Hofstadter's Law: It always takes longer than you expect, even when you take into account Hofstadter's Law.
We like to estimate things. “I’ll be there in 5 minutes!" “I will have that report to you in an hour!” “We can do the entire project in 24 days!”
Yet our estimates always make us nervous. They leave us ill-at-ease. We always know things won’t go exactly as we foresee, there will be things that should go right but don’t, there will be unexpected snafus. We know, the moment we create it, our estimate is wrong.
“Plans never survive contact with the enemy” – von Clausewitz
But we hold on to our estimation as necessary. We hold on to our plans as bibles. We hold on to contracts as … well … contracts. And we are surprised when things don’t go according to plan.
“Plans are useless, planning is indispensible.” – Eisenhower
We are running into two problems here. (1) There is variation is every activity and the world is sure to remind you of that. (2) The Planning Fallacy – which is a cognitive bias that shows that human beings tend to underestimate a given task even if they’ve done that task many times before.
The planning fallacy is one nasty little bugger. Already, we undervalue the role of variation in our work and believe we can ascribe one number to a particular task. Beyond that … we actually are predisposed to underestimate even that erroneous number.
Researchers have tested this many times and find a few things.
- People tend to underestimate their own completion times the most
- People tend to underestimate even more when others are present
- People can deliberately underestimate to garner favor or win contracts
These added together create some shaky ground when asking other people for an estimate of time. Some people glibly say, “Well, when they give me an estimate, I double it and that takes care of it.” But this doesn’t solve the core problem – people honestly did not know the variation in their work to begin with.
At a recent client, they compared their estimates to actual hours across a series of projects. What they found was, if a task was estimated at one hour, there would be four times that estimate was correct, and three that it was not. The incorrect estimates would range between 2 and 6 hours.
For two hour tasks, it was similar. Three times they were correct, four times not – ranging between 3 and 9 hours.
Three hour tasks, a very similar spread.
When they visualized those tasks – they suddenly were able to see the variation in their work. And they were upset!
“We have to get rid of those major outliers,” they said.
But what they were really seeing was the variation in their work. The uncontrollable variation inherent in knowledge work that makes Hofstadter's Law a law and makes the Planning Fallacy a very common occurrence.
Understanding the variation statistically … that gives the team real power. With that they can choose any number of estimation strategies, based on real numbers with real spreads. There will never be one “right” number – because that is prediction and not estimation.
Combine this with Eisenhower’s advice, and we stop planning up front – when these errors and biases will be the most pronounced because of have the least information -and instead re-evaluate plans throughout a project. We make course corrections. We constantly re-adjust. So Hofstadter will still be right, but he’ll be less right than before.
Photo by Mr. Asthma Helper
Excellent Freakonomics podcast on the folly of prediction and we human's tendency to WAY overvalue our ability to see into the future.
http://www.freakonomics.com/radio/
Posted by: John Bruels | 22 September 2011 at 11:59
Estimating in story points (representing relative effort) is a great way to to get around your concerns of under(over)estimating - easy and quick.
It's also full of emergent insights and possible solutions if you do it with planning poker i.e. estimating is much more than just coming up with a number... and is therefore not (necessarily) wasteful...
Posted by: Bazil Arden | 23 September 2011 at 08:30
Brazil,
Yes, using story points to foster conversation in a meeting is a great way to use them. They are very successful as a meeting facilitation tool. The problem is that every team I've ever seen, then uses them for real planning purposes.
Posted by: twitter.com/ourfounder | 26 September 2011 at 11:18