Joachim Andersen

Accelerated Learning

Red Teaming: Why I Invite Someone to Attack My Plans

Two antique chess knights facing each other on a dark polished walnut chessboard at night, one in polished gold brass and the other in deep red lacquer with gold detailing
A plan is only as strong as the best attack it has survived.


Red Teaming is the practice of deliberately asking someone to attack my plan, argument or system as an opponent would, so that I find its weaknesses before reality does. Its simpler cousin is playing the devil’s advocate: arguing against an idea, not because I believe the opposite, but to test how well the idea holds up.

The problem it solves is that we can get blind to e.g. our own assumptions, or flawed logic. Also, once we have invested time in a plan, we typically see the reasons it will work much more clearly than the reasons it might fail. A group of people who agree with each other can make this even worse, because nobody wants to be the one who spoils the mood.

I have learned that a plan nobody has tried to break is a plan I don’t really understand yet. The criticism I invite early is almost always cheaper than the criticism I get later from customers, competitors or reality itself.

The essence of Red Teaming is to ask: “If someone smart wanted this to fail, how would they do it?”


Origin: The Devil’s Advocate

The term devil’s advocate comes from the Catholic Church. When a candidate was considered for sainthood, a canon lawyer known as the Promoter of the Faith, popularly the advocatus diaboli, was appointed to argue against the canonization. His job was to take a skeptical view of the candidate’s character, look for holes in the evidence, and argue that any miracles might have been fraudulent. The office was established in 1587 during the reign of Pope Sixtus V, and Pope John Paul II reduced its power and changed its role in 1983.

The term red team comes from the Cold War. In the early 1960s, the think tank RAND Corporation ran simulations for the United States military in which the red team represented the Soviet Union and the blue team represented the United States. Today, red teams are used in the military, in intelligence, in business and especially in cybersecurity, where they try to break into an organization’s own systems.


Key Concepts

  • An Assigned Opponent
    Someone is given explicit permission, and often the duty, to disagree, so criticism is expected and not personal.

  • Thinking Like the Adversary
    The red team tries to act as a real opponent or a hostile environment would, not as a polite colleague.

  • Independence
    A red team works best when it’s not controlled by the people whose plan it’s testing.

  • Authentic Dissent
    A person who genuinely disagrees often challenges a group more effectively than someone who is only playing a role.

  • Acting on Findings
    The value of red teaming lies in what changes afterwards, not in the exercise itself.


Some Examples

  • The Office of Saying the Opposite
    The 1973 Yom Kippur War nearly took Israel by surprise, despite ample evidence of an impending attack. Afterwards, Israel formed a group known as Ipcha Mistabra. Its duty was to always present a contrarian, unexpected or unorthodox analysis of intelligence reports, so that things would be less likely to be overlooked.

  • The War Game That Was Restarted
    In 2002, the United States military ran Millennium Challenge 2002, a large exercise involving 13,000 troops. The opposing red force, led by retired Marine Corps Lieutenant General Paul Van Riper, used unexpected tactics, such as motorcycle messengers and light signals instead of radio, and a pre-emptive missile attack sank sixteen blue warships. The exercise was suspended and restarted with the blue forces restored and the red force so constrained that, as some described it, the end was scripted. The lesson the red team found was nearly thrown away.

  • The Hired Intruders
    Many organizations hire security specialists to try to break into their own networks, buildings and routines, a practice known as penetration testing. Their findings are often humbling, because the weakest point is frequently not the technology but ordinary habits, such as a door that is held open for a stranger or a password that was never changed.

  • The Real Dissenter
    In 2001, Charlan Nemeth, Keith Brown and John Rogers compared groups that faced a person who genuinely disagreed with groups that faced a devil’s advocate playing a role. The authentic minority led to more and better solutions than all three forms of devil’s advocate they tested. Real disagreement, it seems, is hard to clone by role-playing. This is most likely due to e.g. reduced motivation to find valid opposing arguments, and limited ability to hold a true contrarian viewpoint.

  • The Plan I Asked Someone to Break
    In projects prior to laying out a plan, or prior to a presentation, I very often ask someone I trust very well who has the ability to hold different perspectives, to play devil’s advocate with me. This usually takes the form of a semi-structured conversation, and sometimes we use white boards as well. Usually, we get a very fruitful discussion that leads to several weaknesses being quickly identified, and then we discuss how to best fix or mitigate them. It most often leads to a better outcome, and a stronger confidence in the plan or presentation overall.


Applying Red Teaming

I don’t need a military exercise to use red teaming. This is the simple process I follow:

  1. Pick the Right Target
    I red team decisions that are important, hard to reverse or based on assumptions I haven’t tested.

  2. Choose People Outside the Plan
    I ask people who didn’t make the plan, and ideally someone who already doubts it.

  3. Give Real Permission
    I make it clear that finding flaws is the job, and that I will thank them for it.

  4. Ask How It Fails
    I ask them to describe how a competitor, a critic or bad luck would beat the plan, and I complement this with the pre-mortem I describe in Inversion.

  5. Decide What Changes
    I write down what I will change, monitor or accept as a risk, so the findings don’t just disappear.


Pitfalls and Limitations

  • A token devil’s advocate can make a group feel that it has considered the alternatives when it hasn’t. If everyone knows the role is an act, the arguments are easy to dismiss.

  • Red teams can be ignored or overruled, as Millennium Challenge 2002 shows. Without leaders who are willing to hear bad news, the exercise becomes theater.

  • A red team can be wrong too. Its job is to raise possibilities, not to have the final word, so I still have to judge which threats are realistic.

  • Constant opposition can wear people down. Used on every small decision, it slows things down and makes people defensive.

  • The evidence is mixed. A 1990 meta-analysis by Charles Schwenk found that devil’s advocacy tended to improve decisions compared with relying on an expert recommendation, but how well it works depends on how it’s carried out.


Related Mental Models

Playing devil’s advocate and Red Teaming reminds me that the best time to find a flaw is before it matters. By inviting someone to attack my plans, giving them real permission to do it, and acting on what they find, I can turn criticism into one of my strongest tools.



To learn more about mental models and philosophy,
check out: Wise: How to Attain a Deep Understanding of the World, History, Our Place in It, and How to Live Well



📚 Sources and References