Occam’s Razor Revisited

When Should We Keep It Simple?

Occam’s Razor

In a Nutshell

The Principle of Simplicity

William of Occam, or Ockham, was a fourteenth-century philosopher and theologian whose name has become associated with an apparently simple idea – that we shouldn’t make an explanation more complicated than it needs to be. More precisely, if two theories explain the evidence equally well, we should prefer the theory which requires fewer assumptions. It’s become known as Occam’s Razor, and we use it to shave off those unhelpful additional assumptions. A word of warning, though. Sometimes the evidence favours a more complex explanation. 

Occam’s Leprechaun

Perhaps the easiest way to see the force of Occam’s Razor is with what seems on the face of it a somewhat wild example. Here goes.

We walk into a room and find a broken vase. I have a theory that a leprechaun did it, and I’m convinced it’s true. You look around but see no leprechaun. I have an answer for that. “Because he’s invisible”.  

You suggest  spreading flour across the floor to detect his footprints. I have an answer for this too. “He floats”.

You suggesting listening for him. No need, I say. “He never makes a sound”. 

And so it goes on.

Every time you suggest a way of testing the theory that a leprechaun broke the vase, I just keep modifying my story. But that’s the problem. The more assumptions I have to add, the less convincing the leprechaun theory becomes.

These modifications are sometimes called saving hypotheses. One awkward piece of evidence appears, so we add an assumption to save the theory. Another appears, so we add another. Occam’s Razor tells us to be suspicious when this starts happening.

Perhaps there isn’t an invisible, floating, silent vase-breaking leprechaun after all!

Does Occam’s Razor Always Favour the Simplest Explanation?

Occam’s Razor doesn’t suggest that the simplest explanation is always the best one. This is a common misconception. There’s no law of nature saying that reality must always take the simplest possible form. Nor are we allowed to reject evidence merely because the explanation required by that evidence happens to be complicated. Think in terms of a murder investigation. Perhaps the simplest initial explanation is that the victim died accidentally. But fingerprints, CCTV footage and forensic evidence might eventually reveal a complicated conspiracy involving several people. At that point, insisting on the simpler explanation would no longer be sensible. It would mean ignoring evidence. Occam’s razor cuts away what we don’t need, not what we do. 

Prediction Matters

There is another reason to value simpler theories. We don’t merely want theories that can explain what has already happened. We want theories that tell us something about what will happen next. It’s like drawing a wavy line after the event to join up every single point. It will fit the data points exactly but doesn’t necessarily best predict what happens next. A simple trend line might be much better at doing that. It’s called the problem of “overfitting” the data. Put another way, a complex theory can sometimes be constructed after the event to explain almost anything. But can it make successful predictions? A simpler theory might well do that better, 

In sum, a theory which explains a large number of observations, makes successful predictions and does so with relatively few assumptions has something important going for it. It is doing a lot with a little.

And that, perhaps, gets closest to the real appeal of Occam’s Razor.

So Should We Accept the Simplest Explanation?

Not always. We should prefer the simplest explanation that adequately accounts for the evidence. Those last few words matter. So do we need that additional assumption? It’s an important question. And it’s at the cutting edge of Occam’s Razor.