The MPs are out shopping at John Lewis
I heard Gordon Brown on Radio 4 say that the reason previous attempts to improve the House of Commons had failed because they tried to achieve all-party consensus.
Browny said that by trying to involve and please all parties you ended up with the lowest common denominator of change. His big idea now is to appoint one, independant body with the powers of the statute book behind them, and they will enforce change on the House.
In my previous post Free the Gladiators! I was making the same point from a different angle. A funny line also suggest itself: Don’t let the inmates run the mad house!
If an organization is clearly broken, behaving in a way that is not expected of it, and failing to produce the results expected, then what are the chances of that organization fixing itself?
It is said that many MPs are blameless, do not have piggy snouts and haven’t been anywhere near a trough: but it only takes a few bad apples to spoil a whole batch. Is there a corrollary in IT, where even if some people work more efficiently, effectively and scientifically they can be brought down by poorly performing colleagues? Is this why IT staff give up before the start to improve things, because they can foresee the unlikely outcome because of a few bad eggs?
Is this why out-sourcing is used? My mess for less. Internal IT can’t get its act together, so I’m going to use a large hammer to crack this nut. The out-sourcer comes in with a cold business view, improves focused productivity and reduces cost. Non-critical projects (!) are canned. Heads roll. Process are replaced wholesale. Does this create value, though? What is value? Is it a return for monies employed? 
After a few years, when the IT tree has been cut back, it has a chance to grow again from smaller, more efficient roots: but what are the chances of the organization going bad again?
Is the root cause beneath the organizational structure, beneath the processes, and beneath the tools employed: is it about a culture and a mindset of good to great. Building core strengths and ruthlessly exploiting them, but what are these strengths for IT?
The Visible Ops authors (Kevin Behr, Gene Kim, George Spafford) all found out these answers via years of empirical research, and they summed up four steps that you need to do to be great – because these were the things that were consistently visible in the operations of high performing IT organizations, but not in “the rest”. These were:
- Stabilize the patient with operational controls, especially change management. A brilliant line in Visible Ops: “Change management is like the brakes on the car, that lets you go faster.” Old, tired change management doesn’t cut the mustard: a robust and innovative approach to change is required. Reward people who improve the efficiency and effectiveness of change, and punish those that break the rules.
- Catch & Release and Find Fragile Artifacts - operational controls (yes, again!), so you have a complete, accurate picture of your environment and you are tracking changes to configuration items and alerting on them. How many CMDBs do you have? Are they kept up to date manually (wrong)? What happens if a CI is changed: is there a check-and-balance?
- Establish a Repeatable Build Library – Release processes that are deeply integrated with your operational controls such that no, slow, error prone human is in the way and that you are 100% confident of what it actually is you are releasing, 100% confident of the success of that release because of your testing, and you have metrics across the whole lifecycle that make it easy to understand things like the capacity you are using, SLAs.. etc.
- Continuous Improvement. I talked about this in my post The VMM is Upside Down. For me this is critical, and the biggest difference between high performance and the rest – this is where a scientific approach to IT, where each change is like an experiment where you hypothesise about what you’re about to do, then you test it, then you measure the result, and prove the hypothesis or rework the hypothesis, the process, or the person doing the work. Invest more in the earlier part of release, rather than spending on heroics in production when it all goes wrong.
As I pointed out in Free the Gladiators!, this kind of change can’t be executed by one person or one team but , as Gordon Brown found, at the other end of the spectrum, working by consensus, might end up with very little change at all. All you need is for the change management team to refuse or take an age to alter their processes to adapt to virtualization (vMotion – RFC, routine maintenance?) and things slow down: when virtualization slows down, there is a direct $$$ impact because the more you virtualize, the more money you save – and the less you do, the more money you have to spend. So what’s the sweet-spot? Is it a SWAT team? If it is, is it internal or external: an internal team didn’t work for MPs.
Out-sourcing is one way to crack the consensus nut: in effect, it is the CIO or higher stating that the IT culture has to change and then, secondly, using an outside, unemotional and non-religious agency to effect that change.
Whilst Gordon Brown hasn’t gone as far as outsourcing the whole of parliament, he is applying the Visible Ops approach and has outsourced the governance because the British public is demanding their mess for less
No related posts.
I should hope that IT isn’t as bad as politicians, who can only retain their jobs by making sure they look good!
Cultural change is hard, so hard, in fact, that it cannot be done by consensus. It has to be led by a single person or group, dedicated to the task and both willing and able to look beyond the cultural taboos to do what needs to be done. Cultural changes really shouldn’t be the first objective to be tackled — that’s too hard. Better yet, focus on rolling out process changes to lead people to the cultural change (all the while educating them about what’s coming).
I don’t know that outsourcing is the answer — bringing in new employees and making change their imperative might be good enough, as long as their is the C-level backing to make it happen.
Process and people, then, as you start to get results, you can take on the culture with those results as your ammunition.
I don’t think outsourcing is a panacea either… I also think the change is deeper than process: how can you change a process that doesn’t exist? OK, so you introduce a process: but how can you introduce a process to a group that is not rewarded for creating or maintaining them, who plain hate them? This is where the scientific approach comes in: turn critics into problem solvers, let them see the impact of what they do on the business, let them start by continuously improving their own processes and interactions with others – it doesn’t have to be a Big Event, just incremental but fundamentally different… I’m going to post an example for vSphere next. Thanks for the great comment, Johnathan!