Life's Little Mysteries

What is Occam's razor?

If you're contemplating a complex problem, think of Occam's razor, as simplicity is often best. (Image credit: Laurence Dutton via Getty Images)

Occam's razor (also spelled Ockham's razor) cuts through complexity with a no-nonsense approach. The philosophical maxim "Numquam ponenda est pluralitas sine necessitate," written by 14th-century Franciscan friar William of Ockham, translates to "Plurality must never be posited without necessity." In other words, all else being equal, simplicity is best. 

So is this actually true? Is the simplest explanation usually the best one? 

Not exactly. Ockham never said complexity is inherently inferior to simplicity, nor did he declare complex explanations inherently wrong. Complex scientific questions often demand complex answers, and that's not at odds with Occam's razor. The principle merely states that unnecessary complexity is, well, unnecessary. 

"Occam's razor is about finding the simplest solution that works," Johnjoe McFadden, a professor at the University of Surrey in the U.K. and author of the book "Life Is Simple: How Occam's Razor Set Science Free and Shapes the Universe" (Basic Books, 2021), told Live Science in an email. "It never fails so long as you remember the necessity clause."

Ockham was not the first to promote simplicity. Aristotle held that "the more limited, if adequate, is always preferable," and Ptolemy considered it best "to explain phenomena by the simplest hypothesis possible." Some three centuries after the genesis of Occam's razor, Isaac Newton would declare that "we are to admit no more causes of natural things than such as are both true and sufficient to explain their appearances." About 200 years after that, Albert Einstein would agree that "everything should be made as simple as possible, but not simpler" (which is, in fact, a simplification of his original quote).

Related: Are humans inherently violent?

When used correctly, Occam's razor works. If two computer programs accomplish the same task, the one with less code is inevitably more efficient. The simplest medical diagnosis is usually correct; hospital interns are often taught to think of horses, not zebras, when they hear hoofbeats. One implication of the second law of thermodynamics (disorder increases for any spontaneous process) is that such processes always use the least possible energy.

"Copernicus came up with the heliocentric model of the solar system solely on the basis that it was simpler," McFadden said. "The existence of a single Higgs boson was the simplest solution to the equations of particle physics. Between these points are a thousand scientific advances that depended on simplicity."

When misused, however, Occam's razor can become a blunt instrument of overgeneralization. The principle does not mean, for instance, that we blindly follow the simplest theory, whether right or wrong. "Very often the simplest hypothesis is too simple," Elliott Sober, a professor of philosophy at the University of Wisconsin-Madison and author of the book "Ockham's Razors: A User's Manual" (Cambridge University Press, 2015), told Live Science in an email. "The simplicity of a hypothesis is one consideration, among others that are relevant to assessing whether a hypothesis is true."

When it comes to data science, Occam's razor may cause more problems than it solves. In this case, "The simplest approach is usually wrong," said Pedro Domingos, professor emeritus of computer science and engineering at the University of Washington in Seattle. When Domingos studied the applicability of Occam's razor to machine learning in the early 2000s, he found that a simpler model is superior to a complex one only if it is just as good at predicting new data. 

"As modern machine learning has shown over and over again — in model ensembles, deep learning, et cetera — it's usually the most complex approach that's right, Domingos told Live Science in an email. "And that's not surprising; the phenomena we're modeling are almost always more complex than the models, and the closer to their true complexity we can get, the more accurate the models."

Occam's razor nonetheless remains a useful tool for trimming the fat off of bulky assumptions, at least in our day-to-day lives. "The universe is a complicated place, but it's sometimes made more complicated through the invention of complicated explanations that suit a particular ideology, philosophy or political persuasion," McFadden said. "Occam's razor tells you to forget about all of those."

Joshua A. Krisch
Live Science Contributor

Joshua A. Krisch is a freelance science writer. He is particularly interested in biology and biomedical sciences, but he has covered technology, environmental issues, space, mathematics, and health policy, and he is interested in anything that could plausibly be defined as science. Joshua studied biology at Yeshiva University, and later completed graduate work in health sciences at Cornell University and science journalism at New York University.

  • Hayseed
    Occam's Razor is why, after 100 yrs., we still have no narrative for mass and matter. We have proposed explanations for light and gravity, but no narrative for the source of light and gravity.....mass.

    I don't believe that modern science would except or permit a simple narrative.

    I think physicality is much simpler than our science proposes. And our science will never consider it.
    Reply
  • bolide2
    If the true explanation is "much simpler than our science proposes," then this is not the fault of applying Occam's Razor, but rather of not applying it.
    Reply
  • Hayseed
    I thought that's what I explained. Occam's Razor is a principle. The razor comes from not using it.
    Reply
  • Ridahoan
    It's often been said that in Biology Occam's Razor is best suited to cut one's own throat.

    But it often comes down to the purpose of the explanation or model. Sure, given enough data, more complex models will often produce better predictions, but without those data, or data that reflects fundamentally different distributions, more complex models tend to overfit and make lousy predictions on cases worthy of prediction -- as Niels Bohr said, 'Prediction is difficult, especially of the future.'

    There are several often used entropy based measures to help find the right balance between complexity and simplicity -- Akaike information criterion is a popular one.

    Explanation or models meant for teaching or human understanding of course have limitations in complexity. Everyone knows that simpler models that delineate cause and effect are heuristics, but often immensely valuable ones.
    Reply
  • friedy07
    Ridahoan said:
    It's often been said that in Biology Occam's Razor is best suited to cut one's own throat.

    But it often comes down to the purpose of the explanation or model. Sure, given enough data, more complex models will often produce better predictions, but without those data, or data that reflects fundamentally different distributions, more complex models tend to overfit and make lousy predictions on cases worthy of prediction -- as Niels Bohr said, 'Prediction is difficult, especially of the future.'

    There are several often used entropy based measures to help find the right balance between complexity and simplicity -- Akaike information criterion is a popular one.

    Explanation or models meant for teaching or human understanding of course have limitations in complexity. Everyone knows that simpler models that delineate cause and effect are heuristics, but often immensely valuable ones.
    A math teacher once told our class that Occam's razor was used to shave Plato's beard.
    Reply
  • abc123
    the razor cuts ones throat to kill them
    Reply