Tuesday, May 29, 2012

On Bazalgette @govuk

I chuckled when I saw a quote from Mike Bracken, speaking at a recent conference.  He is reported to have talked about:
The 'public service ethos' of Sir Joseph Bazalgette, who designed the main drainage and sewerage system for London in the 1850s, Bracken said that the Government Digital Service's (GDS) first focus was on meeting user needs, with making savings a secondary driver.
It's probably not often that Government's main online service delivery is compared with sewage transport mechanisms ... though the last time I'm aware of was when I made a similar link in September 2004 at an IDeA conference:

The picture in the top left corner is of the monument to Sir Joseph which is, obviously perhaps, on the Embankment in London.

Friday, May 18, 2012

If IT Is Broken ... Have We Got The Right Fix?

The main problem with Government IT, many will say, is that contracts (and so power, revenue and so on) is concentrated in a small number of very large vendors, mostly in single prime contracts. Even when they're not in single prime setups they are in dual prime (e.g. MoJ has Atos and Logica, Home Office has Atos and Fujitsu, HMRC has CapGemini and, wait, Fujitsu and so on). So the fix, we are told, is to no longer allow such single prime contracts and, as a consequence of deciding that, break the market open to new players. Job done.

A new model has been crowned that goes by the name of "towers". In another time, it might have been called best of breed. Essentially, several different suppliers are chosen, each of will be skilled in a specific tower - where those towers will be functions such as hosting, applications development, security, desktop support and so on (some tower models have seven such towers, some as many as thirteen). Such towers sometimes exist within the context of a single prime model but, this time, it is the contracting authority (i.e. the government department) who will own all of those contracts. They will, the thinking goes, have complete visibility of all of the prices and eliminate any "margin on margin" that results from the prime holding subcontracts. The result? A better deal will be had by government.

There is a very special kind of tower though. We might call it the sine qua non of the tower model. It's various called the service integration tower or the SIAM (Service Integration and Management). In effect, it's the old prime contractor (though not necessarily one of the companies that holds those contracts now) re-appearing as the company that manages all of the different towers, but doesn't actually hold the contracts.

If we know, then, that the prime contractor model was so broken that it needed to be replaced, are we sure, I wonder, that the new model is going to fix all of the problems inherent in the old world? Well, let's see:

- Lack of transparency ... fixed to some degree ... the contracting department will now see the cost of each of the components of the service. That doesn't mean that there is full transparency within each contract of course. Whilst everything is "open book" there are a million definitions of what that means.

- No margin on margin ... fixed to some degree ... the absence of a prime means that one layer of margin is gone, but there will be other layers within the towers (particularly where small companies are encouraged to play by coming in under the "safe" wing of a bigger company)

- Shorter contracts so more regular competition ... fixed to some degree ... Whilst G-Cloud encourages one year contracts (the framework forces such contracts though there is nothing to prevent renewal at the end of the period), the new models seem to encourage durations somewhere between 3 and 7 years. Better than 10 years but still gives plenty of room for prices to keep up (the regular iterations of G-Cloud services will provide very useful benchmarking though as is already being seen)

- Lack of innovation ... uncertain ... it's probably true that big IT companies struggle to bring new capabilities to bear but I suspect that it's equally true that government has struggled to adopt such interesting developments as became available. Splitting contracts into smaller chunks doesn't necessarily make them lighter weight and easier to change, but it might if government carries on with its plans to change the protective marking of data (and if the 50% of new spend via the public cloud promise is held on to, there is certainly scope for more innovation, but I don't see that as related, particularly to the towers model)

- Greater involvement of SME ... Uncertain ... Breaking what were very large contracts into smaller, shorter contracts should certainly allow new players into the game, but it's not clear if small players will make the cut. The recent PSN framework perhaps demonstrates that with only two small players involved - though that's still two more than before. Without a wholesale shift away from complexity towards commodity, small players will still struggle to navigate the arcane bureaucracy of most government contracts and so will likely need to shelter uner the wings of the bigger players for some time to come

- Better delivery ... Uncertain ... I guess we have to wait and see. There will, though, be several schools of thought. The large players will say that they can only deliver if they have control of everything; others will say that the more complex the interactions between the contracts, the harder it will be to deliver; still others will say that competitive tension between the suppliers and the knowledge that the contracts are much shorter (or that individual pieces of work can be competed) will improve performance all round. How the SIAM looks and works may turn out to be the key here - how they operate, influence and drive change could make all the difference to delivery (for better or worse)

- Lower risk or, at least, better and clearer risk transfer ... Also uncertain ... With many more moving parts and overall control resting with the customer, aided in some way yet to be determined by the SIAM, the risk picture certainly looks more complicated from the outset.

The net of that is that, in my view, it's unclear if this new model solves the problems of the old model. Much of whether it does will be in the detail of the contracts and the behaviour on the ground, by which time it's too late to do much about it until the end of the first contract term - and at least that is a shorter period than it has historically been.

What worries me most about the new model is not whether it fixes any problems of the old model but how it will actually be put together. There are three stages that need to be got through, each of which will be more challenging than the equivalents ever were in the old model.

1. Buying it all. With the prime model, there were many potential suppliers at the beginning, a few in the middle and 2-3 near the end. Negotiations completed with just one. With the new model there will be multiple, parallel, inter-dependent commercial negotiations underway. That will put a huge burden on the client side buying team. In the past those teams have been heavily supported by external parties; that may not be possible this time, although some will put the SIAM or an equivalent in place first to mitigate that problem. Of course, several parts of government will be doing this at the same time putting pressure on customers, suppliers and potential partners.

2. The transition. There have been relatively few changes of contract over the last ten years. HMRC moving from EDS to Cap is one, for instance. That was largely a one to one transition. With the new models there will, if the point of the model is realised, be multiple transitions to manage - staff will be parcelled up and moved to any one of perhaps a half dozen suppliers (some may go back to the customer), systems will move to any of several data centres, support for apps will move (supplier, location and perhaps even country) and so on. That's going to take a lot of management. Departments may say "that's what the SIAM is for". Suppliers may be giving work away for one contract at the same time as they are taking work on as a result of winning another contract. That could get interesting.

3. Running day to day. All of those moving parts, everyone looking at each other when a problem occurs, many pointing fingers away from themselves. How to diagnose a problem? Who moves first? Who pays service credits? Who proposes, funds and benefits from improvements? What happens in a crisis? All to be figured out. Again, some will give much of that role to the SIAM.

I used "SIAM" in each of those paragraphs deliberately. I get the feeling that it's the role that everyone thinks will fix the problems of the past. Yet whoever operates there will not have contractual leverage unless they are, actually, the client themselves (that is, the owner of the contracts). At the same time, the SIAM looks a lot like a prime, without the ability to take on / share / divest / pass back risk. It isn't, in my view, as simple as breaking up the contracts and creating a phantom integrator who somehow brings it together.

I wonder if the analsysis - and the sharing of understanding, lessons learned, best practice etc - is in place to support such a comprehensive and largely parallel implementation of the new model. It's going to take a lot of work from all parties to make it work and, even then, it may turn out to be no better than the old model in some ways. It may be worse in some, better in others. But we'll be in it, across the board, by then. So best to do all of the thinking now.