How Good an Estimator Are You?

Chapter 2 of Software Estimation: Demystifying the Black Art opens with a quiz designed to test your estimation abilities. It's an interesting exercise, so I thought everyone might like to give it a shot.

This is a companion discussion topic for the original blog entry at:

The trouble is the amount of knowledge a person has on any of these subjects will help in the estimation.

For instance the year of Alexanders the Greats birth is more a piece of history triva than estimation.

A real estimation test would be more like show a group of people a jar full of marbles and asking them to guess how many are in there. This would provide a much more level playing field.

My favorite estimation question is:

How many piano tuners work in NY?


The key to this kind of estimation is binary searching. Humans are much better at comparisons (is a blue whale more than 8 tons) than guessing.

Total length of the coastline of the Pacific Ocean? US coastline? Including or excluding Alaska? What about Hawaii? Or Mexico? North and South Pacific? East and West? Islands too? This one’s way too vague. Perhaps it’s meant to mimic really, really bad software requirements…

Pops: Total length of the coastline of the Pacific Ocean seems straightforward enough. If it needed to be qualified, like your examples, then the question would have had to be longer.


In retrospect, it does. Two factors contributed to my confusion: I was, for some reason thinking within the context of the US only; and, I missed the part of geography class where the borders of the oceans are defined – it’s not like there are fences or painted lines or anything. And do the borders with other oceans count as “coastline”?

I don’t mean to belabor the point. I bring it up because sometimes in the arena of estimating software, things aren’t spelled out because there is an assumption that the requirements are crystal clear, whereas if they are not clear and (nearly?) identical in the minds of all parties then they aren’t adequate.

One odd quirk - for some reason this post doesn’t show up in the “June 2006” archives (or July 2006), so I only saw it when it was linked to from part 2.

Hey there,

I only got two out of ten questions correct.

It’s a good book, I am a little impatient to get to the meat and bones, but it has been a pretty good read so far. Especially for just $26 used on Amazon.

Thanks for mentioning it as I bought it after reading your initial post on it.


Matthew Purdon

@Peter Bridger:

The whole point of the exercise is that you are not supposed to know what you are estimating. Many times in software development you are asked to estimate development time on a project in a field that you have little or no experience in.

I won’t give away the secret of course, but I am sure Jeff will explain it fully tomorrow. I estimate that your opinion will change.