The following is a model of how scientific knowledge is produced: groups of scientists investigate and collect data about the natural world through observation and experiment, they interpret this data by putting it into conversation with larger theoretical issues and questions, and if this interpretation is seen to be valid, these results are accepted by a larger scientific community. My claim is that scientific knowledge production is fundamentally precarious, because the wider scientific community can always question the data these scientists have gathered through their investigations, and can doubt the ways these scientists have interpreted this data. To put it in words that will soon become clear, scientific knowledge is public, solidified in communities. But it is created through a private process of experimentation and theorization. This contradiction of creating public knowledge from a private process is one that, as we will see, has shaped the way scientific activity is done.
To see this, let’s go to Shapin and Schaffer’s excellent book on the beginnings of experimental science, Leviathan and the Air Pump. They trace the birth of experimental science to Robert Boyle, who grounds the kind of knowledge experimental science produces in matters of fact. A scientific community establishes matters of fact in the following way: First, a scientist or group of scientists create an empirical experience, through experiment or some other means, from which they form beliefs; second, they need to assure others in the community that their rationale for forming these beliefs are adequate; and third, if enough people in the scientific community come to hold these beliefs, then they are codified into matters of fact. But this last part – the codification into matters of fact – is fraught with issues, as Thomas Hobbes’s critique of experimental science shows (yes, this the same Hobbes you read in your Introduction to Political Philosophy class).
Hobbes claims that because experiments are performed by a small group of scientists instead of the whole community, and there are multiple ways to theorize the data experiments produce, experimental science is essentially private. In contrast, geometrical reasoning – on Hobbes’ view – provides a firm foundation for public knowledge. Geometrical proofs start with a socially agreed upon set of axioms, and proceed by incontestable method to irrefutable knowledge. The method is incontestable because it is built around a public feature of our humanity – human reason – and the exact steps needed to proceed from hypothesis to conclusion are visible to all in the form of proof, so the conclusions are irrefutable.
Abstracting a bit from this, we might say that a public process of scientific knowledge production has two important features: its members discuss and produce transparent experiences that can be seen equally by all, and they share a universal capacity to theorize and form beliefs about these experiences, thereby minimizing disagreement. We can use these categories – transparency and universality – to derive two worries that place pressures on the form of scientific knowledge production. The first worry is that the scientist’s collection of facts is improperly done. This places a sociological pressure on experimental science to become more transparent. The second worry is that the beliefs the scientist has formed are not appropriate given the evidence she has available to her. This places a sociological pressure on experimental science to develop more universal methods of belief creation.
To get a better sense of how these sociological pressures have helped determine the form of experimental science, let’s play with them a bit. What does a scientific community that is largely private – without transparency and universality – look like? Without transparency, we have that only a select group of researchers in the scientific community has access to the relevant observations. Moreover, without universality, we have that only a few select researchers have the ability to appropriately theorize these observations. Lorraine Daston and Peter Galison in their work Objectivity seem to describe such a state of scientific affairs when they write about “truth-to-nature” naturalism. Here, naturalists were sage-like figures who spent their life studying what was essential or typical about a certain class of natural objects, such as seashells or skeletons. What was shared amongst the scientific community were these idealized drawings, representing not what the naturalist actually saw, but their interpretation of “underlying types” between the class of objects being studied. So, the larger community did not have access to the phenomenon being studied as they weren’t transparently available, and even if they did have access to these phenomenon, only the naturalist who had spent their life studying them had the epistemic authority to interpret them – so the capacity to appropriately theorize the phenomenon was not universally distributed.
But private processes of belief creation don’t provide a firm ground upon which to create public knowledge, so there was some push-back to this state of affairs. As Daston and Galison write, the “truth-to-nature practices of selecting, perfecting, and idealizing” as they were seen to be the “unbridled indulgence of the subjective fancies.” This “fear of the naturalist” – that the scientist is reading into his results and picking out ones that fit his theoretical expectations – cast doubt on the scientific knowledge that was produced. They describe the push-back of this to be “mechanical objectivity” – a dialing up of the transparency of natural phenomenon. Instead of publicizing drawn and idealized images, there was an impetus towards photography, to produce images “untouched by human hands.” We see a similar impetus in modern scientific papers in detailed methods sections that provide extensive and detailed reports of the experimental procedure and what was observed, as well as occasional pictures of the experiment and apparatuses used.
The social sciences take this even further with the general insistence that researchers let the “data speak for themselves.” Moreover, this push to make the knowledge produced seem objective – uninfluenced by the “subjective fancies” of individual researchers – also in part explains the growth of statistical tools and mathematical language in the sciences, which lend some of the public nature of Hobbes’s geometrical reasoning to the kind of private reasoning scientists do. Essentially, we can come to see the development of various forms of distributing scientific knowledge – such as photography, detailed methods sections, and statistical approaches – as a way to get around the sociological pressure imposed by a lack of transparency: fear that the data the scientist is working with is improperly collected.
However, dialing up transparency in this way has its own problems. If what a scientific collective wants is to create scientific knowledge, it is not enough to make the natural phenomena that lead to belief formation as transparent as possible. Without universal methods of belief formation, this will lead to even more confusion, as different portions of the scientific community will claim direct access to the relevant facts, but without a set method by which some interpretations can be shown to be better than others, the kind of communal agreement necessary for creating matters of fact will not exist. We can see “trained judgment,” the next formulation of scientific knowledge production that Daston and Galison theorize, as a way of dealing with this conundrum. Trained judgment dealt with the crisis presented by mechanical objectivity by developing social structures that helped create universality. Now, trained judgment did not display the transparent nature of mechanical objectivity, as scientists did not hesitate “to enhance images or instrument readings to highlight a pattern or delete an artifact.”
But the increased role of the scientist in interpreting the objects he was studying did not descend into the subjective worry posed by the mechanical objectivists. This worry was resolved by a “newfound confidence” amongst scientists that was “born in professional training,” mediated by social structures such as universities. This universalized and standardized training meant that the scientist was now trusted to use his intuition and judgment to “organize experience into patterns.” But this intuition was not purely subjective and individual to the scientist, as it was honed in apprenticeship: all scientists learned how to think, process, and intuit the natural world in similar and universal ways. Trained judgment should remind us of current scientific practice: scientists get PhD’s where they learn the standard practices and discourses of their field, then move on to get apprenticeships and Post-Docs, and have to show that they can produce work that meets standards of the larger scientific collective they are part of. Moreover, this work is increasingly distributed in peer-reviewed journals, where both data collection and data interpretation are reviewed by others in the field. So, scientific knowledge is increasingly vetted, and as a result the worries I posed above become less troublesome.
But will this central contradiction between public and private always trouble scientific practice? The radical point that Hobbes seems to be suggesting is that technologies of publicity – such as university degrees, or detailed photographs – are only partial solutions, bound to fail: do we believe him to be right, or are there a set of such technologies under which scientific knowledge production becomes truly public?