The phrase means that whoever holds authority, strength, or influence also inherits an obligation to answer for how it is used — the greater the capacity to affect others, the wider the duty to consider consequences before acting. The binding condition is simple: power without accountability converts into harm, and the line names that conversion as the cost of holding power at all. It applies to parents, managers, engineers, elected officials, and anyone editing the lives of other people. Read it as a structural rule about duty, not a compliment for the powerful.
The Origin of the Line Traces Through a 1962 Comic and a 2002 Film
The canonical source is Amazing Fantasy #15, published by Marvel Comics with an August 1962 cover date, in which Stan Lee and Steve Ditko introduced Peter Parker. The issue closes with a narrator caption stating that "with great power there must also come — great responsibility," which makes the line a narration about a character's failure rather than a speech from a mentor.
Stan Lee's 1974 book Origins of Marvel Comics rendered a shortened version, and the 2002 film Spider-Man gave the line to Uncle Ben in dialogue. That screenwriting choice is why most people now hear the sentence in a father's voice rather than a narrator's.
Know the difference between the "must also come" phrasing of the original and the modern "comes." The original conditional carries a stronger claim: responsibility is not a byproduct of power but a requirement attached to it.
Because two versions circulate, quote the one you mean. Citing the comics version signals the structural claim; citing the film signals the family-advice framing. Both are legitimate, and neither is a misquotation.
Three Real Examples Show the Rule Operating
Start with a hospital pharmacist. A single keystroke change to a dosage order can reach a patient in minutes, and a pharmacist's verification permission is a lever over another person's body. The duty side is checkable: independent double-checks, allergy cross-references, and the refusal to override a flag.
Take a mid-level engineering manager who controls performance reviews and project assignments. That person can stall a career for a year with a single framing sentence. The responsibility shows up as written criteria, a documented review, and the willingness to state the bad news in a way the employee can understand and contest.
Consider the holder of a company's administrative credentials to a cloud account. One clicked grant can expose customer records across a region. The corresponding duty is scope discipline: least privilege, an access review on a schedule, and an audit trail long enough to reconstruct who changed what.
None of these examples involve heroism. Each involves a plain asymmetry: one person's decision, many people's exposure.
The pattern repeats at every scale. Power is anything that lets you shorten the distance between your decision and someone else's consequence.
Responsibility Measures as Answerability, Not Good Intentions
Responsibility here means answerability: you can be called on to explain a decision, the information behind it, and the alternatives you rejected. Good intentions do not discharge it, because the standard is what you can answer for, not what you meant.
That makes the concept testable. If no one can ask you why, or you cannot produce the reasoning, the responsibility side of the equation is missing no matter how much power you hold.
Edmund Burke's line in his 1774 Speech to the Electors of Bristol argues that a representative owes constituents his judgment as well as his industry and must not sacrifice it to their opinion. The phrasing reframes responsibility as a duty to think well on behalf of others, which is not the same as obeying them.
Combine the two ideas and you get a working test. Holders must explain their judgment and must not simply mirror the crowd's preference. Both obligations fall on the same person at the same time.
The Responsibility Gap Widens as Technical Power Grows
Modern technical systems create a specific problem: the person who makes a decision is rarely the person who feels the consequence. A product manager changes a default setting, and a hundred thousand users, most of them unknown to that manager, end up affected.
The gap matters because it is measurable. A single deployment permission reaches more people than any individual interaction ever could, and the time between cause and visible effect keeps shrinking. Detection windows compress, which leaves less room for correction.
Named frameworks have grown around that gap. The Markkula Center for Applied Ethics at Santa Clara University publishes a practical framework for responsible technology development that pushes teams to identify who benefits, who bears risk, and who can contest a decision. Those three questions mirror the traditional phrasing closely.
Practical rule: when your decision reaches people you will never meet, document the decision as if you were writing to them. That habit closes part of the gap faster than any policy memo.
A Four-Step Test Converts the Line Into a Decision Procedure
Here it is as a routine you can actually run, in order, before you sign off on something that affects others. Keep the sequence, because skipping to step three gives you a defense without a diagnosis.
1: Quantify the reach. Write the number of people affected, the size of the realistic worst-case effect per person, and the length of time the effect could persist before anyone notices. If you cannot produce those three figures, you are not ready for the next step. 2: Name the answerable party. Identify the single person who would have to defend the decision in front of those affected, and confirm that person has actually read the plan. A committee name is not a responsible party. 3: Build the correction path. Specify how someone harmed can reach you, how long a reversal takes, and who is empowered to trigger one without seeking approval. If reversal takes three levels of sign-off, the power exceeds the correction capacity. 4: Log the rejection. Record the alternatives you declined and why, with the date. The audit trail is often the only artifact that proves judgment was exercised rather than assumed.
For a worked example, imagine a small team changing a password-reset email template that serves 40,000 accounts. Reach is 40,000 people. Worst case per person is account lockout plus a visible phishing lookalike. Persistence is up to two hours before monitoring catches the error rate. The answerable party is the product lead, and reversal takes one configuration rollback, about 11 minutes based on the team's own change log.
Failure handling matters here. If step 1 or 2 stalls because no one knows the numbers, that stall is itself the finding: the decision is operating beyond its measured reach. Stop and measure first. And if step 3 produces "we escalate to leadership," you have located a real gap — correction has no owner close enough to act in time.
Two Worked Benchmarks Reveal the Cost of the Gap
Consider a hypothetical mid-size online retailer with 250,000 active accounts, examined through the four-step test. The team holds one credential that can export the full customer table, which puts reach at 250,000 people. The export is one command, executed in seconds, with detection currently dependent on a weekly log review.
Run the arithmetic. If detection takes seven days and the export runs in under a minute, the exposure window is roughly 10,080 minutes, or 168 hours. A 20-minute detection latency would cut that window by 99.8 percent compared with a one-week review cycle — a computable difference of 9,960 minutes. That gap, not the size of the export, is what the team can actually fix.
Contrast it with a team of eight people and no customer data, examined on the same dimensions. Reach is eight colleagues, worst case is a delayed deliverable, and persistence is one working day. The same export permission in that context is proportionally harmless, which shows why blanket rules about credential size miss the point.
Both comparisons point to one number per case: the minutes between decision and detection. The framework targeting responsible technology development at Santa Clara's Markkula Center names upward and downward accountability among organizations, not credential counts, as a core question — and detection time is what makes downward accountability visible.
Comparison: How Each Framework Assigns Duty
Four well-known formulations handle the same problem with different mechanisms, and naming the mechanism helps you pick one for a given situation.
| Framework | Core mechanism | Strength | Limitation | Fits best for |
|---|---|---|---|---|
| Comic narration (Amazing Fantasy #15, 1962) | Prevents harm from omission | Compact and memorable | No procedure attached | Teaching the principle |
| Burke's trustee model (1774) | Judgment owed to constituents | Defends independent reasoning | Can collide with direct democracy | Elected representatives |
| Markkula Center framework | Stakeholder benefit, risk, and contest analysis | Forces named stakeholders | Requires facilitation time | Product and engineering teams |
| Four-step reach test (this article) | Quantified reach plus named answerable party | Produces auditable numbers | Demands honest measurement | Pre-launch sign-off |
The comic version carries no procedure, which is why it gets quoted and then ignored. Burke's model fits representatives who must sometimes resist their own voters. The Markkula framework spends time facilitation buys back in clarity, and the four-step test trades elegance for numbers you can defend.
Choose the Markkula framework if your team has hours to spend on stakeholder mapping. Choose the four-step test if you need a decision by the end of a meeting. Skip the four-step test for reversible, low-reach decisions; the measurement overhead will exceed the risk.
Failure Cases Show Which Duty Usually Breaks First
Most failures don't start with malice. They start with the least costly duty being abandoned.
A permissions-review skip is the common edge case. An administrator inherits access from a departed colleague's project, never revisits it, and holds dormant privileges for months. Nothing looks wrong until an incident forces a reconstruction of who could have done what, and by then the audit trail may not answer the question.
Another failure is the delegated-but-unnamed approval. A director says "use your judgment," which feels generous. When the judgment goes wrong, no one can say whose judgment it was, and the correction path stalls while the responsibility search runs.
A third is the loud-advisor trap. A representative reads a strong majority opinion on one narrow issue and treats it as pressure to abandon every independent judgment on adjacent questions. Burke's distinction between being a representative of a district's interests and being a delegate for its momentary wishes is the useful correction here.
Watch for silence as a failure mode too. A decision made entirely through defaults, with no one logging the rejected alternatives, produces no evidence that anyone exercised judgment at all. The absence of a record becomes the first problem in any review.
Holding Power Requires Weathering Disagreement Without Bending
The harder half of the sentence is not the duty to act responsibly. It is the duty to stay answerable when answering is uncomfortable.
One practical discipline is to write down your reasoning before you learn the reaction to it. A note dated before the meeting is evidence of independent judgment; the same note written after is a rationalization, and everyone in the room knows the difference even if no one says so.
Another is to keep a standing correction channel that people can use without social cost. If the only route to challenge a decision runs through your inbox, the channel will go unused and you will mistake quiet for agreement.
Stay inside professional limits. None of this grants permission to override required safety reviews, legal obligations, or compliance sign-off. The scope of autonomous judgment is what remains after those obligations are met.
For high-stakes decisions, the habit that protects both judgment and the people affected is a written record sent to the people who need it, kept where they can find it later. Power exercised in the open is easier to correct.
FAQ
What is the correct wording of the phrase?
The original 1962 narration is "with great power there must also come — great responsibility," while the standard modern phrasing drops "must also." Both circulate widely, and the original appears in Amazing Fantasy #15, cover-dated August 1962. Anyone quoting it should specify which version they mean, because the original makes responsibility a requirement rather than a consequence.
Did Uncle Ben actually say it in the comics?
No — in Amazing Fantasy #15 the words arrive as a closing narration caption, and Uncle Ben's death motivates them without him speaking them. The 2002 film Spider-Man reassigned the line to Uncle Ben in dialogue, which is why the phrase is now usually attributed to him. Shifting the quote to Ben is a screenwriting choice, not a comics fact.
Is the phrase actually a good principle for making decisions?
It works as a first filter but contains no procedure, so a team that only quotes it will still need a concrete test. The four-step reach test in this article — quantify reach, name the answerable party, build the correction path, log the rejection — turns the sentiment into something auditable. Without a procedure like that, the phrase tends to become decorative.
Who actually wrote the phrase?
Stan Lee is credited with the narrative script for Amazing Fantasy #15, with Steve Ditko on art, and the line appears in that issue's narration. Stan Lee's later retellings and the film adaptation influenced consumer versions of the phrase. Attribute it to the comic's narrative voice first, since that is the documented publication.
Does responsibility really scale with power, or do some roles carry fixed duties?
It scales in scope, not in kind, because the underlying duties are fixed while the reach expands. A pharmacist, an engineering manager, and a database administrator all owe accuracy, disclosure, and a correction path. What changes is the number of people those duties touch. Scope tells you how much diligence fits the role.
What should you do if a decision is already made and the responsibility is unclear?
Start a written record with three elements: who decided, who was consulted, and who can reverse the outcome. Then run a single measure of the reach and the detection time. A written record with those five items gives the next reviewer something to act on and exposes where the answerable party is missing.
Can you decline power instead of accepting responsibility?
You can decline a role, but not the consequences of a decision you already influence. Refusing a formal title doesn't undo the effect of giving or withholding input, because an omitted voice is a visible decision in most systems. Choose between clearly scoped authority and acknowledged influence.
How does this apply to an individual with no formal authority?
Influence over outcomes — the trust of a team, access to information, the ability to escalate or stay silent — is the power at stake. Declining to record a decision, or choosing not to flag a risk, is the common case. Tracking which risks you flagged and when is a workable practice for people with no formal title.
This article is general educational information about a widely quoted principle and how responsibility applies in professional settings; it is not a legal opinion, compliance determination, or medical or financial recommendation. Required duties vary by profession, jurisdiction, and organization — confirm with the relevant regulator, licensed local professional, or your organization's compliance policy before acting on a high-stakes decision.