Follow the Pointer
A Pointer Moves the Estimate
Suppose the claim is:
Jordan was at the meeting.
You start uncertain.
Then several independent, knowledgeable witnesses say they saw Jordan there.
That should move you.
Not necessarily all the way to certainty.
But toward:
Jordan was probably there.
That is a true pointer.
Now suppose timestamped footage places Jordan across town during the meeting.
That should move you too.
In the other direction.
Again, not because one fact has been granted magical authority.
Because the information actually bears on the proposition you are testing.
A pointer does not need to settle the question.
It needs to legitimately change your estimate.
Pointers Can Point Toward
A positive pointer makes the claim more plausible.
You are asking:
“Did Marcus send the payment?”
You find a bank record showing a transfer from Marcus's account for the correct amount at the expected time.
That points toward:
Yes.
Maybe there are still questions.
Was the account actually his?
Was the transfer reversed?
Did it go to the correct recipient?
Fine.
Those can be inspected separately.
But the bank record still gets to do its job.
It points toward the claim.
Pointers Can Point Away
A negative pointer makes the claim less plausible.
Suppose someone tells you:
“I called you five times yesterday.”
You check your phone records.
No incoming calls from that number.
No missed calls.
No voicemail.
That does not necessarily prove the person is lying.
Maybe they used a different number.
Maybe there was a technical problem.
Maybe you are looking at the wrong day.
But the absence of the expected calls legitimately moves the estimate away from:
“They called me five times yesterday.”
You do not need proof before information is allowed to count.
And Sometimes the Pointer Is Neutral
Some information does not move the external proposition at all.
Remember:
P is true.
Observation:
I desperately want P to be true.
Does your desire make P more likely?
No.
Less likely?
Also no.
Toward the proposition itself, the pointer is neutral.
That does not make the information useless.
As we already saw, it may indicate motivated reasoning.
Different destination.
Different job.
Neutral toward one question does not mean meaningless everywhere.
A Pointer Is Not a Verdict
This matters because brains love dramatic conclusions.
You find one piece of supporting evidence.
“PROVED.”
You find one piece of conflicting evidence.
“DISPROVED.”
Easy there, courtroom thunder machine.
A pointer changes the model.
It does not automatically finish it.
Several independent witnesses placing Jordan at the meeting is meaningful evidence.
Timestamped footage placing Jordan across town is meaningful evidence.
If you have both?
Now you have a problem worth inspecting.
Maybe the footage timestamp is wrong.
Maybe the witnesses are mistaken.
Maybe there were two meetings.
Maybe Jordan traveled between locations.
Maybe somebody has been cloned and we have developed a much more interesting website.
The point is not to invent excuses for every inconvenient fact.
The point is that evidence accumulates.
A pointer gets a vote.
It does not seize the government.
Strong Pointers and Weak Pointers
Not every pointer should move you the same amount.
Suppose you are asking:
“Did my friend attend the concert?”
Evidence A:
A stranger says, “I think I saw someone who looked like him.”
Evidence B:
Your friend appears clearly in a timestamped photo taken inside the venue.
Both point toward attendance.
They do not carry the same weight.
The first might move you a little.
The second should probably move you a lot more.
For now, you do not need an exact numerical update.
No:
“Photo evidence has increased confidence by precisely 23.8 millifucks.”
We have already decorated that grave.
You only need the simpler habit:
How much should this fact reasonably move me?
Ask Whether the Evidence Would Look Different if the Claim Were False
Here is a useful rough check.
Suppose the claim is:
“My partner forgot our anniversary.”
Observation:
They apologize and say:
“I completely forgot.”
Does that point toward forgetting?
Somewhat.
Now imagine the claim were false and they had deliberately ignored the anniversary.
Could they still say:
“I forgot”?
Yes.
So the statement alone may not be an extremely strong pointer.
Now suppose you learn they had accidentally scheduled a work trip over the date months earlier, had not mentioned the anniversary in recent conversations, and looked genuinely shocked when reminded.
Those facts may collectively move the estimate further.
You are asking:
Would I expect to see this evidence more often if the claim were true than if it were false?
You do not need to calculate anything.
Just notice whether the evidence actually discriminates between the possibilities.
Do Not Demand Proof Before Updating
There is an opposite failure mode:
“It does not prove anything.”
Therefore:
“I will not change my mind at all.”
No.
Evidence does not need to prove a conclusion before it is allowed to move you.
Suppose there is a 40 percent chance of rain.
Then dark clouds roll in.
The clouds do not prove rain.
You may still reasonably think rain is now more likely.
Mental models work the same way.
A pointer can be incomplete.
Ambiguous.
Imperfect.
Still useful.
The question is not:
Does this settle everything?
It is:
Does this legitimately move the estimate?
Do Not Force Every Fact to Move You
But do not update just because a fact feels dramatic.
Suppose someone is accused of cheating and becomes furious.
That fact may feel important.
It is important.
But toward:
“Did this person cheat?”
anger may not move the estimate much at all.
Guilty people can become angry.
Innocent people can become extremely fucking angry when falsely accused.
If the observation does not discriminate well between the possibilities, it should not get a large vote on that proposition.
Emotionally vivid is not the same as evidentially powerful.
Let the Pointer Do Exactly Its Job
When you encounter a piece of evidence, ask:
What proposition am I evaluating?
Does this fact point toward it?
Away from it?
Or neither?
How strongly?
What other explanations still fit?
And most importantly:
How much should this actually move me?
Not:
Does this prove it?
Not:
Can I dismiss it because it is imperfect?
Just:
What direction does the arrow point, and how much weight has it earned?
A pointer is not a verdict.
It is a reason to update.