The Multistakeholder Meeting Illuminates Various Problems of Objectivity

At the NTIA Multistakeholder Meeting this week, it was suggested that all those studies submitted and funded by an outside source be disclosed. Later, when the issue of objectivity came up again, it was suggested that the group should rely on professors to provide objective views.  While I am for disclosure in practice and not necessarily against the academy, both of the suggestions belie a flawed, but pervasive, vision on the connection between objectivity and research.

Sure, corporate funded research has lead to foibles in the past, but there is nothing to suggest the mere connection to a company delegitimizes it; nor are researchers swayed by the intentions of the companies they work with, despite what some of the most vociferous commenters might say. The reason why an individual would engage in research or advocacy of this sort is because their interests are aligned beforehand. Furthermore, companies are able to give researchers the kinds of technical and financial resources that cannot be found in the academy. Corporate labs like Bells Labs, PARC, and from an even earlier time, Thomas Edison’s Menlo Park helped to invent and bring to market the carbon microphone, the lightbulb, the transistor, the laser, the personal computer, the laser printer, Ethernet, and the graphical user interface, just to name a few. Each of these were hugely expensive inventions that might not have been pursued had it not been for the freedom of their arrangement.

Moreover, scientific work is not value-free. As the Nobel Prize winning economist Gunnar Myrdal pointed out,

There is an inescapable a priori element in all scientific work. Questions must be asked before answers can be given. The questions are an expression of our interest in the world, they are at bottom valuations.

The values of a person will affect the range of problems that they are willing to analyze, as well as the choice of variables to be included, which in turn determines the grounds on which a study will be judged. This helps to explain why objectivity in research of society and business is more complex than physics or biology. These features, however, are not magically down away with when you get funded by the National Science Foundation.

Even the NSF funded work is not foolproof. Just last week, the blog Neurobonkers posted a lecture by Prof. Dorothy Bishop, a highly respect professor of developmental neuropsychology, who detailed how brain scans and other typical neuroscientific methods attaches gavitas to hypotheses that are just wrong. As she went on to explain, studies like this are rife in her field. In other words, people love to believe a wrong conclusion back up by “neuroscience,” especially when graphs and charts seem scientific but can hardly be deciphered.

All of this is to say that funding sources do not necessarily separate out the objective from the subjective.

Privacy policy is especially troublesome, but for other reasons related to objectivity. Most of the research that is trotted out in favor of stricter privacy regulations comes from polling data. However, a poll does not actually express risk, but the public’s perception of it. There is a fundamental disconnect then between actual harm and the perception of harm.

In a classic study on the concept of risk, participants were first asked to consider two causes of death and then estimate the rate of each and ratios between the two. When compared against recent statistics, the estimations were off by factors. Participants said that both disease and accidents were about as likely to cause death even though death by disease is 18 times more frequent. Strokes cause almost twice as many deaths as all accidents combined, but 80% of respondents judged accidental death to be more likely. Tornados were seen as more likely to kill than asthma, even though the latter causes 20 times more deaths.

Daniel Kahneman, winner of the Nobel prize for his work on the subject, later said,

The world in our heads is not a precise replica of reality; our expectations about the frequency of events are distorted by the prevalence and emotional intensity of the messages to which we are exposed.

This is why during the Multistakeholder Meeting my colleague Berin Szoka asked that we rely upon experimental data and not polling data. He is correct in wanting a level headed approach because there is a lot of hyperbole effectively swaying the range of debate. He hopes, as well as I do, that calmer voices will prevail in the debate because there are tradeoffs inherent in restricting the use of personal data in the name of privacy.