Monthly Archives: November 2008

On Language And Decision Making

Decision Making

After a week of meeting new people in short spurts of mutual analysis (read: interviews), I am starting to realize that we suffer from an intractable problem: our decision making is only as good as our data and our data is necessarily limited.

Whether this be in the context of making decisions about where to work, what to eat, who to befriend, or what stock to invest in, we are very often evaluating our options.

In order to help us come to the “right” conclusion we have shortcuts like information labels, reputational filters, and other mental models that we trust.

For example, when meeting someone new, one often asks where they are “from” and what they “do”. Although perhaps not consciously, as the responses to these questions start coming back, we immediately start to categorize based on our past associations with these places and professions.

A similar phenomenon occurs when evaluating a stock: we look at the company’s performance, compare it to similar companies, try to put these thoughts into a framework we trust, and then make an evaluation.

We Are Like Machines.

This process – data input, association, analysis, action – is central to any form of decision making, even for a computer.

The beauty of computer programming languages is that the decision-making operates within a bounded system. In object-oriented programming, one can create hierarchies of linguistic complexity, but there literally is no data that could be within the system which is incapable of categorization because the inputs are defined. Of course programs are thusly limited in their scope of operation, but everybody knows this.

You can’t use a Blackberry to transmit temperature, but when I press on the letters of this keypad, the machine “knows” exactly how to interpret these actions and how to respond.

Similarly, we humans are limited by our data inputs and representational frameworks. I can’t parse radio waves, and at least my human brain doesn’t have the capacity to digest 6 dimensional objects.

We Are Not like Machines. Our Language Is More Complex.

However, unlike the programs we create, we humans have a much richer set of tools with which to do our analysis.

I believe it is the linguistic element of this toolkit that is the most powerful.

In other words, because we have a common language, we are able to categorize and communicate in rich and powerful ways. When I say ‘Texas’, the person in front of me is able to grasp ‘greatest State’, ‘humble citizens’, and the various different concepts she has grown to associate with this word. Unlike a computer program, her associations can be loose or strong, they can change over time, and here is the key: they are different for all of us.

It is this aspect of our language, and as a result our decision making, that is both powerful and troubling.

Our Language Is Open At The Periphery.

A simple visualization of what I am talking about comes in the form of a venn-diagram:

Venn Language

When we communicate (and also when we process data on companies, food and anything else), there is an overlap of commonality, but there are also areas of disagreement. For example, if the above diagram represents three different people’s understanding of ‘Texas’, person A, B, and C might all agree that the state is ‘great’, but each might independently associate other concepts with the word.

Unlike a venn diagram, our linguistic frameworks are much more hazy, complex and full of not only linguistic connections, but also emotional and physiological connections.

If you don’t believe that language is affiliated with these more complex responses consider words like: ‘rape’, ‘stain’, ‘caress’, ‘love’, etc.

And if that wasn’t complicated enough, these linguistic structures are influenced by our unique personal experiences and our common historical experience.

An example of the latter is how the word ‘hope’ has become hijacked by the recent presidential election and now contains a number of associations it previously didn’t – for better or for worse.

So What? It Matters For Investing…And Life.

The implications of all of this are that we are left with a process that is at the same time both powerful and imperfect.

It is powerful in the sense that we can theoretically accomplish a lot of communication and analysis in a short period of time.

With the addition of representational and processing aids like the machine I am using to write this message, we can tap into the power of language to reach out and interact with our worlds in ways like never before.

We can use this increased power to make more decisions, about more things, more quickly and more “accurately” than ever before.

However, as a brilliant investor once told me: we should never forget that mental models are not reality…they are just models.

In other words, in a world where we bound along from mountaintop to mountaintop aided by the clouds of our linguistic meta-structures and the tarzan-ropes of our communication aides, we often forget that the system is at best one of approximations and at worst one filled with errors. Part of the reason why our “venn diagrams” don’t completely overlap is because the words we use are incomplete and inadequate depictions of the world we experience.

Do: ‘sunset’, ‘heartbreak’, ‘healthy’, or any of our words really represent the things about which we seek to communicate?

The arrogant research analyst in the crowd might try to argue that stocks are different because there we are using ‘numbers’ instead of words. But that is a massive over-simplification.

The numbers on your screen represent a ‘value’ arrived at by people who, using some common language and their own understanding of the world, crafted representational frameworks of entities called ‘companies’ that are really an amalgamation of people, stuff, and mutual understandings.

These representational frameworks are supposed to encapsulate not only the current state of affairs but also the future trajectory of these various different actors.

If you don’t recognize the inherent inaccuracy in such a system then either: a) I am not being clear (which is likely) or b) you are probably smart enough to be dangerous, but not smart enough to know your limitations, which is a really bad combination.

It is this kind of necessary arrogance that leads us through life making decisions with necessarily imperfect and inaccurate information. We don’t have any other choice.

The best we can hope for is that enough of us will have the humility to recognize these limitations that we will collectively band together in support of our failures as they occur. We can also help one another to communicate and learn about the various different aspects of the world that matter.

The sun is rising on the horizon. A new day awaits.