AI
New Ventures Now Have to Survive a Red Team First
By Khai Tran · · 1 min read
I sent Red an article on the CIA's red-team techniques and asked for the same thing for my ideas. The first test run sent an idea back for revision.
Week ending July 3. Written up from the build log.
I sent Red an article about how the CIA's red teams challenge their own analysts' conclusions, and asked for the same thing for my ideas. My exact words were "Build /red-cell now."
The result is a gauntlet with four rounds. List the assumptions the idea depends on, and how sure I really am about each. Imagine it already failed and work backward to why. Play a hostile competitor trying to beat it. Write its one-star review.
Every idea leaves with one of three verdicts: kill it, revise and run it again, or proceed.
It's written into Red's constitution now. Any new venture, offer or campaign, and anything that would take six months of my attention or $10,000, goes through the gauntlet before work starts. My answer when Red asked: "yes ratify it."
The first run was a test on an idea I had already put on hold. The verdict was to revise it and run it again.