Are Claims Evidence? A Bayesian Response to “Claims Are Not Evidence”

There is a slogan, repeated often enough to sound like common sense: “Claims are not evidence.” It has a certain rhetorical force. It sounds cautious, even scientific. But like many slogans, it trades precision for punch—and in doing so, it quietly misleads.

The philosopher’s task is not to be impressed by slogans, but to ask: what, exactly, would have to be true for this to be correct? And once we ask that question carefully, the slogan begins to unravel.

What Do We Mean by “Evidence”?

Let’s begin with a simple clarification.

In probability theory—specifically within Bayes’ theorem—we say that some piece of information E is evidence for a hypothesis H if it raises the probability of H.

Not proves it. Not guarantees it. Simply raises it.

Here is the formal rule:

P(H | E) > P(H)

If learning E makes H more likely than it was before, then E is evidence for H.

That’s it. No mysticism. No rhetoric. Just arithmetic.

A Simple Analogy

Suppose your friend says: “I bought a football yesterday.” Before hearing this, what do you think? Perhaps your friend buys one football per year. That gives roughly a 2% chance he bought one this week.

Once you hear his claim, does the probability remains 2%? . . . Of course not. Even if you’re cautious, and even if people sometimes lie, you now think it’s more likely than before. Perhaps much more likely. And that means, by definition: The claim is evidence.

Perhaps not conclusive evidence. Nor irresistible evidence. Yet, it is evidence nonetheless.

To deny the claim is evidence isn't being a bold skeptic. It simply a refusal to update one’s beliefs in light of new information.

When Evidence Is Weak (But Still Evidence)

Now consider a more exotic claim: “I bought a spaceship that can reach another galaxy in an hour.” Here, your prior probability is vanishingly small. The world, as you understand it, simply does not contain such things.

So when your friend says this, what happens? The probability increases—but only slightly. You move from “almost impossible” to “still almost impossible.” And yet: The probability did increase. Which means—even here—the claim is still evidence.

This is where the slogan fails most obviously. It confuses strength of evidence with existence of evidence. Weak evidence is still evidence. A whisper is still a sound.

Another Analogy (a diagnostic test)

To sharpen the point, consider a classic example from probability.

A disease affects 1 in 1,000,000 people. You take a test that is 99.9% accurate, and it comes back positive. Surely you have the disease? . . . Not so fast.

When you apply Bayesian reasoning, your probability rises—but only to about 1 in 1,000. Still unlikely. yet, before the test your chances were 1 in 1,000,000. Now, after the test your chances are 1 in 1,000. That is a thousand-fold increase.

So the test result is clearly evidence—even though it is not sufficient to justify belief.

This is the crucial distinction: Evidence need not be decisive. It only needs to be relevant.

No Evidence Works Alone

At this point, someone might object:

“But a claim only works because of background knowledge—about honesty, plausibility, and so on.”

Quite right. But this proves too much.

A fossil is only evidence of prehistoric life because of background knowledge about geology, biology, and time. A fingerprint is only evidence because of background knowledge about identity and uniqueness. Remove all background knowledge, and nothing counts as evidence.

So if the standard is:

“It only counts if it works all by itself,”

then nothing counts at all.

The slogan collapses into self-defeat.

A Subtle Confusion: Content vs. Act

There is, however, a more charitable interpretation of the slogan.

Perhaps what is meant is this:

“The content of a claim is not evidence for itself.”

That is true. The statement “I bought a football” does not prove itself simply by existing as a sentence. But that is not what is at issue.

The real question is whether the fact that someone made the claim counts as evidence.

And here the answer is plainly yes.

The act of assertion carries information:

• People usually believe what they say

• People are often truthful

• People have access to their own actions

These are not certainties—but they are probabilities. And probabilities are precisely what Bayesian reasoning measures.

Why This Matters

This is not merely a technical dispute. It cuts to the heart of how we reason about the world. Almost everything we know depends, at least in part, on testimony: history, science (through reports and papers), news and everyday life. If claims were not evidence, then testimony would collapse—and with it, most of human knowledge.

The real question is never:

“Is this a claim?”

But rather:

“How strong is this claim as evidence, given everything else I know?”

The Balanced Conclusion

We can now state the matter with clarity:

• Claims are not automatically sufficient evidence

• Claims are not equally strong evidence

• Claims must be weighed against prior probabilities and other data

But:

Claims are, in general, evidence.

To deny this is not skepticism—it is a misunderstanding of the nautre of evidence and how it functions. And once we see this, the slogan “claims are not evidence” turns out to be not a principle of reason, but a slogan in need of correction.

-- I am much indebted to the work of Nathan Hawkins for the content of this article.