The Planning Fallacy
Posted by Rick Ellis on March 23, 2008
A universal phenomenon among humans is that we are terrible at estimating how long a project will take to complete. Cognitive psychologists call this “The Planning Fallacy”, and the more complex or numerous the steps involved, the more likely that we will be wildly wrong.
The problem has to do with the way we visualize projects, focusing only on the successful steps needed to complete the task. We visualize optimistically, rarely taking into account all of the things that can, and will, cause each step of the process to slow down.
Studies have found that people almost always underestimate, even when researchers ask them for a worse-case scenario. In fact, the accuracy tends to be about the same whether we plan with great detail, or just estimate a date off the top of our head.
For those of us in creative fields, like writing software or designing websites, estimating how long a project will take is even more difficult because there is an exploratory nature to these tasks. We’re not building widgets on an assembly line, we’re working through a series of challenges.
Sure, we might know what we want to accomplish, we might even have a pretty good idea of how we intend to accomplish it, but along the way there are endless smaller challenges to solve, each overlapping in different ways, and bringing new problems, ideas, and opportunities. It’s simply impossible to envision how a project will evolve until we’re in the thick of it.
Some people have suggested rules for estimating projects, like coming up with a timeframe, then doubling it. That’s not a bad idea (especially with budgets), although it further reinforces just how terrible we are at estimating, and just how wrong we will be no matter what we say.
Why do I mention all this? To explain why we at EllisLab never publish a software release date until we are nearly finished with the project. In my experience, any date that is more then 30 days out will always be wrong.
That’s also why we don’t publish a road map that describes where we’re going with our applications. There is such a fluid and dynamic nature to technology that anything we plan for today will probably be irrelevant a year from now. We just do not have enough information today to plan correctly for tomorrow. Our development cycles, and the internet at large, are moving too fast.
I wish there were a methodology that could be used to meet goal dates with precision, but thus far, I’ve never found one. Until that day, we’ll have to stick to our policy of not disclosing released dates until we’re nearly finished with the project.
