Guess How Much The Pumpkin Weighs?

Posted on Posted in Books, KPIs, Learners

guess the weight of a big pumpkin

If you had to guess?

How would you guess the weight of this pumpkin?

James offers a smart estimate based on the height and circumference of the pumpkin. Sean thinks he has an idea based on the load capacity of the truck that’s carrying it. I am tempted to google the answer by finding the pumpkin at a past event. I don’t, but only because I feel pretty confident. I can estimate the weight. I don’t need to guess. The officiator, Ryan, accepts my guess: “the average of all guesses.”

After the submission deadline passes I find out I am not in the winners circle because Ryan forgot to calculate my answer. “Let me add you in.” He says, “But it is going to be high. There were a some really high guesses.” I wait for him to calculate the average.


“oh, Crap.”

At this point I know I’ve placed because Ryan says, “oh, crap.” The winners have already been chosen and I’ve bumped one out of the circle. The problem, Ryan already told him he won! I estimated the weight to within 10 lbs of the true weight!*


How to Guess the Weight of any Pumpkin.

In 1907, Sir Francis Galton asked villagers to guess the weight of an ox. None of them guessed correctly. When Galton averaged their guesses, he arrived at a near perfect estimate. This is the classic example of the “wisdom of crowds”, or collective intelligence. More recently, the show, Brain Games, demonstrates how to guess the number of gumballs:

guess the number of gumballs

Pumpkin. Ox. Or gumballs in a machine. The wisdom of crowds has it’s moments. Each person’s perspective holds true information plus some degree of error. The individual errors of each guess get rounded out; they offset each other. Whats left? An estimate of the truth. Everyone getting it wrong, gets it right. It’s the diversity of eccentric perspectives that makes it work!  If we were to plot the answers on a graph you might assume the scatter-plot forms a bell curve: i.e. a lot of people guess close to the right answer. Not necessarily (See below). For sure, it’s not required to. The average is also not popular opinion, that is the mode. In this case the average is the shared consensus we get when we honor our differences.

A Caveat: Averages Make for Poor KPIs.

It would be a mistake to rely on an average as a KPI. To benefit from the “wisdom of crowds” we need to meet certain requirements. These requirements are not likely to exist in the context of an SOP. When you use an average as a KPI you are assuming that  it is representative of the whole, and it isn’t.

I often cringe when I see averages. I’ve been working on a resource to help teams create better measurements. I’ve looked and haven’t found one.  The options that I see lack either sophistication or practicality. I haven’t found a resource with both. I’m creating a resource that will (1) show you how to think correctly about KPIs and (2) how to do it in Excel. I invite you to check out what I am doing. Sign up on my email list for updates.


*Ryan & Hee Haw Farms were good sports. We all won passes. My family saw the giant pumpkin drop.


7 thoughts on “Guess How Much The Pumpkin Weighs?

  1. I love the chart that shows the enormous range of guesses…each one offsetting another to arrive at the correct answer. Going off that logic, if we were to remove elected officials and instead used polling services to decide all policies…in theory we would arrive at the perfect legislation for each individual group. It wouldn’t achieve the best policy for each unique situation (the exceptions), but it would establish the perfect rule.

    1. Thats an interesting thought @Philip87:disqus , but arn’t policies just varying opinions on what is wright or how things should be done, can taking the average of an opinion really produce a correct answer? The reason the pumpkin or gumballs work is because there is a right answer, is there a right answer in policies or just opinions that think they are right?

      1. Good point for sure! It would be interesting to test it out…obviously those polling services would need several indicators. For example, on the topic of abortion it couldn’t just ask, “Are you pro-life or pro-choice?” It would have to combine that question with the lateness of the term, the circumstances, the education and resources provided, etc. When gathering the preferences of an entire group, could you actually achieve the most ethical outcome that pleases the greatest number of people? I think that with substantial information, it could achieve a better outcome than the current format, which is the interpretation of our preferences made by policymakers.

    2. It isn’t logic, it’s math but I get where you are coming from and what you are saying. Two issues with it. First, you can’t have yea or nay as the only options you need to be able to accept other options (and we need to quantify those options). Second, we need the polls to be independent. Ideas and opinions spread like viruses. Once a common idea has spread through the group you have correlated polls and the math will breakdown. *There is a good explanation on wikipedia (

  2. Rho, great article but whats the weight of the pumpkin? But on another note not a big fan of the title, Maybe The “wisdom of crowds” would be a good title (but this does come from someone whose never written anythings lol)

Leave a Reply

Your email address will not be published. Required fields are marked *