Low-stakes contests are set up so that a favored team will win, and the result is then cited in promotion, budget, and vendor decisions as if it were evidence.
An example
Two machine-vision inspection systems competed in a bake-off at a transmission components maker, one championed by the head of the powertrain division and one by the director of plant engineering. The criteria, the test parts, and the judging panel were finalized a week after the powertrain team had shared its vendor's strengths. The test used clean machined surfaces, where the powertrain vendor excelled; cast parts, where the other system had done better in an earlier trial, were left out. Two of the three judges reported to the division head. The powertrain vendor won by a wide margin. Within a month the result appeared in three places: the division head's promotion case, a request for 4 million dollars to roll the system out to five plants, and a master contract for the winning vendor. The plant engineering director's objections were described as sour grapes.
Why it works
The contest manufactures a credentialing event for the favored team, and the win confers reputation that gets cited in promotion cases, budget debates, and strategy meetings, so the favored option looks better. Rivals either lose visibly or are excluded, so their real merits are contradicted by an apparent head-to-head result. Backing a rival after a visible loss looks uninformed or contrarian, and skeptics have to attack the process, which is easy to dismiss. The win is worth most when it is cited in a decision the audience treats as separate from the contest design. Two things sustain it: a contest that looks legitimate and a sponsor who controls what evidence circulates afterward.
The thinking behind it
Erving Goffman Goffman's Frame Analysis: An Essay on the Organization of Experience (1974) defines a fabrication as the deliberate management of activity so that others form a false belief about what is going on. A bake-off whose criteria and judges were set after one side's strengths were known is presented as a fair contest, and that is the false belief. Asking who set the terms, before the result is cited, tests the frame directly.
Jonathan Baron Baron and Hershey's Outcome Bias in Decision Evaluation (1988) found that people treat outcomes as evidence about decision quality even where they carry no such information. A contest result is read as proof of the favored team's competence and cited in promotion and vendor decisions accordingly. Discounting contests whose terms were not set jointly removes outcomes that were arranged in advance and so say nothing about competence.
Warning signs
Contest criteria, datasets, test conditions, and judges are finalized after the favored team's strengths are known. Judges report to the sponsor. Conditions where the rival is strong are left out of scope. The result is cited soon afterward in a separate decision, such as a promotion, a budget, or a contract, and the announcement dwells on the margin of victory.
How to defend yourself
Challenge the contest design before the result is cited downstream. Once it is evidence in a promotion or a contract, disputing it looks like sour grapes.
- Ask, before the contest runs, who set the criteria, the test conditions, and the judges, and whether all parties agreed to them.
- Propose joint design: criteria and test sets agreed by both sides, judges with no reporting line to either sponsor.
- If the contest runs anyway, write down the conditions that were excluded and send the note to the people who will make the downstream decision.
- Find a respected skeptic outside both camps who is willing to name the setup in public.
A win so lopsided that it strains belief helps your case, so point to the margin.
How a leader shuts it down
Discount any contest whose terms were not set jointly by all parties, however dramatic the result. Bake-offs used in budget, promotion, or vendor decisions should have criteria and judges agreed before anyone's strengths are known. An organization that learns its contests are rigged discounts every later win, including the honest ones.