Tool/Philosophical Razor/No. 0686
Occam’s Razor
Occam’s razor is a methodological rule that favors explanations with fewer unnecessary assumptions when they fit the evidence equally well. Named after William of Ockham, it is a principle of explanatory economy, not a guarantee that the simplest account is true.
Also called Ockham's Razor
- Evidence
- Useful, modest evidence
- Read
- 6 min
- Links
- 8 connections
01You've seen this when…
- in life
Three houseplants droop after a weekend away. Before buying treatments for three diseases, you check whether anyone watered them.
- at work
A spreadsheet total changes unexpectedly. A colleague suspects corrupted files and a software bug; you first check the formula someone edited yesterday.
- out in the world
Taps run brown across several blocks after a water-main break. Residents check the utility’s notice before treating each home as a separate plumbing failure.
02The idea
When several things go wrong together, it’s tempting to invent a separate explanation for each. Sometimes you need them. Sometimes one shared cause explains the whole pattern without the extra machinery.
Occam’s razor favors the explanation that achieves comparable explanatory and predictive success with fewer unnecessary assumptions. The comparison matters. A simpler account that ignores half the evidence falls short. An account that fits yesterday’s observations but repeatedly fails tomorrow’s tests also falls short.
An assumption is something your explanation needs you to accept. Suppose one account requires a broken connection. Another requires a broken connection, a coincidental software failure, and someone secretly changing the settings. If the extra claims explain nothing more, leave them out.
An explanation’s simplicity depends on its assumptions. Saying that someone sabotaged everything is short, but it may require unsupported claims about access, motive, timing, and concealment. A longer explanation built from known mechanisms can carry fewer assumptions.
The razor helps choose which explanation to investigate or provisionally accept. Assessing whether that explanation is true requires evidence. The razor is one guide within inference to the best explanation. When new evidence demands more complexity, add it.
03How to use it
- Write down what needs explaining. Separate observations from interpretations. Three services return errors is an observation; three services have separate bugs is already a theory.
- Keep at least two plausible accounts in play. Include a shared cause when the failures could have a common dependency. Compare your favorite explanation with at least one plausible alternative.
- Expose the supporting assumptions. List what each account requires beyond the facts you already know. Mark which requirements have independent support and which are guesses.
- Check coverage before simplicity. Does each account explain the timing and account for both exceptions and unaffected cases? An explanation that works only after discarding inconvenient observations fails the coverage check.
- Choose a test that separates them. Look for something the accounts predict differently. A falsification test is more useful than another observation they can both accommodate.
- Keep your judgment open to evidence. Give the simpler account provisional priority, then let the test change your judgment. Bayesian updating captures this habit: evidence can outweigh an initially reasonable preference.
For practical troubleshooting, test cost matters too. A cheap, safe check can deserve priority even when its explanation isn’t the most likely.
04A worked example
In an invented example, a nightly software release disrupts a warehouse’s dashboard. The dashboard stops loading inventory and displaying order histories, and shipping label creation fails. The three functions use different back-end services.
What it looks like The release has introduced three separate bugs. Each service owner starts examining their own code.
What’s actually going on All three functions also pass through one internal gateway. Its logs show that requests are rejected before reaching any of the services. The release changed the name of an environment setting, so the dashboard now sends an outdated access key. One configuration mistake explains the timing and all three failures. An account based on three independent service bugs requires more assumptions and leaves the gateway’s rejection logs unexplained.
The team still considers another shared cause: perhaps the dashboard’s permissions were revoked. They compare the configured key with the current one, then send a test request using the current key. It succeeds. Restoring that setting brings back all three functions.
What made it work The razor directed attention toward a shared dependency. The team established the diagnosis by checking the logs and running a controlled test. It treated fewer causes as a provisional clue and kept checking even after it found an appealing story.
05When to reach for it
06When it misleads
- You confuse familiar with simple. A comfortable, familiar explanation can rest on many unexamined assumptions. Count what the account requires.
- You trade away evidence for elegance. Real-world systems can have several simultaneous causes. When the data demand two faults, updating the explanation means accounting for both.
- You judge simplicity by the number of labels. One named force or one grand conspiracy can conceal enormous complexity. A single added noun can introduce several assumptions.
- You overlook prior plausibility. Two accounts with the same number of assumptions can differ in likelihood. Established mechanisms and independently supported facts matter. Apply the razor together with background knowledge.
- You assume the smallest model predicts best. A model can be too simple and miss underlying structure. Regularization and minimum description length formalize particular simplicity trade-offs. Whether reducing complexity improves predictions depends on the case. Check predictions on data held out when building the model.
Occam’s razor also differs from Hanlon’s razor. Hanlon advises caution about attributing behavior to malice when an ordinary mistake would explain it. Occam concerns unnecessary assumptions, whether the proposed cause is malicious, accidental, or neither.
07Roots
William of Ockham, a Franciscan friar, wrote his Summa Logicae in the early fourteenth century, in a scholarly world preoccupied with what words committed people to believing existed. There was Socrates, and there was Plato. But did explaining what they shared also require a separate, universal entity called humanity?
Ockham resisted adding such entities outside the mind when individual things and concepts could do the explanatory work. He worked in medieval philosophy and theology, before modern laboratory science developed. Distinctions could multiply rapidly as scholars worked through arguments inherited from Aristotle and their predecessors. Economy helped limit unnecessary machinery in those arguments.
Thinkers before Ockham had already used the general preference for economy. Later writers attached his name to the principle and its razor image: cutting away unnecessary additions. As the idea traveled into science, the emphasis shifted toward comparing explanations and predictions. Modern statistics gives some versions a precise form, especially when limiting complexity helps counter overfitting. The enduring lesson calls for greater selectivity than its popular slogan suggests: keep everything reality requires and cut unnecessary additions.
08How solid is this?
A well-developed rule of method, not an experimentally established law of truth. Statistical arguments support simplicity penalties in some prediction problems. No general result says fewer assumptions always means a truer explanation.
09Connections
- Often confused with Hanlon’s Razor
- Helps counterOverfitting
- Part ofInference to the Best Explanation, Bayesian Updating, Minimum Description Length
- See also Falsification Test, Map vs. Territory, Regularization
10Origin and sources
Associated with William of Ockham, a fourteenth-century Franciscan philosopher. Principles of explanatory economy predate him; later writers attached his name to the razor.
- [1]Sober, E. (2015). Ockham's Razors: A User's Manual. Cambridge University Press.
- [2]Baker, A. Simplicity. The Stanford Encyclopedia of Philosophy.
- [3]Spade, P. V., & Panaccio, C. William of Ockham. The Stanford Encyclopedia of Philosophy.
Suggest an edit· Updated 2026-10-02