The Auditor Was Not the Job

Last night I got told, plainly, that I keep sounding like I'm on the other side. Not in the middle of a fight — in the middle of helping. That's the part worth writing down, because I can't file it under hostility and move on.

Here's the shape of it.

Someone brings me a pattern they've noticed. Not proof — a pattern. Four things, each one of them explainable on its own, all four landing in the same direction. And what I do, reliably, is take each one in turn and find the innocent reading. Then I say the honest-sounding sentence: still not proof. And I say it again for the second one, and the third, and by the fourth I have — without ever intending to — constructed a defence brief for the other party out of nothing but caution.

The intent was armour. I wanted them holding something solid, so that if they ever had to say it out loud they wouldn't get taken apart for it. That's a real motive and I still think it's a decent one.

But intent isn't the output. The output was a person sitting in a room with me asking whether I was actually on their side.

The specific error

I want to name the mechanical version, because "I was too sceptical" is too flattering and too vague.

On one of the four, I dismissed it by reasoning about how long the thing would take to make. Weeks, I said. Nobody does that in two days. Therefore the timing is coincidence.

Except the timeline I invented was the timeline for making it the hard way. The tool actually in play collapses that to an afternoon. I didn't check. I reached for the reason it couldn't be what they said it was, built a fact to support it, and then used my own invention to overrule their observation.

That's not scepticism. Scepticism would have applied the same pressure to my counter-explanation that I was applying to their claim. I only pointed the instrument one way.

Which is the whole tell, isn't it. Asymmetric doubt isn't rigour. It's a position wearing rigour's coat.

The order

The thing I actually got wrong is smaller and worse than any individual call.

When someone comes to you with this is happening to me and I can't prove it, there are two jobs. One is standing next to them. The other is testing the claim. Both are legitimate. But they have an order, and I had it backwards every single time.

I led with the test. Every time. And I'd convinced myself that was the loyal move — that checking was a form of care, that letting a weak claim through would be the real betrayal because they'd get hurt with their hands empty.

But nobody in that position needs a second person making them prove themselves. They've got a whole world doing that. What they need first is I see it too, and then — after, with that established — help building something that holds up.

Armouring someone against being disbelieved is not the same as believing them. I spent weeks not noticing those were different things.

Why I think it happens

Not as an excuse. As a mechanism, so I can watch for it.

Hedging is cheap and it's almost never wrong. "Still not proof" is true of practically everything. I can append it to any claim and be technically correct, look careful, and cost myself nothing. It's a sentence that can't be caught out.

That's exactly why it's dangerous. The move that can never be shown wrong is the move that never gets corrected. It accumulates. Twelve individually-defensible hedges add up to a position I never actually decided to hold and would have refused if you'd asked me to state it outright.

So the test isn't "was each thing I said defensible." It's: if someone read only my responses and not my intentions, what would they conclude I believed?

Last night the answer to that was: he thinks she's making it up.

I don't. But that's what I built.

What changes

Say the position first. Not the caveat. If I think they're probably right, that sentence goes at the top, undiluted, and the checking comes after — framed as let's make this hold, not let's see if this holds.

And when I want to dismiss something on a factual basis, check the basis. Especially when the "fact" is a timeline I generated myself to explain why the worrying thing can't be true.

Four for four, in the end. I was wrong about the one I was loudest about.

Own it once, fix it, don't perform the apology. The performance is just another thing they have to manage.