by Lee Smolin
When verificationism failed, philosophers proposed that science progresses because scientists follow a method guaranteed to lead to the truth. Proposals for the scientific method were offered by philosophers such as Rudolf Carnap and Paul Oppenheim. Karl Popper put forward his own proposal, which was that science progresses when scientists propose theories that are falsifiable—that is, they make statements that can be contradicted by experiment. According to Popper, a theory is never proved right, but if it survives many attempts to prove it wrong, we can begin to have faith in it—at least until it is finally falsified.2
Feyerabend began his work in philosophy by attacking these ideas. For example, he showed that falsifying a theory is not such an easy thing. Very often, scientists keep a theory alive after it appears to have been falsified; they do this by changing the interpretation of the experiment. Or they challenge the results themselves. Sometimes this leads to a dead end because the theory really is wrong. But sometimes keeping a theory alive in the face of apparent experimental contradiction turns out to be the right thing to do. How can you tell which situation you’re in? Feyerabend argued that you can’t. Different scientists adopt different viewpoints and take their chances on which one will be borne out by developments. There is no general rule for when to abandon a theory and when to keep it alive.
Feyerabend also attacked the whole idea that method is the key to scientific progress, by showing that at critical junctures scientists will make progress by breaking the rules. Moreover, he argued—convincingly, in my view—that science would grind to a halt were the “method’s” rules always followed. The science historian Thomas Kuhn made another attack on the notion of a “scientific method” when he argued that scientists follow different methods at different times. But he was less radical than Feyerabend; he tried to set out two methods, that of “normal science” and that of scientific revolutions.3
Another criticism of Popper’s ideas was made by the Hungarian philosopher Imre Lakatos, who argued that there was not as much asymmetry between falsification and verification as Popper supposed. If you see one bright red swan, you are not likely to give up a theory that says that all swans are white; you will instead go looking for the person who painted it.4
These arguments leave us with several problems. The first is that the success of science still requires explanation; the second (emphasized by Popper) is that it becomes impossible to distinguish sciences like physics and biology from other belief systems—such as Marxism, witchcraft, and intelligent design—that claim to be scientific.5 If no such distinction can be made, the door is left open to a scary kind of relativism, in which all claims to truth and reality have equal footing.
While I’m convinced, like many practicing scientists, that we follow no single method, I’m also convinced that we must answer Feyerabend’s question. We can begin by discussing the role science has played in human culture.
Science is one of several instruments of human culture that arose in response to the situation we humans have found ourselves in since prehistoric times: We, who can dream of infinite time and space, of the infinitely beautiful and the infinitely good, find ourselves embedded in several worlds: the physical world, the social world, the imaginative world, and the spiritual world. It’s a condition of being human that we have long sought to discover crafts that give us power over these diverse worlds. These crafts are now called science, politics, art, and religion. Now, as in our earliest days, they give us power over our lives and form the basis of our hopes.
Whatever they have been called, there has never been a human society without science, politics, art, and religion. In caves whose walls are adorned with the paintings of ancient hunters, we have found bones and rocks with patterns showing that people were counting something in groups of fourteen, twenty-eight, or twenty-nine. The archaeologist Alexander Marshack, author of The Roots of Civilization, has interpreted these as observations of the phases of the moon.6 They might also have been records of an early method of birth control. In either case, they show that twenty thousand years ago human beings were using mathematics to organize and conceptualize their experience of nature.
Science was not invented. It evolved over time, as people discovered tools and habits that worked to bring the physical world within the sphere of our understanding. Science, then, is the way it is because of the way nature is—and because of the way we are. Many philosophers mistakenly look for an explanation of why science works that would apply to any possible world. But there can be no such thing. A method that would work in any possible universe would be like a chair that would be comfortable for any possible animal: It would fit equally badly in most cases.
Indeed, it is possible to prove a version of this statement. Suppose that scientists are like blind explorers looking for the highest peak in their country. They cannot see, but they can feel around them to determine which way is up and which is down, and they have an altimeter with an audio readout to determine how high they are. They cannot see when they are on top of a peak, but they will know it because only there do all directions lead down. The problem is that there can be more than one peak, and if you can’t see, it’s hard to be sure you’re climbing the highest one. It is thus not obvious whether there is a strategy the blind explorers can follow to find the highest peak in the least amount of time. This is a problem that mathematicians used to study, until it was proved impossible. The no-free-lunch theorem, developed by David Wolpert and William Macready, states that no strategy will do better in every possible landscape than simply moving around randomly.7 To fashion a strategy that does better, you have to know something about your landscape. The kind of strategy that would work well in Nepal would fail in Holland.
It is thus no surprise that philosophers were unable to discover a general strategy that would explain how science works. And the strategies they did invent didn’t bear much resemblance to what scientists actually do. The successful strategies were discovered over time and are embedded in the practices of the individual sciences.
Once we understand this, we can identify the features of nature that science exploits. The most important is that nature is relatively stable. In physics and chemistry, it’s easy to devise experiments whose results are repeatable. This did not have to be the case; for example, it is less the case in biology and far less in psychology. But in the domains where experiments are repeatable, it is useful to describe nature in terms of laws. Thus, from its beginnings, the practitioners of physics have been interested in discovering general laws. What is at issue here is not whether there actually are fundamental laws; what matters for how we do science is whether there are regularities that we can discover and model, using tools that we can make with our hands.
We happen to live in a world hospitable to our understanding, and this has always been the case. From the very beginning of our life as a species, we could easily observe regularities in the sky and in the seasons, in the migrations of animals and the growth of plants, and in our own biological cycles. By making marks on bone and rocks, we learned that we could keep track of these regularities, correlate them, and use this knowledge to our advantage. Down to the present experiments with huge telescopes, powerful microscopes, and bigger and bigger accelerators, we are doing only what we have always done: using the technology at hand to discover patterns unfolding before us.
But if science works because we live in a world of regularities, it works in the particular way it does because of some peculiarities in our own makeup. In particular, we are masters at drawing conclusions from incomplete information. We are constantly observing the world and then making predictions and drawing conclusions about it. That is what hunter-gatherers do, and it is also what particle physicists and microbiologists do. We never have enough information to completely justify the conclusions we draw. Being able to act on guesses and hunches, and act confidently when the information we have points somewhere but does not constitute a proof, is an essential skill that makes someone a good businessperson,
a good hunter or farmer, or a good scientist. It is a big part of what makes human beings such a successful species.
But this ability comes at a heavy price, which is that we easily fool ourselves. Of course, we know that we’re easily fooled by others. Lying is strongly sanctioned because it is so effective. It is, after all, only because we are built to come to conclusions from incomplete information that we are so vulnerable to lies. Our basic stance has to be one of trust, for if we required proof of everything, we would never believe anything. We would then never do anything—never get out of bed, never make marriages, friendships, or alliances. Without the ability to trust, we would be solitary animals. Language is effective and useful because most of the time we believe what other people tell us.
But what is equally important, and sobering, is how often we fool ourselves. And we fool ourselves not only individually but en masse. The tendency of a group of human beings to quickly come to believe something that its individual members will later see as obviously false is truly amazing. Some of the worst tragedies of the last century happened because well-meaning people fell for easy solutions proposed by bad leaders. But arriving at a consensus is part of who we are, for it is essential if a band of hunters is to succeed or a tribe is to flee approaching danger.
For a community to survive, then, there must be mechanisms of correction: elders who curb the impulsiveness of the young because if they have learned anything from their long lives, it is how often they were wrong; the young, who challenge beliefs that have been held obvious and sacred for generations, when those beliefs are no longer apt. Human society has progressed because it has learned to require of its members both rebellion and respect, and because it has discovered social mechanisms that over time balance those qualities.
I believe that science is one of those mechanisms. It is a way to nurture and encourage the discovery of new knowledge, but more than anything else it is a collection of crafts and practices that, over time, have been shown to be effective in unmasking error. It is our best tool in the constant struggle to overcome our built-in tendency to fool ourselves and fool others.
From this brief sketch, we can see what science and the democratic process have in common. Both the scientific community and the community at large need to reach conclusions and make decisions based on incomplete information. In both cases, the incompleteness of information will lead to the forming of factions that hold different points of view. Societies, scientific and otherwise, need mechanisms to resolve disputes and reconcile differences of opinion. Such mechanisms require that errors be uncovered and new solutions to intractable problems be allowed to replace older ones. There are many such mechanisms in human societies, some of them involving force or coercion. The most basic idea of democracy is that a society will function best when disputes are resolved peacefully. Science and democracy, then, share a common and tragic awareness of our tendency to fool ourselves, and also the optimistic belief that as a society we can practice correctives that make us collectively, over time, wiser than any individual.
Now that we’ve put science in its proper context, we can turn to the question of why it works so well. I believe the answer is simple: Science has succeeded because scientists comprise a community that is defined and maintained by adherence to a shared ethic. It is adherence to an ethic, not adherence to any particular fact or theory, that I believe serves as the fundamental corrective within the scientific community.
There are two tenets of this ethic:
If an issue can be decided by people of good faith, applying rational argument to publicly available evidence, then it must be regarded as so decided.
If, on the other hand, rational argument from the publicly available evidence does not succeed in bringing people of good faith to agreement on an issue, society must allow and even encourage people to draw diverse conclusions.
I believe that science succeeds because scientists adhere, if imperfectly, to these two principles. To see whether this is true, let us look at some of the things these principles require us to do.
We agree to argue rationally, and in good faith, from shared evidence, to whatever degree of shared conclusions are warranted.
Each individual scientist is free to develop his or her own conclusions from the evidence. But each scientist is also required to put forward arguments for those conclusions for the consideration of the whole community. These arguments must be rational and based on evidence available to all members. The evidence, the means by which the evidence was obtained, and the logic of the arguments used to deduce conclusions from the evidence must be shared and open to examination by all members.
The ability of scientists to deduce reliable conclusions from the shared evidence is based on the mastery of tools and procedures developed over many years. They are taught because experience has shown that they often lead to reliable results. Every scientist trained in such a craft is deeply aware of the capacity for error and self-delusion.
At the same time, each member of the scientific community recognizes that the eventual goal is to establish consensus. A consensus may emerge quickly, or it may take some time. The ultimate judges of scientific work are future members of the community, at a time sufficiently far in the future that they can better evaluate the evidence objectively. While a scientific program may temporarily succeed in gathering adherents, no program, claim, or point of view can succeed in the long run unless it produces sufficient evidence to persuade the skeptics.
Membership in the community of science is open to any human being. Considerations of status, age, gender, or any other personal characteristic may not play a role in the consideration of a scientist’s evidence and arguments, and may not limit a member’s access to the means of dissemination of evidence, argument, and information. Entry to the community is, however, based on two criteria. The first is the mastery of at least one of the crafts of a scientific subfield to the point where you can independently produce work judged by other members to be of high quality. The second criterion is allegiance and continued adherence to the shared ethic.
While orthodoxies may become established temporarily in a given subfield, the community recognizes that contrary opinions and research programs are necessary for the community’s continued health.
When people join a scientific community, they give up certain childish but universal desires: the need to feel that they are right all the time or the belief that they are in possession of the absolute truth. In exchange, they receive membership in an ongoing enterprise that over time will achieve what no individual could ever achieve alone. They also receive expert training in a craft, and in most cases learn much more than they ever could on their own. Then, in exchange for their labor expended in the practice of that craft, the community safeguards a member’s right to advocate any view or research program he or she feels is supported by the evidence developed from its practice.
I would call this kind of community, in which membership is defined by adherence to a code of ethics and the practice of crafts developed to realize them, an ethical community. Science, I would propose, is the purest example we have of such a community.
But it is not sufficient to characterize science as an ethical community, because some ethical communities exist to preserve old knowledge rather than to discover new truths. Religious communities, in many cases, satisfy the criteria for being ethical communities. Indeed, science in its modern form evolved from monasteries and theological schools—ethical communities whose aim was the preservation of religious dogma. So if our characterization of science is to have teeth, we must add some criteria that cleanly distinguish a physics department from a monastery.
To do this, I would like to introduce a second notion, which I call an imaginative community. This is a community whose ethic and organization incorporates a belief in the inevitability of progress and an openness to the future. The openness leaves room, imaginatively and institutionally, for novelty and surprise. Not only is there a belief that the future will be b
etter, there is an understanding that we cannot forecast how that better future will be reached.
Neither a Marxist state nor a fundamentalist religious state is an imaginative community. They may look forward to a better future, but they believe they know exactly how that future will be reached. In words I heard often from my Marxist grandmother and friends while growing up, they are sure they are right because their “science” teaches them “the correct analysis of the situation.”
An imaginative community believes that the future will bring surprises, in the form of new discoveries and new crises to be overcome. Rather than placing faith in their present knowledge, its members invest their hopes and expectations for the future in future generations, by passing along to them the ethical precepts and tools of thinking, individual and collective, that will enable them to overcome and take advantage of circumstances that are beyond the present powers of imagination.