Models as indexing, and the value of Google

There are many arguments for the use of models in economics, including notions of rigor and transparency, or that models can help you to see relationships you otherwise might not have expected.  I don’t wish to gainsay those, but I thought of another argument yesterday.  Models are a way of indexing your thoughts.  A model can tell you which are the core features of your argument and force you to give them names.  You then can use those names to find what others have written about your topic and your mechanisms.  In essence, you are expanding the division of labor in science more effectively by using models.

This mechanism of course requires that models are a more efficient means of indexing thoughts than pure words or propositions alone.  In this view, it is often topic names or book indexes or card catalogs that models are competing with, not verbal economics per se.

The existence of Google therefore may have lowered the relative return to models.  First, Google searches by words best of all.  Second and relatedly, if you have written only words Google will help you find the related work you need, kicks in too.  In essence, there is a new and very powerful way of finding related ideas, and you need not rely on the communities that get built around particular models (though those communities largely will continue).

It is notable that open access, on-line economics writing doesn’t use models very much and is mostly content to rely on words and propositions.  There are several reasons for this, but this productivity shock to differing methods of indexing may be one factor.

Still, it is not always easy to search by words.  Many phrases — consider say “free will” — do not through search engines discriminate very well on the basis of IQ or rigor.


Note well that economists are constantly cheating wirh the words they use to name models and their complements -- stealing cognitive content from ordinary language to which they are in no way entitled. The examples are endless, eg "perfect competition" for a construct in which there is _no_ competition, the constant labeling of elements of formal constructs as "money", "capital", "prices", "economies", "agent", etc which in no way resemble that thing in the world that are appealing to in order to give significance to the formal construct which is in no manner earned or legitimate. The examples are endless. "Utility" does not exist, yet economists constantly talk about it as if it is a thing people strive for and collect, the hypostasizstion of a conceptual fiction which exists in the same way that an impossible object exists, thought economists are so deluded here that most have no idea that their talk of "utility" is strictly nonsense.


This reeks of lowest common denominator Internet Austrianism. May as well be copied and pasted from for its clumsy (and *mostly* inaccurate) caricature of modern economic modeling.

You are a moron, Will, with a moronic ad hominem response with zero cognitive content.

You were clearly never taught the value of an "as if" proposition or Occams Razor.

you are hand waving and saying things that utterly fail to grapple with the issue at hand. Try again.

It grapples directly with your moronic criticism of utility models. Those models are useful because they accurately predict and explain human behavior. It's not at all necessary for anyone to actually maximize some measure of utility. This demonstrates your complete lack of understanding of what a "model" is. This is so fundamental as to put everything you say into doubt.

What you say about is completely false and what you say about "utility theory" is misleading at best and fails utterly to address my remarks.

Cognitive FAIL.

I note that you dont explain what is false or misleading. You are obviously a troll.

How exactly is perfect competition the absence of competition?

If you're responding to Greg Ransom, I think you misread him, reread.

In everyday language, "competition" arguably means striving to outdo others. Under "perfect competition," in contrast, no single individual or firm thinks what they do will affect the market outcome.

That's a perfect example of equivocation.

Competition, in the economic sense, is short for PRICE competition. Your description belies the fact that you are confusing equilibrium with disequilibrium conditions. Even if prices cease changing (in equilibrium) doesnt mean the competition has ended anymore than an arm wrestling match has ended merely because the hands are not moving to one side or the other.

Clearly there are other forms of competition, but the simple PC model obviates those with a homogeneity assumption. No one ACTUALLY believes there are PC markets. The simple model enables us to see the dynamics of pricing and quantity while holding confounding factors constant. These are USEFUL lessons even if the world is more complicated.

Just about the most facile, and most ubiquitous, critique of any model is the realism of the assumptions. No shit theyre not real. The entire corpus of research is about stressing and stretching those assumptions to see how robust the results are. This requires a literature review beyond a principles textbook.

"Perfect competition" does not refer to price competition, it refers to quantity competition where firms take the market price as given and select their output to maximize profits given the price.

Your arm wrestling analogy is striking, but it is not appropriate in this context. The perfect competition model starts with the assumption that firms take the market price as given.

Two econ Nobelists that you should admire: V. Leontieff, 1973 Nobelist (I/O analysis) and Kantorovich& Koopmans, 1975 Nobelists (Operations Research). They don't make them like that anymore, nowadays it's all mushy "behavioral economics", aka "stuff happens".

Bonus trivia: the Jordan-Gauss matrix inversion method is ingenious! I bet it's named after the famous Johann Carl Friedrich Gauss.

Two econ Nobelists walked into a saloon with a sign on the window: "Martinis Ten Cents." Nobelist #1 orders a martini. Nobelist #2 happily snacks on beer nuts; saying nothing.

Nobelist #1 asks the bartender how he can charge only ten cents for martinis. The bartender responds that he won the lottery and it was always his dream to own a bar and charge minimal prices.

Later, the bartender asked Nobelist #1 about #2, "What's with him?" Nobelist #1 responded, "He's waiting for happy hour."

Adapted from a "walked into a bar" joke seen on Google.

It's an intriguing insight, but I don't know how powerful it will turn out to be. The earlier arguments about rigor and transparency -- forcing authors to clearly and unambiguously define and display what they are talking about -- remain important strengths of models. If I had to guess, the indexing function that Tyler has described is a secondary advantage and thus google's displacement of models will be limited. Quite possibly significant and interesting, but ultimately limited.

It seems like in Academia declaring models as foundations to your thinking is a kind of tribal virtue signalling. You can call it indexing but it's really a friend/foe litmus test or doggy leg lifting.

A model is a compressed predictive isomorph - it extracts key features (compression) from the territory and creates a symmetrical linguistic map (isomorphy) while claiming to be able to tell a story about future states of the system with some degree of confidence (prediction).

A model can also be generative, a jumping-point for future patterns that are not so much a matter of predictivity but rather a matter of creating workable artefacts that extend or synthesize current configurations.

An mp3 recording is a compressed isomorph but is not predictive or generative - inferring a theory of music from listening to 10 orchestras leads to a mental model - composing a new orchestra is generative use of a model.

My 2c, anyway.

I think TC has a bit different mental intuition about mental models: his mental concept of a mental model seems to be more along the lines of Daniel Dennett's "intuition pumps" - tools that help jump-start thinking. You could (perhaps) formalize this view as models being linguistic labels that point toward a certain Markov chain (of related concepts / associations) as well as provide a vector for exploring the chain.

Interesting, for sure.

"In essence, you are expanding the division of labor in science more effectively by using models."

I would put it like this: high-IQ cognitive styles throw-off mental models as artefacts of their thinking process. Mental models are the breadcrumb trail of great minds throughout history - each model being a dimensionality reduction and composition of certain elements using elements of analysis, synthesis and semantic associativity.

Thoughts? Kind of abstract, struggling to put into words what I mean exactly.

I'm agnostic on models. Used well they can sharpen your definitions and generate a common language for debate. Of course, precise writing can often do the same.

Bad models make a few assumptions, carefully select data, etc. Make an assumption here, another one there, pretty soon you're in your own little world. Simplification to isolate something that you're trying to study isn't inherently bad. However unnecessarily complicated models frequently just hide a lack of insight.

The ultimate test is, does the model work to explain the real world?

Even a simple demand and supply curve is just a snapshot of a moment in time. A freezing in time and place of a confluence of factors which allows you to change one element at a time and pontificate on the impact of that change. For participants in the game, the demand line may look more like a scatter plot; they can only estimate where it is (or was or will be). The economist looking from the outside can differentiate between movements along the curve or shifts in the curve (and their causes), but participants in the game may have a harder time telling the difference in real time. (And what happens next can depend on those perceptions.) That doesn't invalidate the simple model's power as an analytic tool to understand how the real world works. This simple model gives powerful insights into real-world issues. But when you draw a straight line in economics, like the Marginal Revenue line, perhaps keep in mind that the "line" is the first derivative of the total revenue curve and may only be straight over a relevant area for a moment frozen in time. For the model, the simplification of a straight line doesn't weaken the argument, and most of the time it works fine.

So good models can help communicate an idea or concept to others and that when placed in the real world works. While a bad model often tries to build a world where some events can occur, but that world has yet to be discovered.

It helps if we think of mental models as tool - tools of the mind.

Take reality X, extract information through the optical lens of model Y (selective zooming in/out as up/down axis and calling in related concepts as defined by the model's Markov neighbors as peripheral vision), then mentally rotate/translate/operate-upon the extracted reality, then transpose it back to the real-world. If successful, you have a good tool / mental model.

"Models are to be used, not believed." Wise words from Henri Theil. I've forgotten the class or prof who quoted Theil, but apparently it's a legit quote, on the internet anyway.

Verbal reasoning, as employed by the Austrians, is indistinguishable from a rigorous model when done RIGHT. This is because every written argument can be expressed in mathematical form.

The problem is that verbal reasoning is always so clouded with rhetoric, superfluous words, equivocation, and other forms of deception that it is practically unreliable.

Using Google searches does rely on models, but they're merely poor ones.

You see those things as deception - which they may be depending on the intent and rigor of the speaker - but isn't formal reasoning and "evocative semantics" as surface to depth - ie both are legitimate cognitive modes with differing teloi (at least if you want to talk about the world, not just your analytic model).

Formal reasoning wants to put boundaries around semantics so that truth-values can be stacked and a chain of causal probability and necessity established.

Semantic synthesis wants to flow from scene to scene, opening up possibilities rather than drilling down to the one truth.

I suspect this maps to male/female, and differing evolutionary goal-programs.

I'm reminded of Nietzsche here - who refused to speak "academese" and instead wrote in poetic terms, to convey what he wanted to express.

Ironically, your well written, thoughtful, and eloquent comment proves my point. It is so rich with content to be nearly impenetrable even to a well educated person.

I do grasp the gist of what you are saying, and I agree. Whether or not an approach is "deceptive" would depend crucially on the intent of the writer. But because much writing of the sort we are describing is meant to inform policy, anything that doesnt get down to brass tacks is deceptive.

Because formal reasoning is constraining, this is what provides discipline to decisionmaking. You can justify any decision you want by rigging the process. No one likes to be constrained, certainly not politicians. But the side effect of semantics is sophistry and bad policy.

If your goal is to win the game for your team, semantics may be effective. But it also leads to very poor decisionmaking. As an example, in economics we strive toward efficiency. Of course, efficiency is but one objective function. You might value equity or fairness. This could lead you to support and achieve welfare destroying policies such as minimum wage. This policy has winners and losers, both short and long term. And those winners might be part of a majority of the electorate. Political success! But the policy could lead to a slow economic bleed that hurts the very people you intended to help when, say, their jobs are replaced by machines.

And thus far I have given the hypothetical sophist the benefit of good intentions. What if they are truly nefarious?

Facts, logic, discipline, rigor, and peer review not only support good decisionmaking but protect against evil. Ulterior motives get laid bare in a model.

Stream of consciousness symbol-stream incoming:

Here is my model of Western thinking:

You have three axes of thinking. The first one is the Logic of Discernment, which you can think of as an optical lens. It goes two ways Analysis and Synthesis. Analysis subdivides and selects, whereas Synthesis builds up. It's basically trying to compress Territory into a Map - symbolic representation.

The problem with this mode is it discarding mechanism - what you might call the problem of there being infinite confounding variables.

The first logic is simply the art of creating symbol-streams that zoom in and out of reality, selecting variables, discarding unknowns and creating trees of truth-stacking - it's really rather like music and its predictive power can be thought of as in kind of the same way - theory-system X with such and such geometry of symbolic chain maps well to theory-system Y with a compatible configuration but not to theory-system Z which is based on a different ontology (a different way to "slice reality" into symbolic mappings).

The second mode is the Logic of Associativity - metaphors, parables, mythology, so on. This logic is predicated on connections and open interpretability - where the first logic seeks to discard unknowns to gain truth, this one discards truth to expand possibilities.

A metaphoric tale is really a symbol-stream that seeks to increase brain connectivity (at the fMRI level, I'm sure that pretty much what it does - squares with my phenomenological intuitions).

These two modes of thought explain most of Western thinking - symbol-streams that promote either discernment or associativity.

To do anything interesting - you have to use both - chaining together Discernment and deciding what's relevant with Associativity.

(Geniuses are said to have extra high Associative Horizon - ability to associate more widely - which explains why they can come up with novel solutions where others are stuck).

The power of both modes is really infinite, since you can stack truth-chains and you can make metaphors of metaphors with meta-metaphors on top of that. So we have symbol-systems with infinite potential operating in a finite human nervous system. Which leads us to the third mode.

The third mode is the Logic of Transcendence. You can call it Mu-logic, from the Buddhist concept of "null" or "unask the question". It is basically the axis of perspective - switching from the I-perspective to the objective Being of the entire universe. It is what religious iconography and rituals are trying to "talk about".

This is why Kant was so influential in philosophy - he formally introduced a third category beyond Map and Territory - that of Noumenon or the Radical Other - the ineffable transcendent. He made philosophy theological in a way - leading, I think, to Nietzsche's ideas about the "death of God" and other things like that.

Theology can be seen as the attempt to codify third mode logic into symbolic representation, for instance. Due to the limitations I pointed out above, it won't ever reach a final conclusion however.

The third mode also lays bare some of our assumptions about time, entropy and causality. I think "logic" is simple the perception of time and entropy at some point, it's basically an aesthetic discernment that compares two entropy-sets to see which "caused" which. If you start to look from the objective POV the universe all of this starts to break down. But that is something I need to flesh out more in my mind.

Final note: Western symbol-thinking is fairly limited and there are many (infinite?) varieties of logic - think of "light logic" for instance (diffraction, shadows, colors, ) - how can you even begin to represent that with the limited bit-set of symbolics?

Non-symbolic logic is something we as a human race need to explore more - if we want to break out of the symbol-box and truly get some answers.

"Facts, logic, discipline, rigor, and peer review not only support good decisionmaking but protect against evil. Ulterior motives get laid bare in a model."

Yes, correct. It is a good praxis, similar to a music theory that produces beautiful orchestras . It is not the be-all end-all, of course, one can always go further in discernment rigor and semantic complexity.

I can imagine argument-mapping, even more distributed peer-review and perhaps more extended use of meta-synthesis (qualitative analysis across an entire field basically) to be some tools that could improve on the current praxis, for instance.

End of stream!

What we are doing is casting problems and providing explanation -- exactly what Darwin does in The Origin of Species.

And because you have the task wrong you are obsesses with the wrong thing, which distracts you and takes your eye off the prize.

Read Larry Wright on explanation and teleological and functional categories and we can begin a substantive conversation.

Stipulations that miss the point don't help.

Every word in this post of yours is so vague and generalized as to be completely unintelligible. I have no idea what "what we are doing" you are talking about. You dont explain what "casting problems" means and why this is wrong or how Darwin missed the mark.

Your second sentence is entirely gratuitous, given the vagueness of the first. You might have well just typed, "You're wrong, all wrong."

You then refer to some obscure writer who has obviously put you into a froth. This is equivalent to responding to this discussion by telling us to read the Bible.

And finally you dont explain your vague reference to stipulations.

You are a sputtering mess.

@Greg Ransom:

Is this a good starting point for Larry Wright?

Also could you put into words the difference between functional and teleological explanations. Does it have something to do with attempting to finalize the "why" of a system vs merely attempting to expand the story of "how" and "what"?

While we're on the subject of "free will," here's a great explanation why the question of "free will" might be a nonsense question:
Excerpt: "What matters isn’t whether there’s free will or not, the real question should be how differently would I live my life if I knew the answer."

Free will is a misnomer because it tries to refer to something that doesn't have any real semantic substance. We have a will, yes, but what does it mean that it is "free"? Free at what level of system? Free to override passions? Yes, obviously. Free in the sense of being uncaused? Possibly, but does that make it "free"?

I think the semantics are standing in the way here. A functional description is that some field of qualia is following around a central nervous system which does certain operations with - among other things - symbolic inputs, and causes the body to move around in certain ways based on that. Occasionally there is the phenomenological phenomenon of identification with thoughts which causes a weird tunnel-like kind of state of being where one is not really awake but rather being led around by a stream of thoughts. This is what gets called "self" in the West, and to shoe-horn the concept into various theological systems the notion of "free will" had to come along as well.

Read "The Ego Tunnel" by Metzger.

I think it's likely that the human nervous system is the locus of some kind of causal trickery - whether that involves pulling information from the future (retrocausality?) or whatever it may be, I don't know. But semantically this is not free will, just causal weirdness.

Free will would have to operate at the level of universe to make sense - uncaused action that doesn't follow any kind of pattern, just does what it wants.

This post confuses internal mental dialog with (external) communication. Gosh, who knew words were so useful? Turns out that they can be used as "indexes". Wow! Also: can anyone explain WTF his last sentence means? "free will" as a phrase doesn't discriminate based on (its?) IQ nor (its?) rigor?? No, no. he's gotta be saying that the search engine he is using is unable to "discriminate" its hits on the basis of assigning IQ or rigor to the hit (paper? video? talk?) (and then ranking them using that value/vector). Shocking!
Who knew? As an aside, I feel lucky if the results returned actually include the word(s) I used to initiate it. Google is bad, and is becoming worse, at actually doing this (i.e. performing as expected). Becoming more like my ex-wife: you know - acting as if she can know my mind/intent without any attempt to validate/verify her internal model.

Typing words into the google search box uses a model owned by google to determine results. Unless you use strong anonymization it also correlates with your own search history and other available data. The relations are made, not found.

Comments for this post are closed