Danny Cohen wrote his proposal “On Holy Wars and a Plea For Peace” in 1980, hoping to put an end to some petty online bickering and adopt an important internet standard. That’s how long technologists have been arguing zealously with one another about the superiority of their preferred approaches. Some such conflicts drag on to this very day, frequently devolving into personal attacks.
Clearly, “holy wars” are a toxic threat to high-performance engineering teams, who must make smart decisions fast, and line up behind them fully, despite often strong opposing viewpoints within the team. To succeed, they must not only do all this, but do it often and well.
Furthermore, any one of these conflicts can appear as hard to settle as an actual religious war. Yet I have found a method that has worked for me repeatedly to get teams to do just that. I am eager to share it.
The secret is to address the fundamental reason that such conflicts persist ad infinitum–which is that humans are fundamentally irrational beings attempting to appear otherwise. (For more on the irrational nature of human decision making, I highly recommend the work of behavioural economist Dan Ariely.)
Put simply, one of the biggest problems is that, “People make up their minds and then make up their reasons.”
This is why the debate around the coffee maker or Slack channel never ends–because when you are committed to a course, you can always find a new reason or rebuttal, justifying both your position and your self-image of rationality.
It is therefore no coincidence that engineers, who have a reputation for rationality, are often most susceptible to such petty conflicts–for the opposition strikes not just at a principle, but a sense of self. This is precisely what someone disparaging one’s religious beliefs can feel like as well.
Externalisation is a key to objectivity. In the absence of documenting the reasons people cite for this-or-that approach, conversations loop back on themselves. However, conversations “documented” in endless threaded messages often become so verbose as to make the rationale on either side mind-numbingly difficult to contextualise.
The key to settling such arguments therefore is to do it comprehensively, all together, simply, and once. Here’s how it works.
Step one: gather the team in a room, and draw a 2×2 matrix on a whiteboard:
List the two decisions at top, and good/bad down the side as shown. (For those with a data science, statistics, or medical background, this is similar to a binary confusion matrix used to predict the sensitivity and specificity of a classifier, but used to represent thought experiments rather than ex-post results.)
I call this the “Speculation Articulation Matrix”. “Sam” for short.
It involves speculation in the form of thought experiments, articulated into a specialised form of decision matrix that aims to be comprehensive, hypothetical, and collaborative.
Sam’s purpose is threefold:
- To give everyone an opportunity to get all their thinking out and be heard
- To externalise all the considerations so that they can be addressed systematically
- To align the team on a course of action fully and finally
Step two: encourage everyone present to respond to a thought experiment: “We chose Option A, and it was a good idea. Why?” List every response in the upper-left box.
This is not a time for discussion or extensive elaboration, just for listing the reasons why, in simulated hindsight, this outcome could be true. Keep asking, “What else?” and wait until the room has gone quiet for at least 10-20 seconds before moving on to the next box.
Do this for all four boxes. What you will end up with may look something like this:
Step three: give everyone a minute to absorb what has been depicted. One time when I did this exercise, after a few seconds of silence, the most passionate advocate for “Option A” stood up suddenly and said, “OK, let’s do ‘Option B’!”
In other cases, it may take a bit more time and discussion.
Step four: add up the diagonals, for example in the case of the above diagram this would result in a score of seven for “Car 1” and four for “Car 2”. People may argue that this is too simplistic, that the points should be weighted, that some positive statements are duplicate opposites of negative statements, etc.
Bring them back to the fundamental purpose: to gather enough information to make a reasonable educated guess at the most probable best course of action, align on it, and move forward.
Sometimes, this exercise can be a bit like flipping coin, getting heads, and realising that you secretly wanted tails. If so, and the group aligns on a choice that does not weight heavily in terms of numerical arguments, but has some factor so compelling as to sway them toward it, go with that. The point is to go in “eyes wide open” to the considerations.
Step five: sometimes, this exercise may prompt the need for further research. If doing this research a) will go a long way toward tipping the decision one way or another and b) is fairly quick and easy to do, decide on what needs to be done, by whom, and set a date to reconvene.
Research should be used as a tie-breaker, not to endlessly spin out the discussion by arming one side or the other with further reasons to support their already-made-up minds. If the research won’t make a substantive difference to the overall picture, encourage the group to decide based on what is present on the whiteboard instead.
Step six: once the group has made a decision, get them to also agree that the phrases, “I told you so” and, “I knew this would happen” are not allowed. Any “told you so” argument has already been weighed, on balance and through a structured process, and the group has agreed to go forward in this direction together. Also, “I knew this would happen” is out–because this exercise is the time that anything you suspect could go wrong should be articulated. Sitting on a concern to use it later is sabotage.
So, no second-guessing the honest best efforts of the group mind. From here, we go as one.
Facilitating this approach well can take a bit of practice and time, and there are some subtleties to encouraging constructive conversation even when all the considerations have been articulated. That said, good group decision-making is one of the hallmarks of high-performing teams of all types, and therefore one of the most significant investments you can make in developing engineering team effectiveness.
Get it out. Get it all out. Look at it honestly. Decide. Then get on with it.
If only peace could be this easy everywhere.
If you use this process, I’d love to hear your results. If you would like help implementing this and other methods to develop team performance in your organisation, please get in touch.