Sunday, November 25, 2007

Conway's Law

I recently had a discussion with one of my colleagues, where I said that our employer would not be able to radically change the way its products were constructed unless it changed its organisation. "Why?" asked my colleague. "Because the design of a product reflects the organisation that builds it" said I. I then gave as one example, the first of our products to integrate design from both the UK and France; you can see the English Channel if you look at the layout of the chip.

I don't know where or when I acquired this belief - I don't believe I suddenly had this insight - on the other hand, I don't recall ever reading it, or anyone telling it to me. Whatever, I'm delighted to have discovered that the idea is not new - it is set out very clearly in a 1968 paper, "How Do Committees Invent?" written by Melvin E. Conway. In its third to last paragraph it says "Any organisation that designs a system (defined more broadly here than just an information system) will inevitably produce a design whose structure is a copy of the organisation's communication structure" - this is now know as "Conway's Law".

The paper is really good. It has several insights which are still true nearly 40 years after the paper was written. Apart from the law itself, I particularly like the section which explains why organisations design systems using many people when the system design would be better done with fewer. "Assumptions which may be adequate for peeling potatoes and erecting brick walls fail for designing system". Amen.

Thanks to Ycombinator Hacker News for the lead.


Anonymous said...

Great article Roger.
I particularly like the following quote:
First, the realization by the initial designers that the system will be large, together with certain pressures in their organization, make irresistible the temptation to assign too many people to a design effort.

and the one about many design disasters being due to large design teams with nothing to do.

It is really remarkable how accurate it is despite it's age and the fact that it is generalizing the problems of large system design.

I think the microelectronics industry is still young compared to, say the aircraft industry which started in 1900 and just finished the A380 last year.

Maybe we'll get it right in another 50 years.


Roger said...

Thanks for the comment.

Reading the paper made me think hard about various things that I might actually write a full blog post about. But in short, I misinterpreted "Mythical Man month" when I read it. I thought "What stupid people."; I didn't appreciate it was a profound insight into how the world *really* is.

Search This Blog