Wednesday, December 16, 2009

Occam's Razor

Does it make any sense to have a preference for simpler explanations over more complicated explanations? Is there anything to Occam's Razor?

Yes, a little. There's more that can go wrong with a complicated explanation, or one that makes more assumptions, than a simpler explanation. Simply, when there are more elements to an explanation there is a greater possibility that one of them is inaccurate.

Is it useful or practical to invoke Occam's Razor? It is very useful for making guesses about explanations when you aren't heavily invested in actually being correct, and don't want to invest time and effort in evaluating the content of an explanation that you've already decided you don't like. It's sort of useful for dismissing extreme-case explanations where the number of assumed unknowns is very large (alternatively, you could just point out that there is uncertainty about all those unknowns - this requires you to make some kind of evaluation about the magnitude of the uncertainty rather than just using Occam's Razor as a sort of appeal to authority).

My point is simply that the more complicated explanation could be the more accurate one. Correctly invoked, Occam's Razor is a specific criticism about the content of an explanation. Incorrectly invoked, Occam's Razor is an intellectually lazy dismissal of an idea without addressing the content.

Generally speaking, it is probably better not to directly invoke Occam's Razor, but instead to simply be specific in your criticisms.


  1. As with Dawkins, I agree that invocation of authority is auxiliary at best.

    For me, Occam's razor is the naming and glorification of the key notion that Mother Nature favors simplicity in her fundamental principles, and I think it's wise to keep that in mind whenever evaluating competing explanations for natural phenomena. Even then, it's silly (I think) to say, "It is always so," and be done with your commentary. It's a supporting argument, so there'd better be something more to support.

  2. Bob,

    It's the part about Nature favoring simplicity that I doubt. After all, the standard model is shaping up to be fairly complicated. How many fundamental particles are there in how many variations? Surely there's a more simple explanation (though possibly a less accurate one).

    Maybe that's a bad example because perhaps the standard model will one day be simplified through the insights of future physicists.

    But in any case, how does anyone know for sure whether Nature has preferences for or against simplicity? Two more examples: evolution of life on earth has proceeded from the extraordinarily simple (and extraordinarily successful!) to the extraordinarily complex. Even the Universe started simple and then went all complex (under the understanding of the big bang).

  3. Excellent points, guys.

    I think mathematicians would definitely say that nature favors simplicity (and probably beauty, too), but then how you go about defining simplicity is something else.

    This reminds me of something else I was reading about lately. I am no fan of myers-briggs stuff, but I took the test anyway and it told me I was either an INTP or INTJ. On wikipedia there is this quote about INTP's which I found dead-on:

    "They can demonstrate remarkable skill in explaining complex ideas to others in simple terms, especially in writing. On the other hand, their ability to grasp complexity may also lead them to provide overly detailed explanations of "simple" ideas, and listeners may judge that the INTP makes things more difficult than they are. To the INTP, however, this is incomprehensible: They are merely presenting all the information."

  4. Justin,

    I hope it doesn't knock me down a notch in your eyes, but I actually like the myers-briggs stuff! However I also tend to get taken in by cold readings.

    But you NEED to present all the information! Not only is it necessary for completeness, it's also interesting! I think I'm an INTJ, but that's close enough to be pretty sympathetic.

  5. For what it's worth, I consistently come up as an INFP. To me, it's an important and essential skill to be able to read one's audience in a very dynamic way, i.e. "Is what I'm doing right now helping to make my point, or hurting?" and adapting accordingly.

    Robert, I have many thoughts about nature vs. our mental models of nature and where the notion of simplicity comes from. I simply can't type it all in this little box.

  6. Bob,

    I was wondering if you were going to call BS on my reasoning by analogy. I'm actually not sure if what I presented amounts to false analogy, but I see it as being a distinct possibility.

    Simplicity definitely needs some definition. I sort of started to define it in the post by talking about how many elements are involved in an explanation.

    Ultimately, I just think of Occam's razor as being over-rated. It's an heuristic, not a description of how the universe actually behaves. And it's subject to abuse, as I discussed in the post.

  7. Hey, Robert.

    My main point is that I think Occam's Razor is appropriately applied to theories/models of Nature and not to Nature herself - and the two are very different. Even then, I think it only makes sense when evaluating competing theories that are equally effective at describing and predicting observed phenomena. In that case, why would anyone prefer the more complicated model? If they're equivalent, hang onto them all; some will fall out when new observations come along. This approach has been very effective for a very long time.

    And yes, Nature demonstrates incredible complexity phenomenologically, but that doesn't mean that the fundamental principles are themselves complicated. One of the hallmarks of any "good" model is that, while it is simple, it facilitates the expression of diverse phenomena. To me, this is purely a human quest for simplicity of description. It doesn't necessarily mean that "Nature is simple."

    I think of Occam's Razor as a criterion for preference between two or more otherwise equally effective models.

  8. Oh, I forgot to add: Using the article "an" before "heuristic..." Now that's too complicated! ;)

  9. Robert, I'm with you on the myers-briggs stuff. I think it is fun and enjoyable to read about personality stuff but I do not trust their accuracy since people have the tendency to see themselves in all kinds of descriptions.

  10. Bob,

    I feel sad as I see the 'an before H' rule falling into disuse!


    Yeah, that's what I meant about cold readings. It's all so seductive, and flattering!

  11. Robert, do you say "an helicopter?"

  12. Well, I But maybe I will start!