Debate

Reach for Debate before you commit to a plan, not after it has gone sideways.

• The call is expensive to undo. Architecture, security, data and positioning decisions have costly wrong answers.
• Nobody has argued for the option you are not taking. The loudest voice carries the plan, and everyone else reacts to it.
• The result has to be checkable later. A "the group agreed" conclusion leaves no record of what each option gave up.
• You are working alone. The plan has only been read by the person who wrote it.

A small call still gets a panel of three and a short matrix. Debate does not build anything and does not make the call. It shows you the cost of each path so the owner can decide.

Most decisions go to whoever argues hardest in the first meeting. The others react, the first idea gains weight by being first, and nobody is assigned to build the best case against it. So the case against the chosen path never gets written down. It surfaces later, after the work is half done, and teams call that a retro.

Debate settles a decision before anyone commits to it. It takes one decision, reads the notes about the work it touches, seats a small panel of expert views, and makes the panel argue. One seat always attacks the premise itself, which means it asks whether the decision should exist at all.

The output is a matrix. Every option gets a row, including the ones the panel threw out, and each row says what the option wins, where it breaks, what it risks, what it costs and when it is the right choice. A reader can check that line by line, which a verbal "the group agreed" never allows. Debate advises and never builds. You read the matrix, you make the call, and the decision goes into the record.

Seat The Panel

Three expert seats by default, and up to seven for a big call. One seat is the premise seat, and it asks three questions:

• What breaks if the outside service the answer depends on is down for four hours?
• What breaks first at ten times the load?
• Can we undo it without touching records?

The other seats come from an archetype bank. A data engineer, a security reviewer, an implementation lead, a strategist and a scope-cutter each see a different risk. An archetype is a lens for asking a question, not a replacement for someone who knows the system. When the cast cannot answer from real knowledge, the honest answer is "unknown, ask the person who knows."

Posture Follows The Decision

Converge means the panel ends on one path. Diverge means it ends on two live directions. The owner's explicit choice overrides the default.

Converging domains have a right answer and costly wrong ones. Two production systems is a cost nobody budgets, and a security gap has one correct fix. Diverging domains depend on the reader or on a judgment call, so forcing one answer hides the real choice.

Argue Every Option

In round one, each seat states a position. In round two, each seat attacks a rival position by name and gives the reason it fails. The path least likely to be taken gets argued anyway, so you hear its strongest version before you rule it out. Two rounds matter because a position stated once is a preference, while a position that survives a named attack has been tested.

The Decision Matrix

The matrix has six columns, in this order: Option, Wins, Breaks, Risk, Effort and Best when.

• Wins says what the option gets right.
• Breaks says exactly where it fails, in concrete terms such as a condition, a date, a volume or a dependency.
• Risk, Effort and Best when stay short. A hidden cost goes into the option's first row so it cannot be missed.

"Could be risky" cannot be checked, and a reader cannot decide against it. "Breaks when the second site's records do not fit the import" can be watched for. Wins and Breaks are mandatory on every row, including the options the panel rejected.

Resolve And Hand Off

Converge ends with one path and the one condition that would make the panel change its answer. Diverge ends with the two strongest surviving directions, presented as a live fork. Either way, the decision is the next move, and the owner makes it. A decision record is saved only when you ask for one, or when the call is about architecture, security, data or a high-impact product decision.

What Debate Does Not Do

Debate never builds anything and never edits the work under debate. If the panel says "agreed" with no matrix behind it, that is theater, and the owner should be told there is no record. Run Context Gap first to see whether something already exists. Debate runs before anyone builds, and Second Opinion reviews the work once it exists.