In Praise of the Silo by Neil Ward-Dutton
From the post:
Every organisational feature – including silos – is an outcome of some kind of optimisation. By talking about trying to destroy silos, we’re denying the sometimes very rational reasons behind their creation.
I’ve been travelling to conferences throughout North America and Europe a fair bit over the past quarter, and I’ve seen a lot of people present about business transformation, business architecture and BPM initiatives. One thing I’ve heard presenters talk about in a reasonable number of sessions (I estimate somewhere around 30%) is the need to ‘destroy silos’.
I have a background in architecture and integration, and for a long time I used to think the same. Silos are bad. Silos beget duplication; wheel-reinvention; contradiction; waste. Silos are really bad.
Except…
It turns out that ‘bad’ here really depends on your point of view. Silos aren’t actually ‘bad’, or ‘good’ for that matter. They’re optimisations – just as everything that every organisational, social or technical feature is an optimisation that serves one purpose or other. Silos are what happens when you optimise part of a business for expediency.
(…)
Another aspect of silos that should be mentioned is compartmentalizing information. Avoids the unlikely scenario where some sysadmin can copy masses of data that should not be commonly available.
Silos were reduced following 9/11 in the mania to “connect the dots.”
“Connecting more dots” isn’t a bad information strategy, as least as far as sound-bite strategies go.
The critical question of “who” was responsible for connecting “what” dots was left unanswered.
For example, the treasure trove of diplomatic cables leaked by Private Bradley Manning was just a data dump from the State Department.
Not really suitable for anything other than being a leak.
How do you connects “dots” with cables that run from catty remarks about personal appearance to reports about low level activities in Afghanistan?
Assuming meaningful access was solved, who is responsible for looking at the material to make the connections?
Having a large store of data you could look at but don’t, doesn’t solve the “connect the dots” problem.
Rather than reducing silos because it is another popular sound-bite information strategy, why not ask who needs access to the silo and for what?
What is the business case for creating mappings into and out of the silo?
And who will use the results of those mappings?
If you can’t answer those last two questions, you need to reconsider breaking silos that are otherwise serving some purpose unknown to you.