Thomas Kuhn, 1962

• Idea that most science is about producing novelty that is not unexpected, to get the unexpected is to have failed
  • Scaling laws is an easy example of this, but also the design of processes, equipment, and other applications of paradigm
  • Normal science is for boring people, or maybe for value capturers? For those doing “mop-up”
• Ephemerides are the tables with trajectories of celestial objects. Kuhn notes computing ephemerides as significant but spurned because of the tediousness and lack of novelty in the process
• He has some fixation/weirdness with ideas about what a scientist is? Specifically about concern with understanding the world, looking for disorder and applying a process to do empiricism
• Pre-paradigm period is about debate over what constitutes a legitimate method and what the rules to the puzzle are and figuring out what the problems are.
• Has this analogy of paradigm articulation being like a puzzle sometimes, in that it has a solution and has many parts
  • I dislike the framing of puzzle, eg we (many in the field + Demis) treat AGI as a puzzle but it has well-defined pieces and yet a poorly defined solution. Also he cites cancer as something that may not be a puzzle, because it may not have a solution but I think this is an example of the “puzzle” model not having any applicable bearing
• Quantum mechanics (he also cites Newtonian dynamics or electromagnetic theory) is a paradigm used for multiple fields but is not the same paradigm for each.
  • For example, whether a helium atom is considered a molecule (yes for chemists, no for physicists)
• Distinction between discovery and invention, fact and theory are exceedingly artificial <3.
  • Discovery is an “extended episode”, ending when the paradigm theory is adjusting so that the anomaly (necessary precursor to discovery) becomes expected.
  • Kuhn also claims that adding the fact into the theory is not sufficient, but the scientist needs to really internalize it for the discovery episode to complete, I think this is only slightly true?
  • In discussing discovery, his prime example is the “discovery” of oxygen, of which there are three possible claims. C.W. Scheele had work that was published after the discovery elsewhere was announced, Joseph Priestly sort of identified it but stuck with phlogiston theory and then Lavoisier followed up on the Priestly experiments and actually concluded that the gas from the Priestly experiment was distinct
    • I do think there are cases where discovery is discrete and instantaneous, eg the transformer architecture but I guess it lacks Kuhn’s requirement of being prefaced on anomaly
  • X-Rays as accidental discovery, which also fits the pattern, as many must’ve seen the evidence of x-rays through the numerous cathode ray tubes. Unlike the discovery of oxygen, the anomaly was not that the theory of the paradigm (phlogistons) was incompatible but that it was deeply unexpected because they expected the equipment to work a certain way, and the presence of x-rays meant they failed to control an important variable.
    • Importance of instrumental and theoretical expectations were also sort of in the oxygen discovery, via the process they used to test the “goodness of air”
    • I don’t really understand what the variable was or what the equipment expectation was here
• There’s a psychological experiment where people are shown invalid cards (red spades) and with a short exposure, people don’t notice anything wrote and with longer exposures people not the “anomaly”. Kuhn claims that precision and good retrieval to identify the anomaly is an important skill.
• Introduces Copernican astronomy as a crisis, following the Ptolemaic system (which was things go around the Sun, actually) even though the Ptolemaic system mostly worked, but was never quite perfect
  • People starting globally recognizing the failure in the early 16th century,
  • Kuhn describes the crisis as breakdown of the puzzle solving activity 😒
• There was crisis with phlogistons in chemistry, where by the time the Lavoisier work was done, people were kind of stuck with too many anomalies.
  • “perhaps phlogiston had negative weight” or “fire particles or something else entered when the phlogiston left” re: a weight gain post combustion/oxidation
• Strong claim that to achieve a new paradigm, there has to be a crisis. Cites that people did have the ideas for oxygen and Copernican astronomy before the revolution but they did not take because they were insufficiently in crisis mode.
  • Aristarchus (astronomy) and Rey, Hooke and Mayow (gas)
  • This seems to be an old science problem, where now we have new paradigms sans crisis (I think, deep learning is one of these, as are many areas of biotech).
  • This idea also leaves us hanging on how we make progress if we fail to encounter crises — perhaps physics is there
• Wolfgang Pauli had a personal crisis moment where he felt Physics was “confused” and wish he hadn’t done it but felt better after reading Heisenberg’s paper on matrix mechanics felt that it was “again possible to march forward” 🥺
• The transition of paradigms can look a lot like pre-paradigm science, and it’s also the case that a paradigm is not quite dropped until the new is discovered (scientists see anomalies but don’t throw away their theories just yet, maybe normal science can save them). Also the case that for some evidence, there are usually multiple viable theories so in the crises stage, paradigm candidates compete.
  • Makes a bold claim of “to reject one paradigm without simultaneously substituting another is to reject science itself”, as to avoid being seen “as a carpenter to blames his tools”. This seems bad, seeing as how the practices/theories/tools of a paradigm can be epistemically hindering, tearing away as much as possible seems beneficial.
• Not all anomalies are studied, or equally pressing, or lead to crisis. How are they prioritised?
• “extraordinary science” is what Kuhn calls the science that seems to be of the next paradigm without the crisis? Seems like he’s claiming they eventually take off at a crisis point anyway.
• At the time, the idea that things were not continuous improvement (but that previous paradigms had to be torn down for new ones in the process of revolution) were really rare and he spends a lot of time talking about this
  • Lots more phlogistons, but also corpuscles
  • Talks about Moliere? French playwright who in The Imaginary Invalid talks about sleep-inducing properties of opium from virtus dormitiva (sleep essence).
• Paradigm shifts as genuine worldview changes (ties in with idea that the shift has not completed until it’s fully internalized)
  • There’s a distinct midpoint, where the scientists acknowledge the new paradigm as superior but still frame things via the old paradigm, with Kuhn’s example being “I used to see a planet, but now I see a satellite” to “I once took the moon to be a planet, but I was mistaken”. I kind of feel like scaling is stuck here, where people believe scaling laws and perhaps overly seek cheats around it
• Effluvium theory (Lucretius), there’d be an efflux of particles going out that attract things. Tried to explain magnetism. Kuhn is most outdated in that the sciences he studies were about more concrete incorrectnesses than we currently work with, and so his ideas about worldview changes and puzzle pieces don’t land super well.
  • Still need to find a Kuhn-branded review of biotech, which I think might have the nuances of direction over theory as a paradigm, the underlying technologies used to do science and other specifics of pre-paradigmaticness and extraordinary science
    • Basically now a lot of sciences are much more empirically done rather than theory? Theorizing about existing known behaviours is somehow harder to do than producing experiment to upgrade or confirm existing theories
• Describes psychology as so untheorizable and presupposing of paradigm (duck rabbit)