Wednesday, July 05, 2006

Parkinson's law

There is a concept called Parkinson's Law which simply states "Work expands so as to fill the time available for its completion". We've all seen this. Let us say, if you are required to do an assignment and submit it in 3 weeks, you'd do the same job that you'd do if you had to do it in 1 week. Somehow, if you are aware that you don't need to rush, you tend to take it easy till the last minute.

I hereby propose a modification, from now known as employment entropy theory, that states "work expands so as to keep employed the people allotted to it". If a team of 5 people can do some work X in time T, then a team of 50 people would do the same work in same time (or less). The additional 45 would not be wasting time, and they'd be busy in their own right. But, the increase in entropy (loss of useful time now spent in co-ordinating, managing, dividing and then integrating the work of these people) would mean that the cumulative per-capita useful work is now less.

My reference is to the performance of large companies v/s small companies. Vista is delayed, WinFS has been aborted, IBM Lotus Notes uses 1980's UI, Google can't do great things beyond its search engine (to their credit, they atleast try) and we are talking about super-rich companies with super-brilliant people. Yet, they can't do things fast. How can we explain this ?

The problem relates to work environment and the recruitment of people. Recruitment methods don't scale up. Nothing that I've studied so far in HR (though it won't mean anything - I got a DD grade in HR) talks about mass-recruitment. Recruiting 3 people is not like recruiting 300 people. Simple probability means you do tend to choose a lot of bad apples (even if proportion remains the same, the absolute number goes up), and even if they are few, their attitudes can affect the good ones. This, in turn, affects the work environment and nobody can do real work, because for that you are dependent on approvals, authorizations, assistance from people who are not serious or interested. It starts the rot slowly, and by the broken window theory, at mind-numbing pace, it envelops the whole organization, till nobody inside is aware that they can do things faster.

In his legendary book, the The Mythical Man-Month, Frederick Brooks estimates that the difference between a poor programmer and a good one is in the order of 10. So, when there are a lot of people, all the people, good and average will start performing to their lowest levels of their competence, and naturally, work suffers. But, given the levels and levels in hierarchy, is anybody concerned ?

Having worked in a startup, where you get things just done, and now being in a large company, however temporarily, is a torture. I'm confident that most things that people estimate would take a year or so, could be done within a couple of months, if the team is small and motivated enough. That means getting the "real coders" to do the work, and ruthlessly chopping off the deadwood. My solution to this problem would be: use the smallest number of people enough to do the project.

Update: Just after I posted above, I got the information from the horse's mouth, as it were. There, Prof. Parkinson mentions something very hilarious: Officials make work for each other. He also analyzes the reasons and his article quite simply eclipses my post above :-)

3 Comments:

At Thursday 6 July 2006 at 1:51:00 PM IST, Anonymous Anonymous said...

Very nicely written..and it echoes a lot of my own thoughts
-Shruthi

 
At Friday 7 July 2006 at 1:36:00 PM IST, Blogger Sridhar Narasimhan said...

@shruthi:
heh, nice to know that we agree on *something* :)

 
At Friday 14 July 2006 at 9:55:00 AM IST, Blogger Aravindan said...

super post..

but u see, companies themselves wd not over allocate ppl for a job for cost reasons..but i think looking at ur exp, i got it totally wrong..

 

Post a Comment

Links to this post:

Create a Link

<< Home