Chapter 7: The Null Hypothesis and the Burden of Proof

You now have the basic grammar on the table.

In Chapter 6, you learned to separate questions, claims, and evidence. You can tease a precise claim out of a vague statement. You can tell the difference between a strong feeling and something that actually counts as evidence. That's enough to start using the first explicit tools of epistemological skepticism.

This chapter gives you two.

They work together:

• The Null Hypothesis is a starting stance: "not yet persuaded."

• The Burden of Proof is a rule of thumb for who needs to provide evidence, and how much.

Neither tool is new in the history of science or philosophy. What matters here is how you can use them in daily life, without equations or special training.

The Null Hypothesis: starting from "not yet persuaded"

In statistics, the Null Hypothesis is a formal device: an assumption that there is "no effect" or "no difference," which you try to dislodge with data.

In this book, we're using a simpler, more human version.

The Null Hypothesis is the practice of starting from "not yet persuaded" whenever a new claim attempts to change your map.

It does not mean "I believe the opposite." It does not mean "this is false." It means:

"I hear the claim. I am not committed either way yet. I am willing to be moved—but only by evidence and reasoning, not just by confidence, repetition, or pressure."

This stance has two parts:

1. You don't treat new claims as true by default.

2. You don't treat them as false by default either.

Instead, you hold them at arm's length for a moment and ask:

• "What exactly is being claimed?"

• "What kind of question is this trying to answer?"

• "What sort of evidence would bear on it?"

Only then do you let your confidence start to move.

You have probably felt versions of this stance already.

When a stranger in a parking lot offers you an unbelievable deal, you don't immediately accept or reject. You go to "not yet persuaded," ask follow‑up questions, and look for signs that something is off. When a friend you deeply trust tells you something surprising, you may still pause briefly: "Wait, really? Tell me more." The Null Hypothesis is about making that pause a conscious, portable habit.

Why this stance matters now

In a slower, more stable world, it was less costly to assume that many claims were probably fine.

Information moved slowly. Fabricating convincing data or images was hard. Institutions had longer‑term reputational stakes. Your inherited epistemology—trust certain kinds of language, defer to certain roles, believe what "everyone knows"—worked well enough.

You don't live in that world anymore.

You live in a world of synthetic fluency: texts, images, audio, and video that can mimic the surface features of authority without having earned it. You live in a world where repetition is cheap and outrage is profitable. You live in a world where some actors have strong incentives to bend your map in their favour.

In this environment, the old default ("believe until you have a reason not to") is too generous.

The Null Hypothesis updates that default to something like: "hold neutral until the case is made." Not forever; not in a way that leaves you paralysed; but long enough to see whether a claim has any real weight behind it.

Misuses: when "null" becomes a shield

Like any tool, this stance can be misused.

Here are two common failure modes:

1. Weaponised doubt.Someone pretends to hold the Null Hypothesis but is, in fact, committed to rejecting anything that threatens their identity or interests. They say "I'm just asking questions" while refusing to engage with strong evidence. This is not epistemological skepticism; it is bad faith.

2. Paralysis.Someone uses "not yet persuaded" as an excuse never to commit, even when action is required. They keep asking for more and more certainty in domains where only partial, noisy evidence is possible, and where waiting has its own costs.

The stance this book advocates is different.

It asks you to start from "not yet persuaded," then update honestly as evidence appears, and act when the balance of reasons and stakes calls for it—even though some uncertainty remains.

The Null Hypothesis is a doorway, not a bunker.

The Burden of Proof: who needs to show what

If the Null Hypothesis is your starting stance, the Burden of Proof tells you who needs to move first.

In everyday life, you're constantly dealing with competing claims:

• "This supplement will improve your focus."

• "This policy will reduce crime."

• "This person is dangerous."

• "This AI is safe enough to deploy."

The burden of proof is the principle that the person making a claim is responsible for providing sufficient evidence for it, especially when the claim is strong, unusual, or high‑stakes.

Put simply:

If you want others to change their map, you carry the burden of showing why.

If someone claims that a new technology is harmless, they have to back that up; you don't have to disprove every possible failure mode before you're allowed to be cautious. If someone claims that a vaccine is dangerous, they have to provide good evidence; you don't have to refute every rumour.

This sounds obvious. It gets violated all the time.

How the burden gets shifted (and how to notice)

Here are a few common ways people try to shift the burden of proof onto you:

• "If you can't prove me wrong, you have to accept I'm right."This is backwards. The inability to disprove a claim does not make it true. "No one has disproved that X causes cancer" is not evidence that it does.

• "Everyone knows this."Appeal to "what everyone knows" is often a way of hiding the fact that no one has actually done the evidential work. It tries to recruit social pressure instead of evidence.

• "Prove to me that it's safe."In high‑harm domains (drugs, bridges, nuclear plants, AI systems), the burden is on the person proposing the risky action to show that it is safe enough, not on others to show that it is dangerous. You cannot prove perfect safety; you can demand adequate evidence in proportion to the stakes.

• "If you doubt this, you're a bad person / on the wrong side."This is moral blackmail in epistemic clothing. It tries to make doubt socially or morally costly so that you stop asking for evidence.

When you see these patterns, you can gently redirect:

• "I'm not claiming the opposite; I'm saying I'm not yet persuaded. What evidence are you relying on?"

• "You're making a positive claim; can you show me something that supports it?"

• "Given how high the stakes are here, what have you seen that justifies this level of confidence?"

You are not demanding mathematical proof. You are asking the right person to carry the right burden.

Matching burden to stakes

Burden of proof is not an on/off switch.

It scales with at least three things:

1. How strong the claim is."This might help a bit" needs less evidence than "This will cure the disease for everyone."

2. How unusual the claim is."This new phone has a slightly better camera" is different from "We have overturned a century of physics."

3. How much is at stake if you're wrong.A new dessert recipe and a new nuclear reactor design do not deserve the same evidential bar.

You do this informally all the time.

If a friend tells you about a new restaurant, you might try it on the strength of their recommendation alone. If a stranger on the internet tells you to move your savings into a sketchy scheme, you (hopefully) demand much more. The more unusual and high‑impact the claim, the heavier the burden of proof you should place on the person making it.

Later, in Chapter 9, when we talk about proportional scrutiny and the evidence hierarchy, we will formalise this scaling more explicitly. For now, it's enough to notice that your "how much evidence?" question should track stakes, not just curiosity.

Using both tools together

Let's see how the Null Hypothesis and Burden of Proof work as a pair.

Suppose someone says: "This AI assistant is fully safe and will never cause serious harm."

• You start from the Null Hypothesis: "not yet persuaded."

• You note that this is a strong, unusual, and high‑stakes claim.

• You place a heavy burden of proof on the person making it.

You might respond:

• "That's a big claim. What do you mean by 'safe'? Safe for whom, and in what sense?" (Clarifying question.)

• "What evidence do you have that serious harms won't occur? What tests have been run? What would count as failure?" (Calling in the burden of proof.)

If they can't provide anything beyond assurances, branding, or "everyone's using it," your Null Hypothesis stays in place. You might still choose to use the system—but you will do so with calibrated caution, not with borrowed certainty.

The same pattern applies at smaller scales:

• A friend says, "This diet fixes everything."

• A pundit says, "This policy will obviously solve the problem."

• A headline says, "Scientists prove X beyond doubt."

In each case:

1. Start from not‑yet‑persuaded.

2. Ask what is actually being claimed.

3. Ask who carries the burden of proof.

4. Ask whether they've met it, given the stakes.

The burden applies to you too

Here is where this gets uncomfortable.

The Burden of Proof is not a weapon you wield against others while exempting yourself. When you make a claim, you carry the burden.

This includes:

• Claims you make in conversation.

• Beliefs you hold and share.

• Opinions you assert as if they were facts.

• Stories you tell about yourself, others, or the world.

If you expect others to provide evidence for their claims, you must be willing to provide evidence for yours. If you expect others to start from "not yet persuaded" about your assertions, you must start from "not yet persuaded" about your own.

This is harder than it sounds. Most of us have a collection of beliefs we have never seriously examined. They came from parents, culture, trusted authorities, or just repeated exposure. They feel true because they are familiar, not because we have tested them.

The Null Hypothesis, applied to yourself, means treating your own beliefs as provisional—always open to revision if the evidence shifts. It means noticing when you are defending a belief without evidence, and asking yourself: "If I were hearing this claim from someone else, would I accept it on these grounds?"

A small practice: "who owes what?"

Here is a practice you can use this week.

Once a day, when you encounter a strong or emotionally charged claim (in conversation, news, or social media), pause and ask yourself three questions:

1. What is the claim?

   Can I state it clearly in one sentence?

2. Who carries the burden of proof?

   Is it on the person making the claim? On me? On an institution? Am I accidentally taking on a burden that isn't mine?

3. Given the stakes, has that burden been met?

   For something this strong or this risky, have I seen enough to move from "not yet persuaded" to "provisionally convinced" or "willing to act as if this were true"?

You don't have to confront anyone. You can do this silently, as a way of training your own reflexes. Over time, you'll find that you are less easily bounced into agreement or disagreement by tone or urgency alone. You will have a place to stand, and a sense of who owes what before your map should change.

Next: Chapter 8 – Falsifiability and Failure Modes