The Endurance of Research Programs
Part 5 of the essay series Research and Knowledge Accumulation
Euclid vs. Galileo vs. Freud
Some thinkers’ ideas are refined and built on for decades, centuries, or millennia. Other thinkers’ ideas are explored but then abandoned.
Euclid and Galileo are examples of the former. Euclid’s Elements in 300 BC presented a framework for what came to be known as Euclidean geometry. Mathematicians built on this for over two thousand years and are still studying it now. Galileo in 1609 AD used a handheld telescope to make surprising observations about the moon, Venus, Jupiter, and Saturn. This led to the development of better and more powerful telescopes, higher quality observations, and eventually — four hundred years later — the current field of astronomy.
Freud is, or is becoming, an example of the latter. He presented a range of new ideas: the id, ego, and superego; the Oedipus complex; the oral, anal, and genital phases; Eros and Thanatos. His followers studied and modified his techniques, leading to the development of many different schools of psychoanalysis. But now, nearly a century later, psychoanalysis is declining. The average age of practitioners is rising; fewer practitioners are being trained. What once seemed to many people a promising research program is more and more being displaced by other approaches.
Why did Euclid and Galileo’s research programs endure for so long, and why does it seem that Freud’s will not? Let’s first consider some of the general conditions under which research programs endure. Then we’ll look in turn at Euclid, Galileo, and Freud.
To anticipate, we will not find that Euclid and Galileo succeeded because they were right or that Freud failed because he was wrong. On the contrary, we will see that all three thinkers were on to something, but that they differed with respect to how their ideas, observations, and methods interfaced with the logic of knowledge accumulation. Euclid and Galileo’s ideas enabled accumulation, and so research programs based on their ideas lasted centuries or millennia and are still going strong today. Freud’s ideas did not, and so the research programs he inspired are fading away.
The pattern of endurance
Before looking at each specific case, let’s state the general pattern that underlies the development of a new durable research program.
First, a thinker develops and communicates some new ideas. These ideas may be true or false; either way, they must display promise. It must seem like the thinker is on to something and that further investigation will yield fruit.
If the ideas seem promising enough, people will talk about them. People will raise questions, suggest alternatives, form opinions. Unfortunately, ideas do not automatically generate enough intellectual common ground among researchers to allow researchers to fruitfully investigate those ideas together. So at this point, most would-be research programs dissipate.
In some cases though, the thinker’s ideas — perhaps in conjunction with instruments, descriptions of techniques, reported observations, and so forth — do generate enough intellectual common ground. In such a case, the ideas generate an intellectual Schelling point. It becomes much easier for researchers to talk to each other, share ideas, and check each other’s errors. This makes the investigation more fruitful for everyone involved. If an investigation is fruitful enough and the ideas attract enough initial attention, a new research program will be born.
At this point, the research program depends for its continued existence on two things: replenished promise and continued cohesion. If there are no new avenues forward, researchers will move on. If there are ways forward, but nothing that helps bring the researchers back together again, then as researchers fan out across the investigative landscape they will lose their ability to communicate with each other. Again, intellectual common ground does not arise automatically, and the conditions that generated it initially may not persist as researchers move away from the initial Schelling point.
What is needed then is the generation of new intellectual Schelling points, so that as researchers make progress, they can come back together again, share information, check errors, and so forth. These new intellectual Schelling points then serve as waypoints, lights along the path to the truth, that bring present and future researchers together and continue to orient them on the truth as the investigation continues. The result is an enduring research tradition.
The original ideas might be confirmed by this process, or revised, or refuted and replaced. It doesn’t matter that much.
Now let’s look at Euclid, Galileo, and Freud. In each case, we can ask: Did their ideas have initial promise? Did they generate intellectual Schelling points that brought researchers together? As researchers investigated, did further investigation continue to seem promising? Were further intellectual Schelling points generated to bring the researchers back together as they made progress? Any “no” heralds the failure of the research program; the first “no” will shed light on the specific mechanism of the failure.
Euclid — early geometry — the Elements
Around 300 BC, Euclid composed the Elements. In the Elements, Euclid states definitions and axioms that constitute a system of rules for constructing geometrical figures and proving theorems about those figures. Euclid then gives many examples of such constructions and theorems.
The Elements quickly supplanted all available geometry texts. Euclid’s system went on to define geometry for over 2,100 years until non-Euclidean geometries were developed in the mid-nineteenth century. Researchers are still working on open questions in Euclidean geometry today.
Why was Euclid so successful? We can’t say that it’s because Euclidean geometry is true of all regions of physical space. Physicists today accept general relativity, which says that space is non-Euclidean. We can’t say that it’s because Euclidean geometry is approximately true of macro-space. While astronomical measurements seem to indicate that macro-space is flat within the margin of error, those measurements weren’t taken until a few decades ago. Barring retro-causality, that’s over two thousand years too late. We also can’t say it’s because Euclid’s constructions and theorems were logically flawless. There are multiple flaws in the very first construction, at least one of which was known quite early. The axioms themselves weren’t made logically complete until 1899 AD.
Instead of looking at truth-values, which are frequently only established very, very late in the game, let’s look at the factors that contribute to research program endurance. Euclid’s ideas had initial promise; this was a feature shared with previous geometers. What Euclid brought to the table that others did not was a system that created intellectual common ground among geometers and made it easy for future geometers to generate more common ground. Euclidean geometry, with its definitions, axioms, constructions, proofs, construction methods, and proof methods was an intellectual Schelling point and it made it easy for geometers to create further constructions and proofs that themselves were intellectual Schelling points.
Whether you agreed with Euclidean geometry or not, whether it was true or not, and despite its flaws, it brought researchers together and enabled them to share ideas, check each other’s work, and spot errors in a way that continued to orient them on the truth. This, combined with the fact that the discovery landscape continued to have more and more things to find, made it possible for geometers to form a cohesive research program that lasted for thousands of years and continues today.
Now let’s take a look at Galileo.
Galileo — early astronomy — the telescope
In 1623, Galileo used a telescope to examine the heavens. He observed the moons of Jupiter, the texture of the Moon, the phases of Venus, and what he at first thought were small planets around Saturn.
This kicked off a race between astronomers to build bigger and better telescopes, to correct each other’s observations, and to make new observations. This continues today, with the construction of many extremely large telescopes including, for instance, the Extremely Large Telescope, which will cost more than $1 billion, gather 100 million times more light than the human eye, produce images 16 times sharper than the Hubble Space Telescope, and allow us to make detailed studies of many things, including planets around other stars and supermassive black holes.
Why was Galileo so successful? We can’t say it was because his observations were all correct, because some of them weren’t. (After thinking Saturn had small planets, Galileo switched to thinking it had handles. It wasn’t until 1655 that Christiaan Huygens figured out that Saturn has rings.) We can’t say it was because Galileo’s correct observations were widely confirmable or replicable, because very few people had their own telescopes and many initial users of his telescope couldn’t see the same things through it as he could. (Galileo’s telescope was extremely weak by modern standards. It was also handheld, requiring a steady hand and substantial practice to use properly.)
Instead of being right, Galileo’s ideas had promise. It was plausible that telescopes — which people agreed could magnify objects near the surface of the Earth — could give us greater knowledge of the heavens. Galileo’s use of the telescope, and his plausible-enough claims that the telescope revealed things that contradicted existing theory, was enough to bring researchers together and orient them on the discovery of the truth. There was then a straightforward path for researchers to follow: build bigger telescopes and cross-check observations with each other.
It didn’t matter whether Galileo’s initial observations were correct. There could have been dust in the lens and he could have reported eight moons around Jupiter: the history would have come out the same. It also didn’t matter that no one understood how the telescopes worked. (Optics was poorly understood at the time.) What mattered was that Galileo unlocked a clear and essentially unbranching path for researchers to follow. Then, even if the telescopic astronomers sometimes disagreed with each other, they would still have enough intellectual common ground for the research program to endure.
Things turned out differently with Freud.
Freud — early psychology — psychoanalysis
In the 1890s, Sigmund Freud debuted a method of psychological investigation which he soon named “psychoanalysis.” Psychoanalysis involved questioning a person and carefully analyzing their responses. Hesitations, slips, associations, dreams — all of these provided clues that were supposed to allow the analyst to gain access to hidden parts of the person’s mind and determine the real causes behind their actions and feelings.
Psychoanalysis became famous. Many learned it, including notable theorists like Jung, Adler, Horney, Erikson, Rogers, and others. It peaked in the 1950s-1960s and went into decline. Now, it seems to be fading away.
Why is this? Freud was certainly wrong about some things. There may not be an Oedipus complex or oral/anal/genital phases. But Euclid and Galileo were wrong about some things too. We cannot explain the dissolution of a research program simply by noting errors. In fact, it is part of the purpose of a research program to identify and correct errors. If perfect accuracy were a prerequisite, no research program would ever get off the ground.
Instead, we should look again at the factors that permit research programs to endure. Was Freud on to something? As much as his detractors might not like to admit it, it certainly seems that he was. There were strange patterns in people’s reports. Trying to bring unconscious elements to the surface did have surprising and often noteworthy effects. And many of the patterns and structures Freud thought he found had enough resonance that they still inform our understanding of the mind today. In particular, Freud brought us the unconscious mind, analyzable structures, and defense mechanisms; the psychological importance of early childhood, traumatic events, and sexuality; projection, sibling rivalry, and Freudian slips. You, the reader, probably believe in many of these things, specifically because of Freud.
Freud’s theories and psychoanalytic techniques were enough to bring together an initial group of researchers. There was, in this regard, an initial intellectual Schelling point. But Freud’s ideas and techniques worked much less well to help researchers maintain cohesion than did Euclid’s or Galileo’s. It was hard to agree precisely on what exactly Freud’s theories or techniques were. It was also hard to maintain whatever agreement was reached. There were so many different directions to go, so many ways to improve the theories and methods. The result was that Freud generated a large amount of initial interest and subsequent effort, but rather than the researchers staying close enough together to produce the requisite cycle of refinement, the tradition splintered into a thousand pieces, with every researcher having their own theories, their own techniques, and examining different and varied parts of the human psyche.
This was Freud’s problem. Not the non-existence of the Oedipus complex. Not an over-focus on sex. These would be errors that would be worked out during the refinement process, had Freud’s theories and techniques enabled a research program with enough cohesion to permit refinements to accumulate.
Euclid vs. Galileo vs. Freud, recap
These examples illustrate some elements of the logic of knowledge accumulation. In them, we see that success or failure depends not on initial mistakes, but on the shape of the research programs the initial thinkers helped to generate, and the nature of the discovery landscape the subsequent researchers were operating within.
Euclid gave his fellow geometers a system that made it much easier for them to share and check each other’s results. This made it possible for them to stick together, even as they investigated the vast and complex landscape of geometrical truths. It was possible to challenge or change Euclid’s system, but it was much easier to work within it. So geometers mostly worked within it, and their research program retained adequate cohesion.
By pointing a telescope at the heavens and announcing provocative results, Galileo opened the door to what was essentially a single long non-branching hallway. There was one obvious direction to go — make the telescopes better, repeat earlier observations and add new ones — and so it was easy for researchers to stick together.
Freud opened a door to a maze of mirrors and many branching paths (the human psyche) and gave some instructions, tips, tools, and rough sketches of the landscape. This was enough to send a troop of researchers in but not enough to keep them together. The result was splintering, with a vast number of tiny research programs, related but distinct, and researchers less and less able to communicate with each other. With less cohesion, it became harder to share results, check errors, and accumulate knowledge. With less progress, it was hard to maintain momentum and continue to attract new researchers. Thus psychoanalysis’ decline.
A lens on the present
Having illustrated the logic of knowledge accumulation with a few examples, in the next essay we will turn to the state of modern academic and scientific research. Knowledge of successful and unsuccessful cases of research programs will help us shed some light on the present state of affairs.
Next essay: [in process]
Previous essay: Intellectual Schelling Points