Murphy's law
Sod’s law · Murphy law · anything that can go wrong
Anything that can go wrong will go wrong. Coined on an aerospace test program in 1949, where it was a design rule rather than a complaint: if a part can be installed backward, someone eventually will, so make it impossible.
In practice
The question is not whether a step can fail. It is whether the arrangement lets the failure through — and the honest answer usually comes from someone who has run the step, not from whoever designed it.
The common mistake
Hearing it as pessimism. It is the opposite: a claim that failures are structural and therefore designable-against, as against the view that they are bad luck and nothing can be done.
The line is usually delivered as resignation. Its origin is the opposite.
Edward Murphy was an engineer on a US Air Force project in 1949 measuring the deceleration a human body could survive. A technician had wired a set of sensors, and every one of them had been wired the wrong way round. Murphy's remark — that if there is a way to do it wrong, someone will — was not about the technician. It was about the sensor, which had been built so that wiring it backward was possible.
The program's lead quoted it at a press conference, credited their safety record to designing around it, and the phrase escaped into English and lost its meaning on the way.
The instruction underneath
The useful version is a design rule: if a failure is possible, assume it will happen, and arrange things so that it cannot.
Not be careful. Care is a resource that runs out, and a system depending on everyone being careful on a Friday afternoon in year three is a system that has already failed and has not noticed. The rule asks for the possibility to be removed.
The engineering word for this is a poka-yoke — a connector that only fits one way, a cap that cannot be replaced until the part beneath is seated, a form that will not submit with the field empty. What all of them have in common is that the correct action is the only action, so nobody has to remember.
The business versions are unglamorous and effective. A payment process that cannot be completed by one person alone. A deploy that will not run with the wrong configuration file. A handover that cannot be marked done until the receiving person has confirmed it. None of these rely on anybody's attention.
Where it stops being true
The law as stated is false, taken literally. Most things that can go wrong do not, most of the time, and a business that designed against every conceivable failure would never ship anything.
So the honest version has a rate attached: anything that can go wrong will go wrong eventually, at a frequency set by how often the step runs. A step performed twice a year and a step performed two hundred times a day deserve entirely different amounts of protection, and treating them alike is how a process ends up with elaborate defenses around the rare event and none around the common one.
Two questions do most of the work. How often does this run? and what does it cost when it fails? The steps that are frequent and expensive are the ones to make foolproof. The rare and cheap ones can be allowed to fail and be caught.
The related error
Attributing the eventual failure to the person it happened to. Murphy's remark was about the sensor and it is routinely retold as though it were about the technician — which inverts it completely and produces the response that never works, namely telling people to concentrate harder. See Hanlon's razor, which reaches the same conclusion from the other side, and the five whys, which is the method for getting from the person to the arrangement.
Concept web
Open the full webQuestions
What is Murphy's law?
Anything that can go wrong will go wrong. It was coined by engineer Edward Murphy in 1949 on an aerospace test program, as a design rule: if a sensor can be wired backward, eventually one will be, so build it so it cannot.
Is Murphy's law pessimistic?
No. It treats failures as structural and therefore designable-against, which is more optimistic than the alternative view that they are bad luck. The instruction is to remove the possibility, not to be more careful.
How do you apply Murphy's law?
Make the correct action the only available action — a connector that fits one way, a process that cannot complete with a step skipped. And weight the effort by how often the step runs and what a failure costs.