The secret to cutting the costs of public sector computer systems

One of biggest problems facing the public sector is the question of how to make ICT work to bring productivity and savings. On the face of it, technology has advanced in so many ways, the potential is obvious: but experience shows that the public sector is not very good at using technology effectively. What has gone wrong?

I am going to propose that the reason the public sector fails in its use of ICT is that it acts within a system based on false premises. I hope I will engage your interest early by laying out those false premises. They are:

1. Procurement is an effective way of selecting the best supplier of a service.
2. Competitive tenders provide a competitive environment that force suppliers to deliver competitive prices.
3. A new computer system that delivers a single service to the UK population or the population of Scotland must be delivered by a single provider. To do anything else would be expensive and more complicated.
4. Strong governance is critical to ensuring that services are delivered in budget and on time.

Ineffective procurement

Procurement of new IT systems is different from many other forms of procurement. Creating IT systems is a creative task: not one of production and delivery. As such the procurement process is much like the process of recruiting a member of staff, choosing an architect for your new building or the artist who will produce a portrait of the Queen. We can look at previous work and assess it but there are no guarantees that the end result will be what we wanted.

However, the added, and very critical, complication is that government procurement is not, usually, about contracting the services of an individual: it is usually about contracting the services of a large company. So there is no guarantee that the individuals delivering past results will be the same individuals serving you. Indeed the service you receive may be subcontracted and delivered by a smaller company that the government would not have considered awarding the contract to because it had no known track record in the public sector.

As many people in the industry understand: at best, procurement is a good guess at which supplier might end up delivering on what is required. At worst, it is just a process that ensures that, out of a range of suppliers, the government can be seen to have been objective in making a random decision.

Competition in name only

Competitive tending sounds like competition—it has ‘competitive’ in the name! But how many large IT systems can you remember that were delivered on time and in budget?

It seems that the competition, which takes place at the point of tendering, is not a competition that determines the cost of delivery. In other words, the price of the end product changes after the contract has been awarded. What is worse is that we all know this. Somehow, we just choose to ignore this awkward truth.

Competition is only effective if it continues to be competitive not just once every five years, but year upon year, week upon week.

Competitive tenders cannot work if the resulting contracts give suppliers a guarantee of being able to provide a service for any substantive length of time. This is not a criticism of any supplier it is simply a fact about what competition is. Competition is competition: a contract for five years is not a competition!

A single service does not demand a single provider

Computer systems have evolved whilst the public sector was sleeping.

How many of us are aware that when we log onto Amazon to buy something we are not all logging onto the same computer? Are we aware that the computer we actually log onto may be in any number of countries in the world and that the computer system may have selected us to trial a new service which is different from the service being offered at the very same time to the person sitting beside us on another computer logging onto Amazon?

The age of the single centrally run mainframe computer are past. The computer industry has discovered that it is cheaper to run vast numbers of anonymous and relatively small desktop like computers than trying to develop and maintain highly complex single supercomputers. (Spot the irony!) The industry is learning to take big tasks and break them down into small numbers of small tasks which can then be farmed out to lots of computers and processors to carry out the work and report back. This system is far more robust because the failure of one computer system is a small setback that can be worked around whereas the failure of a dedicated mainframe may be a disaster. Indeed, modern distributed computer systems expect computers to fail.

So when we log onto Amazon, we are identified and allocated to a computer that deals with our requests. There is no reason why everyone in the UK or Scotland should have their requests serviced by the same computer, nor even the same service provider.

But we are wedded to the idea that centralisation and large scale delivery brings ‘efficiencies of scale’ and mass production. We still believe in the mainframe, even if it is a thing of the past!

But suppling the same service from multiple providers would be complicated and expensive

How could two different suppliers provide the same service to different groups of the public?

This question turns out to be as bizarre as the answer is familiar. How about if I asked, “How could two different plumbers fit gas central heating to different groups of the public?” Well of course they could!

So, why can’t two different providers provide the same IT based public service to different groups of the public? As soon as we see the reference to IT, we focus on the IT system and not on the service. We start to be convinced that the computer system has to be the same. But we don’t mind if two plumbers do their job differently as long as they do a good job and the system works. If one supplier does a bad job, we won’t use them again. It’s called competition!

Do we really believe that plumbers would be cheaper if there was one central plumbing company for everyone serviced by their own central call centre?

I understand that some people still believe in a non-privatised public sector run by people of integrity concerned to deliver a good service. But we are not talking about public servants here. We are talking about public services that are already delivered by the private sector.

For “governance,” substitute, “delay and cost”

The word ‘governance’ is all the rage. Wikipedia says, “Governance is the act of governing. It relates to decisions that define expectations, grant power, or verify performance”. It seems to me that most of the governance I hear about is none of these. It is a governance centred around avoiding doing the wrong thing. It reflects a politics more fearful of mistakes than inspired to deliver innovation and greatness.

I would go further and say that governance in the world of public sector ICT in Scotland is all about avoiding making mistakes.

Sadly we will never innovate if we don’t allow people to make mistakes. We may all grow old and die but the ‘governance’ we are talking about here will never give us innovation. I am minded to point people to read Tim Harford’s book “Adapt”. A book which presses home the critical importance of experimentation in innovation whilst reflecting on why this is so hard for governments.

Why are we fearful of mistakes? We are fearful of mistakes because the costs of mistakes are enormous. Why are the costs enormous? Because we believe in delivering centralised services from single providers for contracts lasting as long as ten years. Why do we do this? We do this, because we believe it is the only option.

It isn’t.