People and Machines

Leveraging our impact with technology means certain things. It means substituting machine processes, which are good at certain kinds of thinking, for intellectual processes, which are good at other kinds of thinking.

In terms of “recommender engines” or other systems intended to connect people with information automatically, it means relying on aggregate data and averages. The attempt is to predict without trying to understand (without trying to understand the user, the question, or the resource/object).

Using machines for decisions and judgment means measuring popularity or consensus – quantifiable factors that statistically analyze everyone’s opinions and characteristics. People’s reasoning is not examined.

There are feedback loops in the formation of the opinions and preferences that are tabulated by these systems. Most people do a little thinking for themselves, but are heavily influenced by what other people think and by the impressions that they get from the way information is presented to them, often through the same systems that are polling them for their opinions.

At any given moment, I think it is a relatively small minority of people – not the majority – whose views on anything are well reasoned. They are not necessarily geniuses. It is simply that rationality is a virtue that most people do not care much about, so they are content with lazy thinking. Also, people are busy, and don’t have time to spend thinking about most questions the way an expert has to.

The populist idea of finding the truth in the average degrades public discourse by mathematically privileging such non-thought. Our use of machines is part of the problem because they are so good at computing those averages, and their power in this regard results in more of our thoughts being put back to us in quantifiable terms. For example, a numbers-based system related to movies will try to tell you how “good” each movie is, or how “good” you will probably think it is, on a scale of one to five. It is the nature of the tool that has led to the focus on such unenlightening questions. Questions that don’t have quantitative answers, such as questions about what something means or what we can learn from it, are asked less frequently as we go forward. Their answers cannot be scaled up because they can’t be put in quantitative terms. We answer them in an individual way, through a process of discourse that has a rational element.

Scaling up decisions with technology means bypassing the process of analysis that looks at people, texts, and situations in an individual way and reasons about them. In order to do this, it is necessary to construct tables of rules and assumptions that only apply some of the time, and to only ask questions that can be answered in quantitative language. The result of this is a gradual reconstruction of our shared reality according to this quantitative worldview.

As these systems grow larger, their hunger for data must be satisfied, so we feed them data without putting much care into questions of what the data is actually measuring. I frequently see data sources described in extremely simple terms that positively misrepresent what the data actually is counting. Much, if not most, of the data that provides the computational fodder for these systems is misrepresented by salespeople and marketers, with the result that the answers we get are often determined by preconceived notions rather than sound methods. The very act of determining how things will be counted is determinative of the kind of picture that will be created by the numbers.

The basic problem is that we are gradually turning over our role in the world as thinkers to systems that look only at inputs and outputs and make no attempt to understand what is happening in between, or what things mean, with little or no concern for methodological soundness. As we go down this road, critical thinking, the kind that actually involves reasoning and an attempt to find the truth, is growing rare, despite the lip service we pay to it.

Philosophically, the transition seems to come from a passive decision, accepted gradually and without notice, that “truth” is a meaningless category, that all there is is representation and consensus. The personal search for truth is replaced by the quantitative search for the mean. It is called a “social” transition, but it is important to look at what constitutes the new social fabric versus the old. Is it a meeting of minds, as the word implies? Or is it a way of living that is organized by the principles of our machines, and given over to measuring ourselves against “Das Man,” as Heidegger called it?