money. Being a second-class tech citizen in Accenture, I quickly saw that Investment banking had a better pay rate so I set up shop as a contractor in London, plying my wares on Canary wharf.

Enter GDS

GDS came into being in 2011 but my contact with them started around 2013. At this time they were kicking off a series of exemplars.

GDS had some great tech chops. They were opinionated and they had power. They also could sniff out a weak technical argument like a fart in an elevator.

They cut through a lot of the nonsense arguments that plagued the government. And there was a lot of nonsense. Tech teams suffered from oversight that often crippled them. This could be architecture locking down which libraries could be used, security mandating crazy rules, release management processes that would take weeks to learn. Boom.

I was working as a contractor. In fact, everyone on the dev side was a contractor and pretty much everyone else was a civil servant. It worked really well. The idea was that we'd be eventually replaced by civil servants and we'd go on our merry way.

We could keep a hold of developer talent to make sure we could hire quality. I had a good tech architect (Lee Provoost who has since gone on to bigger and better things). We had access to the tools and tech we needed. Everything was fresh and new. There was movement and we delivered a project that has since won awards. It was great, for a while. Delivery ruled the roost.

No country for old devs

The system integrators got a bit hot under the collar. The outsourcing model had left them vulnerable to skill shortages. Cloud? DevOps? Microservices? They saw it originally as a fad but soon realized they were struggling.

I did get a chance to work with a System Integrator during this time and boy, they were poor. It was almost HelloWorld level of software development. They had been drained of all skills. I'm not sure if all the SIs got stuck in this trap but they were certainly losing out on contracts at this time.

System Integrators start turning up to the party

In 2018, all of a sudden the big System Integrators started showing up again. They got in through a few whispers here and there.

"We have bigger scale", "we can solve your resourcing problems". Their prime targets were a management that didn't quite understand what was going on (of which there were many).

Consultancies that popped up in the GDS era were now under price pressure from SIs. SIs started offering nearshoring and offshoring again. The contracts started moving back in that direction. The small contracting companies started disappearing.

DevOps, AWS, Kubernetes, Serverless started moving into the same place Oracle was in the mid-2000s. There is little thought given to what is appropriate or technically the most sound again.

"If AWS provides it, it must be the best" or something like that

Devs and DevOps are back to throw them in/stack 'em high, bums on seat mode. Everything has to be a managed service whether it's suitable or not. Engineering quality is not really at the forefront anymore. Service support, cost, and stability trump all else.

The stable and chaotic eras

I wonder if this is a standard pattern for innovation. After a stable Oracle era in the mid-2000s, we had chaotic project failures towards the end of the first decade. This gave rise to a new stable era where Cloud and innovation formed the next stable era. Who knows what the next era for Government IT might be?

Perhaps someone will bring it all back on-prem and into monoliths?

Maybe AI will generate our next wave of innovation and generate the complex task management systems that require so much effort to build and maintain.

Maybe I'll be paying my tax in Doge to an HMRC DAO?

One thing I'm sure of, that chaotic eras are always the most fun for engineers.


Jim Collins

Jim Collins