The Cone of Uncertainty

2008-11-26T07:36:00.008-06:00

I've been a technical project manager for over 3 years now, ever since getting out of college. And during that time I've managed a fair number of projects and been asked to provide estimates on timelines and labor for even more. Some of those projects came to fruition, some withered on the tree. However, most of the requests for estimates had two things in common: a very short (and often very vague) description of the problem statement, and a need to provide an estimate ridiculously quickly. I knew very well that any estimate provided in this fashion had a high degree of uncertainty. Even if I scoped down there request into something manageable and said, "here's what we can do and when" there would still be many unanswered questions (and thus room for error) simply because we hadn't had time to do an honest evaluation of the request.

A couple days ago I had the good fortune of talking with a colleague about how to provide better estimates on projects and he got me onto the concept of the The Cone of Uncertainty. Steven McConnell writes that in the initial concept phase of a project, any estimates created by skilled estimators can be inaccurate by a factor of up to 4x on the high or low side. Meaning if you estimate a project at taking one month to complete, it may require 4 months or 1 week. Estimates get continually more accurate as the product definition is approved, requirements are agreed upon, designs are completed, etc. Here's what that process looks like:

Now none of this is rocket science, mind you, but what's so helpful here is that it provides a solid, understandable explanation for why initial estimates have so much variability. And if a more accurate estimate with less variability is needed, then more time needs to be allocated for gathering requirements and answering questions.

I'm going to try out this concept when providing estimates for future projects. I haven't yet decided if I'll give a range (e.g. this will take anywhere between 1 week and 4 months) or just go straight to the top (this will take 4 months). I might start with the range simply because I'm sure I'll get some questions about why there's so much variability in the estimate and that will be a good opportunity to explain what we need to provide better estimates.

Define: AntiFantastic

2008-03-29T08:31:00.005-05:00

Back in the early 2000's I had a keen interest in the massively multiplayer online role-playing games (MMORPG) market. At that time the U.S. MMORPG market was dominated by the "big three" -- Ultima Online, EverQuest, and Asheron's Call. Sure, there were other games, but those were the three that were getting the most media attention and had the lion's share of the market.

Back then I was a bit younger, still in high school and college, and I had a lot more time on my hands. I spent a good portion of my youth playing Ultima Online and EverQuest and became increasingly interested in game design, particularly for these massively multiplayer games. Like other games, MMORPGs had their strengths and weaknesses, but over time EverQuest's weaknesses (such as the ones noted here) started drawing my ire. The game had so much potential, but they were doing so many things to make the game less than ideal. Eventually my frustration with the game grew to the point where I wanted to write a blog discussing MMORPG game design.

Like many of my ideas, the blog never came to fruition. However, I did manage to coin a term for EverQuest's "less than ideal" nature: antifantastic.

antifantastic
Main Entry: an·ti·fan·tas·tic
Function: adjective
: The opposite of being extraordinarily good or great