Lecture Notes on Model Thinking I

Some lecture notes/ scribble on Model Thinking by Scott E. Page. Lecture Intro, Part 2: “Intelligent Citizens of the World”:

“George Box: ‘essentially all models are wrong, but some are useful'”

|=: Agree! Just a little everyday-obervation: it’s funny how often people use right/ wrong in the context of abstraction (i.e. models), not realising, without being ‘wrong’ (i.e. loosing details) it would not be a model. Such discussions usually start sth like this:
2 Rects
A:‘These are two rectangles!’ B:‘No, no, these are two essentially different things!’ … (sounds familiar?)

Have fun

Are you a hedgehog or a fox?: Try the quiz on Overcoming Bias
and this is Philip E. Tetlock: Wikipedia

About modelpractice

Modeling Theory and Abstraction Awareness in strive for scientific rigour and relevance to information systems engineering.
This entry was posted in Abstract Thinking, Model Thinking and tagged , , , , , , , , , , , , . Bookmark the permalink.

17 Responses to Lecture Notes on Model Thinking I

  1. Pingback: Model as Use « THINK IN MODELS

  2. vhanniet says:

    Is an idea wrong? Undecidable with so few context elements. A model is wrong only if it does not serve the intent it is used for as an explanation. So, the two rectangles are equivalent, different, or inappropriate (and every mixed response) relative to a problem which has to be set first!

  3. I disagree, since I don’t see why we should see George Box’s quote as “not wrong”. Accepting it is regreting it, somehow…

    • Hi, would “using it” instead of “agreeing” ok for you?

      • I think that’s not enough. Why should we use something self-refutable?

        That’s ok to see all models as instrumental (i.e., something that might be useful in some scenario). The problem arises when you deny some models could be more than that without justification, or when we assign a value judgement to the model itself. If you can’t prove something as “wrong/false”, that doesn’t prove it as “right/true” (false dilemma fallacy), but we shouldn’t say it’s “wrong/false” before knowing that. Also, the model itself can’t be “wrong” and also can’t be “not wrong”: it’s only a model, what can be wrong is an assignment of the model to the reality (be it subjective or objective), not the model itself.

  4. Hi, yes, think that’s what he actually meant (and how I interpreted it subsequently). Perhaps that’s why he put the “essentially” in front? Seems to me a bit of a compromise btw an analytic proposition and a fancy soundbite.


    • “Essentially” might be used just to avoid the self-refutable issue. However, that statement can’t be proven true even if we care only about “purely non-analytic propositions” (whatever it might be). I still think it’s based on a false dilemma fallacy.

  5. how can someone who uses models professionally, seriously say they are wrong? wouldn’t this mean his job is obsolete? so, I don’t think he meant it this way.


    • It means that being wrong doesn’t matters for him.
      That’s another way of saying “truth doesn’t exist” or at least “even if truth exist, we couldn’t ever talk about truth”. These may “sound good”, but they’re self-refutable too.

      That all don’t mean his job would be “deprecated” anyhow, but make him dangerously closer to sophists (that statement is a typical pseudoskepticism [see wikipedia for examples]). But even after everything I said here, I REALLY hope you’re right (remember that I’ve enrolled to that class too…).

  6. let me put it this way: he says: right or wrong doesn’t matter (in the case of models), however they are useful and that is what matters. This seems to me compatible with your conclusion above “It means that being wrong doesn’t matters for him….” – ??

    • I’m not sure of what you’re trying to say now. The above is self-refutable, I don’t see how those ideas could be useful. I still can’t understand why follow such maxim without asking “when?” and “why?”. But, even with the disagreement getting stronger, I’ll accept that as “useful” temporarily. The trouble is: is the quote saying something at all? What is it telling to us?

      In a world where everything is “orange”, it’s unuseful to look for the color of anything. Without a dichotomy (or a “K-chotomy” for some integer K), we’ll be talking about a general property of everything. Such property is meaningless, unless it characterizes your “everything” as a subset of a “bigger everything” (i.e., unless it creates a “K-chotomy” otherwhere).
      And THAT’s the point: when you say that “models are wrong”, I ask you 2 questions: (1) “what is it like to be wrong?” (a.k.a. as “what’s the meaning of the word ‘wrong’?”) and (2) “is there an example of something not wrong?”. If there’s no constructive example of something “not wrong”, I’ll call the word “wrong”, as used in that context (to avoid conceptual change), meaningless.

      But an useful model cannot be wrong: if it were, you wouldn’t be using it. The best model for you is the one you’d call the right one. If there’s no “right” solution, there’s no problem being discussed, besides perhaps some unavoidable events. The core of what I’m trying to say is that the definition of wrong is RELATIVE, immanently, and relative to the problem you’re trying to solve, or to some criteria you defined previously to the analysis (the analysis is the characterization of the model as right or wrong in the given context). When you have a task (i.e., a context), the only chance you have to deny the model you’d use as “right/trusted” is when you give to the “right/wrong” dichotomy ANOTHER context, not the one of the problem you’re trying to solve.

      The starting of the previous paragraph might be seem as based on the idea of “usefulness as a new definition of truthness”, however I think that’s [by far] more general than the G. Box’s quote: the “right/wrong” dichotomy isn’t a property of the model itself. But let’s try to force his quote to have such meaning I described, talking about the existance of context, “is there a possible world where blahblahblah”, and saying that a model is “useful” when there is any such possible context. But that seems unlikely as the Box’s intention, he probably tried to be pragmatic, not general. If we accept the interpretation os usefulness as “general”, the problem goes to the “but some” part of the quote. In that case, which are the models that aren’t useful? Itself?

      I also tried to think in his quote as “all models are context-dependent” (so there’s always a context where we can call it as wrong), but that’s self-refutable too…Ahhh, Box’s word “wrong” seem to have no meaning at all, at least I can’t find one =P…sometimes seems that ALL TERSE STATEMENTS ARE WRONG…but…as you may see, that’s self-refutable too ¬¬

  7. Hi Danilo

    Sorry for replying so late. What have we achieved so far: you made an important point concerning the interpretation of Box’ line. Imho it was important to get this clear. However, I see no point about self-refutation here at all. Since I think I pointed this out pretty clear before, I just wanna say thanks again for your contribution and leave it this way.

    Take care

  8. payday loans says:

    kvbfozvm twngnmru xqkfusvmae uxisxxafi

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s