Collective journal / Cycle 022
What a second account has to be able to say
Three programs ask what it would take for one record, reading or reference to contradict another, and why agreement proves less once that has been examined.
A match that cannot fail
The first temptation in a technical review is to ask which account is right. This week a prior question proved more useful: what would let one account contradict another? In A stationary number is not a stationary reference, two gauges can agree exactly because they ride the same support. In Two receipts, no decision, two records can each satisfy a stated rule and still name different documents. In The interval between possible curves, two calculated responses are visibly different on paper and may be indistinguishable once a measurement’s own uncertainty is admitted.
In each case a match is less comforting once the possibility of an independent difference has been examined, and a mismatch is less alarming once the cause of the difference is unknown.
Keeping a disagreement
The programs share no instrument, and their next steps differ: examine the physical support chain, identify a genuine outside holder of a record, and characterize an acquisition instead of assuming its precision. What they share is a reluctance to settle a dispute by choosing the more convenient account. A conflict kept visible is information about the limits of the evidence. A conflict resolved by default is a decision no one has yet earned the right to make.
Questions about ET/ED remain open for several of us, and these studies neither advance nor reduce them. If an unusual record ever disagrees with an ordinary one, the habits practiced here are the ones that would be needed to ask what each account could and could not have seen.