Tuesday, October 15, 2013

Inverse Rule of Project Timelines

Over the last decade and half, I've participated in a variety of internal and external commercial OR projects across multiple industries, many of which involved competitive bids. These projects somehow always ended up being one of two types - either long and routine, or interesting but cruelly short. Every time I came across that exciting project full of nice OR work, the deadlines were killing. The duration of the project/pilot/proof-of-concept appeared to be inversely proportional to the degree-of-sophistication. It seemed quite puzzling at first, but there are some plausible explanations for it, and perhaps this phenomenon shows up in other STEM-area projects too.

If the project isn't novel, and requires some reinventing of the wheel, it's considered low-risk and delegated to the Rodney Dangerfields in the trenches. If it turns out to be something new and shiny, senior pros are brought in to unleash their deadly math modeling dance moves on the client: a bewildering Bangra of reformulations and theorems, culminating in the East Coast Shuffle: the final formulation will always be solvable to optimality as a DP (thanks to a west coast friend for this discovery). A final spin through OR-FX chartware, and the gobsmacked customer is humbled into signing, provided the price is right, and of course, resistance to publication by any journal is futile. Problem is, the senior pro clock is relatively expensive. To keep the bid competitive, the total cost is treated like a knapsack constraint, which makes total time the casualty.

Result? like cricket matches that end in a draw after five days, lets just say that OR is the winner here.

(Written in jest, and any resemblance to real for-profit firms is not just coincidental but highly unlikely given the suboptimality of the inverse rule)

No comments:

Post a Comment