I was reading about the covid-19 briefing yesterday and this quote got me thinking:

“President Donald Trump warned Tuesday that the next two weeks will be “very painful” in the U.S. in terms of deaths from the novel coronavirus.”

That got me thinking about the title of this post, which are terms used when I was in software development, at least at companies I worked for. I have seen the same effect as a project manager in the non-IT world too.

Here’s the breakdown: when someone asks for an estimate of how long a project will take, it is a situation fraught with asymmetries it terms of information and incentives.

As the estimator (call them P1), I know a lot more about the facts than the person asking for the estimate (call then P2). However, P2 usually has a lot of power over the P1. So P1 has a high incentive to make P2 happy by giving them the shortest estimate possible.

If P1 really believe the project will take six months, but knows that will make P2 unhappy, they are not likely to give that estimate, unless P1 has rock-solid evidence to back it up.

However, if P1 gives an estimate that make P2 very happy but is in the very short-term, say less than a week, then P2 will be back in a day or two and expect the job to be done or at least want a status of how the work is progressing. If progress does not meet the estimates, P2 will then be very unhappy as well, so P1 does not want to give that estimate unless they are absolutely sure that they can deliver on time.

So what is an estimate that is long enough to allow P1 to get back to work on the project and not have P2 annoyingly coming back looking for updates or deliverables, but also short enough to have P2 go away happy?

You guessed it: Two weeks!

When I have heard any estimate in any context be “two weeks”, what us has invariably translated into is “I am not really sure, but if I tell you my 90% confidence estimate that will go badly for me, so I am telling you something that allows us both to withdraw from this discussion with what we can position as wins and get back to work.”

This practice is, in my experience, endemic in American business and it is worth being aware of it so as not to be part of the process. Reality will eventually intervene into all project plans and estimates done to make all parties feel like they have a win will need to be revised and brought into line with what actually needs to happen.

In fact, as a person getting estimates, a 3, 4, or 6-week (or longer) estimate should make you feel pretty good that is is as accurate as possible at the time, as the team would not give you this estimate unless they had done the work to vet it out, since they will be expecting to be challenged on it and will want backup. The same is true for very short estimates; no one is going to quote you a 2-day plan unless they are very sure they can deliver in two days.

On the other hand, any time you hear any estimate of “two weeks”, that should raise major red flags and you should definitely start to probe for supporting evidence that the estimate is based on good analysis and has a foundation in reality. Teams often use the “Goldilocks Estimate” to hide uncertainties or lack of rigor, figuring that they can close those gaps during the two weeks and then have their firmed-up estimate. But that thinking contravenes the assumption that the two weeks are being used to deliver the project, not do better estimates!

I am not suggesting that the President or the administration is engaged in this practice; but in the role of “American people as P2”, I would want to see a bit more support for the idea that after two weeks we will be beyond the worst of the virus. That is a possibility, but knowing as I do the prevalence of the “Goldilocks Estimate”, anytime I hear “two weeks” for anything, I start to ask for some backup to make sure I know what this estimate really is.