A Critical Examination of Evidence-Based Scheduling
As part of the class I’m taking on software process, I was required to write a short paper on a course-related topic. The topic I chose was Joel Spolsky’s Evidence-Based Scheduling (EBS) technique.
I approached the topic with a critical eye toward its aspects, strengths, and weaknesses. Hopefully, this could serve as an unbiased reference for someone interested in the topic of task scheduling, EBS, and Monte Carlo style simulations. Keep in mind that this paper was not written to be submitted to a journal publication and, therefore, is more of an expert opinion on the subject, rather than a definitive piece of research.
The paper is available online and as a PDF.
Note: The online version has a few formatting problems, for which I blame Microsoft Word, since I let it do the conversion. The PDF version is the truest rendering of the work.
Category: Software | Tags: Auburn, EBS, Evidence Based Scheduling, paper, PhD, school, Software Process 2 comments »
December 11th, 2009 at 1:06 am
Russell,
the paper seems ok as far as a review of EBS goes, but doesn’t put EBS into context alongside other alternatives used for traditional plan-based software development. Also, I just skimmed this, but there didn’t seem to be evidence about how well it stood up to other solutions on offer, which would be interesting to see. Lastly, what happens when the estimate is wrong? Sorry, couldn’t resist, as tend toward agile, lean and evo approaches in this stuff.
December 13th, 2009 at 8:56 am
Bruce,
Good points. You’re right that the paper doesn’t compare EBS to other alternatives. I tried to focus on how well EBS might fit into any arbitrary organization. As for how EBS handles being wrong… In theory, it self-adjusts for future projects but does nothing for the current project. Here’s a place that I disagree with Joel. He says in his articles that a block of padding is added to the project to account for stuff like this. I think this is too loose an approach.