Theory 60/Averaging removes noise
Averaging Doesn't Cancel Bias
The claim
Averaging reduces random error but leaves systematic error untouched, so precision improves while accuracy stalls.
01Mechanism
Bias is common across observations and does not average out.
02Prediction
Confidence should improve faster than accuracy.
03Refutation
Bias reducing with sample size would remove the distinction.