Red team

What it is

A red team is a group given the job of breaking a plan before anyone commits to it. It works from outside the plan: it argues the opposing case, plays the competitor or the regulator, and hunts for evidence the plan's own logic files away. The output is not a mood of caution. It is a short list of named ways the recommendation fails, each with the evidence behind it.

Where it comes from

Red teaming has no single author. The honest account has three strands.

The vocabulary comes from wargaming. Prussian Kriegsspiel, prototyped by Georg Leopold von Reisswitz in 1812 and adopted by the Prussian army in 1824, established the habit of staffing the enemy side and playing it to win. In American defence and intelligence writing, "blue" is your own side and "red" is the adversary. The 2009 US government handbook A Tradecraft Primer says analysts normally work from the position of the "blue" (friendly forces), while a red team "attempts to work in the environment of the hostile forces."

The written method comes from intelligence analysis. That same primer sets out two techniques which business usage now blends. Devil's Advocacy, at page 17, challenges a consensus by building the best case for an alternative. Red Team Analysis, at page 31, models how the adversary reasons, to prevent "mirror-imaging": the assumption that an opponent weighs a decision the way you would.

The institutions came later, and after failure. The US Army founded the University of Foreign Military and Cultural Studies at Fort Leavenworth in 2004 to train red teamers, after a Pentagon review of Iraq and Afghanistan found officers were not challenging their own assumptions. The Army defunded the school effective 1 October 2021. The standard cross-domain study is Micah Zenko's Red Team (2015), which follows red teams through the military, intelligence services, the NYPD and private firms.

What it corrects

Picture a group that has owned one decision for four months. They are not careless: they know the objections, and have answers to every objection they thought of. The defect is the shape of their search. Evidence arrives and is read for whether it fits the recommendation, and what fits is weighted more heavily than what does not. The primer calls this a dominant mind-set, and warns that it makes contradictory evidence likely to be dismissed or given too little weight.

Ordinary care does not repair this, because care is spent inside the same frame. Telling the authors to be rigorous buys more thorough work on the questions they already asked. Someone has to be paid to find the failure, and allowed to succeed.

How it works

1. Write the recommendation down, and beneath it the assumptions it needs to be true.
2. Appoint the challenger from outside the team that wrote the plan, and state the mandate: break it.
3. Choose the assumptions that look most exposed, stated or unstated, and grade the evidence for each as strong, thin or absent.
4. Play one specific adversary, competitor or regulator or lender, and write their move in their words.
5. Deliver one named finding: the mechanism by which the plan fails, and what it has priced too cheaply.
6. Answer it. Hold, revise or drop the plan, and record which evidence moved you.

Worked example

Millennium Challenge 2002 was a $250 million US joint exercise, run over three weeks in summer 2002 with 13,500 participants. Retired Marine Lieutenant General Paul Van Riper commanded the red force. Outmatched in technology, he refused to fight the war the blue force had prepared for. He moved orders by motorcycle courier and light signal to stay under blue surveillance, then struck first as the fleet came into range: missiles from ground launchers, commercial ships and low-flying aircraft, plus speedboats loaded with explosives. Nineteen ships were sunk, the aircraft carrier among them. "The whole thing was over in five, maybe ten minutes," Van Riper said. His finding was a mechanism, not a warning: cheap mass attack against a force that depended on its network.

The rest is the other half of the lesson. Controllers refloated the sunken ships, barred the red force from firing on V-22 Ospreys and C-130s, made it reposition its air defences, and ruled out chemical weapons. Van Riper stepped down six days in. Blue won an ending written in advance.

In a Business Case Weekly case

In the SoftBank case, the sharpest fork is 2016, when Masayoshi Son proposes $32 billion for Arm at a steep premium in the same year he raises a $100 billion fund. A red team at that table does not observe that the cheque is large. It picks the assumption the thesis needs, that Arm's position converts into the coming compute buildout on a schedule SoftBank can finance, and asks who would have to be wrong for that to hold. Then it plays the lender whose debt matures before the thesis pays. The case asks what governance you would demand in exchange; the red team shows which failure that governance has to catch.

In your answer

• "The strongest attack on my recommendation is …, and it works through …"
• "That attack succeeds if … is true, so the evidence I would need to see is …"
• "Playing the competitor, their first move would be …, which my plan answers only by …"
• "I proceed / I revise, because … is what changed my mind."

Common misuse

The counterfeit is the challenge session that raises only objections the plan already answers. From outside it looks identical: a meeting, a nominated sceptic, a list, a plan that survives. The test is whether the challenger was in a position to win. Ask what the red team was permitted to find, and whether any finding on its list would have changed the decision. If nothing could have, the ships were refloated before they sank. The second misuse is red teaming before a recommendation exists, where the challenge replaces the original thinking instead of testing it.

References

• Micah Zenko, "Millennium Challenge: The Real Story of a Corrupted Military Exercise and its Legacy", War on the Rocks, 5 November 2015. The worked example above, from the author of the standard book on red teams.
• US Government, A Tradecraft Primer: Structured Analytic Techniques for Improving Intelligence Analysis (March 2009). Devil's Advocacy at page 17, Red Team Analysis at page 31. The method itself, readable in an evening.
• Micah Zenko, Red Team: How to Succeed by Thinking Like the Enemy (Basic Books, 2015). Red teams that worked, and red teams that were neutralised.
• Bryce Hoffman, "U.S. Army Moves To Close Red Teaming University", 26 October 2020. What the Army built at Fort Leavenworth, and why it was defunded.