The estimation theory behind the card game

Category: Sharing insights        

Table of contents

• Estimation
• Absolute VS relative estimation
• Complexity dimensions
• Consensus VS consent
• Alternative card decks
• From T-shirt Size to Story Point
• Monte Carlo simulation
• The workshop

Estimation

Teams estimate their work for different reasons. Many use estimates for forecasting purposes and many use them to have a conversation about the work at hand to create a shared understanding in the team. Lots of teams do not estimate at all in terms of putting a number on an item, and for good reasons.

Most Agile teams that estimate their work, apply "planning poker" with a Fibonacci card deck and come to consensus to pick a final number. When I co-created a brand-new set of estimation cards, and creating a workshop to release the card game, I had the chance to dive into the world of estimation and the endless other possibilities out there. This article describes some of those aspects.

Absolute VS relative estimation

When teaching estimation in my workshop, I ask participants to estimate the size of a country. The range of answers varies greatly, from way too small to immensely overestimated. When I ask participants how many times a certain country fits into another country, the numbers are much closer together.

The human brain is absolutely rubbish when it comes to absolute estimation. We are way better at comparing things. That is why many Agile teams use relative estimation to guess how much work or how complex something is.

Relative estimation has another great advantage, as Mike Cohn beautifully formulates in his quote:

"Because the estimate for each feature is made relative to the estimates for other features, it does not matter if our estimates are correct, a little incorrect, or a lot incorrect. What matters is that they are consistent." - Mike Cohn

Complexity dimensions

If a team operates in a complex domain, where the work is difficult to plan due to the many unknowns that are still to be discovered, estimation is based on four complexity dimensions:

1. (un)certainty of the outcome. To what extent do we know that the solution will solve the problem?
2. (un)familiarity of the context. How much do we know about the problem and the business process?
3. (un)clarity of the desired situation in relation to the current situation in use.
4. (in)dependence in development. Do we need actions from others? Have we done this before?

"To be uncertain is to be uncomfortable, but to be certain is to be ridiculous." - Chinese proverb

Consensus VS consent

The final result of a team estimation effort can be achieved through either consensus or consent:

• Consensus means everyone is in favor of a decision. There is agreement. In terms of estimating, the team keeps playing rounds of "planning poker" until everyone is on the same page.
• Consent only means that no one is actively against a decision. There is acceptance. In terms of estimating, the team usually plays two rounds of "planning poker" and at the end of the second round majority "wins".
• In both cases team members play a card in each round, which represents their estimate for the work item. Through discussions and refining, the work item is made clearer for everyone and possibly smaller, larger, simpler, or more extensive. The conversation focuses on the differences in individual estimates to come to a shared understanding of the item, and eventually one estimate.

Alternative card decks

Teams can think of a million ways to estimate their work. When applying "planning poker", card values can differ from numbers to cups of coffee to animal names, each with a different goal. A few examples:

Fibonacci

Card values: 1, 2, 3, 5, 8, 13, 21, 34, ...

This deck compensates for Weber’s law, which states that as something becomes larger, our ability to see differences decreases. With each higher value in this card deck, the difference between the values grows as well.

Who is the mole

Card values: Mosquito, Mouse, Mole.

This deck only contains three cards (small, medium, large) and is a simple way to facilitate a discussion in a team where there is a big difference in knowledge or experience between team members. As this deck asks for consensus, the discussions enable a learning opportunity for all team members (not only the knowledgeable ones), narrowing the knowledge gap between them.

Family day

Card values: Much less hassle, Less hassle, Just as much hassle, More hassle, Much more hassle.

This deck is based on a self-compiled and team-specific reference story, which is separate from the team’s work context. The reference story represents a typical family day with an average amount of hassle. The five card values require to compare the work item to this average amount of hassle. This determines whether it is less hassle or more hassle. In case of the latter, it would be wise to simplify the work item. The outcome is based on consent to prevent endless discussions about the family day reference story.

Electricity Hydrogen

Card values: Low voltage, Medium voltage, High voltage, Hydrogen.

This deck has domain specific values, where one of the values represents something that is much more complex compared to the others. This is to quickly identify work items that are too large. In this example Hydrogen is the outlier compared to Electricity. If one or more team members estimate the work item as Hydrogen, the team needs to simplify the item. Once all estimates are in the Electricity domain, team members look for consensus on Low, Medium or High voltage in terms of complexity.

Everything is 42

Card values: 42, not 42.

This deck is suitable for teams that believe estimation is a form of waste and strive to work with items that are approximately the same size. Estimates are based on a reference item with the right size, which in this case is 42, the meaning of life, the universe, and everything. If an item is not 42, it has to be made smaller, larger, simpler, or more extensive, until everyone agrees it is of the right size. This enables the team to forecast based on the number of work items instead of separate estimates. And people happen to be much better at counting things than at estimating things. Suitable for the #NoEstimates fan!

More on #NoEstimates: https://www.dhirubhai.net/pulse/noestimates-forecasting-minimal-estimating-jan-willem-rutten-auuje/

Other estimation approaches

From T-shirt Size to Story Point

If you want to quickly estimate your backlog in Story Points and spend as little time as possible doing it, this method might be for you!

https://www.dhirubhai.net/posts/svanrooden_scrum-sizing-productowner-activity-7163894062041800704-mu9Z

Monte Carlo simulation

If you want to do forecasting, without doing any estimation at all, check out this recording of a meetup with Prateek Singh who explains the use of cycle time and Monte Carlo simulation for probabilistic predictions.

https://www.youtube.com/watch?v=VTRqCMWK5L4

“Planning is everything. Plans are nothing.” - Field Marshal Helmuth Graf von Moltke

The workshop

I have created a workshop (link below) to let participants get familiar with different ways of estimating work and the notion that each approach has its own strengths, depending on the context of the team. Through different exercises and moments for discussion, participants gather insights on estimation in general and how each type of card deck can be applied within their team.

I hope the workshop inspires others to teach teams and organizations about the world of estimation. The workshop can easily be facilitated online as well. If you need help preparing or facilitating the workshop, or even designing a completely new one, feel free to get in touch. I am more than happy to help!

The workshop: https://www.dhirubhai.net/pulse/how-explore-new-ways-estimating-work-team-jan-willem-rutten-snf7e/