People search for, favour and recall information that confirms what they already believe, which is why the evidence against has to be a required field.
Why it matters when the plan changes
Confirmation bias is not a failure of effort. The tendency to search for, interpret, favour and recall information that supports what is already believed operates whether or not someone is being careful, which is why a working judgement will still weight supporting material more heavily and interrogate it less. The only reliable counter is structural: make recording what argues against the finding part of producing the finding itself, built into the evidence for and against that every claim carries, so the search has to happen rather than depending on memory.
The tension is presentational. A judgement that states its own weaknesses reads as less certain than one that omits them, and the less certain document is the more trustworthy one, because a forecast that can later be scored against its outcome depends on what argued against it having been recorded at the time, not reconstructed afterward. That trade is real, it is paid in the room, and anything that lets it be quietly avoided will be.
In practice
A finding says a dependency is unowned, supported by two documents and an interview. The counter-evidence field records that one manager describes an informal arrangement covering it. Both go into the recommendation. The leadership team can now decide whether an informal arrangement is enough, which is the actual question.
Evidence
People tend to search for, interpret, favour and recall information in ways that support what they already believe.
Confirmation bias, Wikipedia (2026)Forecasts that are recorded and later compared with outcomes can be scored, which requires that what argued against them was recorded too.
Superforecasting and the Good Judgment Project (2026)
What it cannot tell you
Recording counter-evidence tells you a search happened, not that the search was thorough or that the two sides were weighed correctly. It does not resolve which side is stronger, and it cannot substitute for the separate discipline of naming what could not be seen at all, which is a blind spot, not disconfirming evidence.
Questions
Because confirmation bias, defined in 2026 as the tendency to search for, interpret, favour and recall information that supports existing belief, operates by default. Making counter-evidence required forces the search to happen while the finding is still being made, the only point at which it can change the outcome.
No. The Good Judgment Project scores forecasts using Brier scores, which only works because what argued against a call was recorded alongside what supported it. A recommendation that hides its counter-evidence cannot be checked against its outcome, and an unchecked recommendation is the weaker one.
That is recorded as none found, with what was looked for. A blank field and an explicit none are different claims: the first says nothing about whether anyone looked, and the second is a statement that can be tested by someone who disagrees.
Anything that would change the finding if it held. That is broader than material which directly contradicts it, and it includes alternative explanations for the same observation, which is usually where the strongest objection to an organisational finding actually turns out to sit.
Counter-evidence is what was found and argues the other way. A blind spot is what could not be seen at all. Both are required parts of a judgement, and confusing them hides the difference between a considered disagreement and an absence of information.