Picture this: you have a string of big projects, and all of them are expensive. All of them are late, and everyone is getting pretty fed up. You keep hearing stories of small, nimble organisations who, startup-like, produce miracles on a shoestring. And you look at them jealously, but no matter what you try, you just can't seem to get that kind of performance for yourself.
This is the problem with almost all development shops where there is Big-IT. Big-IT is characterised by big processes, big budgets, and very, very big projects. It is also characterised by big overruns, and significant budget-blow outs.
It is a problem which is magnified in Government. I say that knowing that doing so will likely get me into quite a bit of trouble. I do it anyway, because now is the first time in ages that there's a chance to change things.
Just why is it that Big-IT always fails to perform well when it comes to delivering big pieces of functionality?
Here's my view. If you want to build a small house, you can put down a concrete slab for a foundation and get started. If you want to build a skyscraper, you dig deep holes, put in very comprehensive foundations, and install significant infrastructures up and down the building to make sure nothing falls down. If you want to build an elevator to orbit, practically the whole building will be supporting infrastructure, and there'd going to be precious little space for anything useful for people.
In Big-IT, we almost always build elevators to orbit, especially in outsourced environments. Outsourcers love these things because all the governance, security, test, and process groupies allow them to sell way more hours than they'd be able to otherwise. They're hardly going to turn up for small projects, now, are they?
In Government, practically all IT projects are big. My argument, though, is that now is a good time to try something very different. Lets build lots of little projects without all the supporting infrastructures and administration first, and then worry about the infrastructure we must have after. Let us let projects get big only after we've proved that whatever-it-is works.
This is a tried approach, known as a Skunkworks. Government needs a Skunkworks, and it needs one badly.
We need one badly because the entrenched ways of doing things - which have worked reasonably well for decades - have stopped working now. It is no longer possible to ignore the fact that groups of citizens, some with practically no training, can build stuff in days that takes Government years and costs millions. Fact is, those millions simply aren't available any more.
A Skunkworks would let us try new things, and more particularly, let us try new ways to do new things. It would also let us fail cheaply and quickly.
This last will be the hardest thing for civil servants to accept, of course, because failure is a very, very dirty word. But quick and cheap failure is really spectacular success. Failures of this kind teach lessons, letting you move on to the right way of doing things more quickly.
My guess is a Skunkworks which celebrated rapid and cheap failure would have significant economic upside, actually. If even 10% of the things it tried worked and produced value, I rather think that those few things would pay many times over for all the failures that went before.
Not to mention that those things would happen quickly.
What's stopping us spooling up a Skunkworks? Nothing but the momentum which continues to carry us down the old path. It's inertia, but, as I said, we're at the dawn of something new. Personally, I'm confident that all manner of things which would have been difficult before will now become possible.
Technorati Tags: Government, skunkworks, failure