"I'm right more often, so it's efficient"

Some readers will grant everything so far and hold onto a last piece of arithmetic. Fine, certainty isn't evidence. But you keep a record, and the record says you're right more often than not. Maybe seven times in ten. So deferring to your judgement is efficient, and the arguing is just the cost of getting to the right answer faster.

Run the arithmetic all the way through. Seven in ten means three in ten where you were wrong, and think about what those three looked like. They were the arguments where you were certain, produced the evidence, and the evidence said Wednesday. Or the memory disputes with no evidence, where you're counting yourself right because nobody could prove otherwise. Those three aren't the cheap ones. They're the arguments that ran longest, because your certainty was full and your record was on the line, and they're the ones your partner remembers, because they were the ones where the calendar turned round.

Then price the seven. In each, you were right about a fact that, by chapter 19's second question, mostly didn't matter, and the price was the evening. Seven evenings for seven plumbers' days. Efficiency is a ratio of what you got to what you spent, and what you got was a date and what you spent was the hour after dinner, seven times. That's not efficient. It's expensive, and the expense doesn't show up in the record because the record only counts the fact.

There's a cheaper way to be right about facts, and it's the one chapter 21 gives for anything with a record. "Let's check later." The calendar is free. It takes four seconds. The argument that precedes the calendar is the expensive part, and it adds nothing to the calendar's accuracy. If efficiency were the aim, you'd go straight to the check every time, without the campaign, and say nothing when it came in. That you don't is a sign that the campaign was the point, and the campaign has nothing to do with being right. It's about being seen to be.

And there's the seventy per cent itself, which is a number produced by the person it flatters. Pronin's bias blind spot applies to records as well as to judgements. A person who's argued to be right for years has a memory of the wins and a thinner memory of the losses, because the losses ended with "whatever" and the wins ended with a calendar held up, and a calendar held up is a more memorable thing. The record is as reliable as the certainty was, and chapter 3 is about that.