Science Cannot Count to Red. That’s Probably Fine.

notes on Kuhn and relativism


coming from Science Under High Modernismvaguely related toEverything is Going According to Plan.


Kuhn is most interesting for examining the uneasy relationship between politics, science, and philosophy, but that’s going to come next time. First, I should address a question that I keep getting asked. I assume it’s the same for anyone who writes about Kuhn:

“Was Kuhn a relativist?”

There are two questions bound together here. The first is whether or not science progresses. The second is over relativism. Kuhn’s answer to the first is his answer to the second. You can deny that his answer to one is a satisfying response to the other, hence the separation.

Towards the end of The Structure of Scientific Revolutions, Kuhn points out that his interlocutors all claim that science advances towards Truth-capital-T, but there’s very little reason to assume such a thing happens. After all, we don’t exactly know what we mean by that, making it somewhat hard to tell if we’re on the right course:

Does it really help to imagine that there is some one full, objective, true account of nature and that the proper measure of scientific achievement is the extent to which it brings us closer to that ultimate goal? If we can learn to substitute evolution-from-what-we-do-know for evolution-toward-what-we-wish-to-know, a number of vexing problems may vanish in the process.

It’s certainly correct to say that science advances (though you’d have to emphasize it does so in stages) but it’s not necessarily advancing “towards” anything. After all, you don’t know that there are anomalies beforehand – if you already did know, you’d have entered a crisis period already. To say that science is advancing “towards” truth presupposes that we know there will be a stopping point, that it’s on the right track to get there, that there’s an interpretation of the end point we’ll agree on.

This is a pretty basic argument, and it’s not really his main one. His own is much better, but it’s a little harder to conceptualize.


Imagine that you’re playing a really boring computer game, the purpose of which is to fill all 100 squares of a 10×10 grid. Clicking on the screen fills the next square, but there’s a catch. Your cursor moves through a random assortment of colors,  and you can only advance when you click with the correct one. Every time you click with the right color, it fills the next square,  and it also changes the color of all-previously-filled-squares. So you watch the colors cycle, finally click blue, and the first box fills up with blue. Then you wait, try a few colors that don’t work, and finally land on red. This fills the second box, and both boxes turn red. Same thing with the third, which turns green, and so on and so forth. The correct color is probably randomized, you’re uncertain, there’s definitely no pattern you can determine. Sometimes they even jump back, so clicking forty-seven makes all forty-seven as red as click two.


If someone asks you, “Does the game progress towards 100?” then you’ll probably say yes. Even if you’re half-way through, still trying to get 52 filled, it’s a reasonable assumption. There’s a continuity there: 1-2-3-…-100.

What if they ask you, “Does the game progress to red?” If you’re at 52, then you have no idea – last box could be cyan, could be black, could be anything. Even if you’ve completed the program and on click 100 all the boxes do, in fact, turn red, it’s still an odd question. The numbers build on one another in a clear way, but the colors are just different. Moreover, box two has to follow box one, but the colors might be randomized. You got green for three, but there’s no reason to suppose that it was necessary to have gotten green. Maybe it was just the luck of the draw. Maybe the program was actually asking for the correctly timed click, but you assumed that it was based on color

For Kuhn, the numbers are puzzle-solving and “amount of puzzles that may be solved.” The colors are “truth,” here considered as Truth-Capital-T, the ontological reality of all realities, whatever the paradgim says is “what nature really is.” The color changes are paradigm shifts, they allow for movement and they change everything before them, but they aren’t continuous. There’s no clear pattern from one to the other. We might get tricked into thinking that there is (1 is blue, 2 is red), but if we want to be rigorous then…

The point of the metaphor is that puzzle-solving continues and progresses, but ontological underpinnings are not easily compared. In Kuhn’s famous phrasing, they’re incommensurable – from one paradigm the problems of the other look different, don’t quite match, don’t really work. Hence, discussing progress as anything more than “amount of puzzles” is almost certainly a bad idea. It’s much like saying, “Indeed, the game tends towards red.” But in what way? Certainly not in the same way as it tends towards 100.

In an afterward, long since accused of being Peasant King of PoMo Mountain, Kuhn clarifies his stance:

Compared with the notion of progress most prevalent among both philosophers of science and laymen, however, this position [“better as solving puzzles”] lacks an essential element. A scientific theory is usually felt to be better than its predecessors not only in the sense that it is a better instrument for discovering and solving puzzles but also because it is somehow a better representation of what nature is really like. One often hears that successive theories grow ever closer to, or approximate more and more closely to, the truth. Apparently generalizations like that refer not to the puzzle-solutions and the concrete predictions derived from a theory but rather to its ontology, to the match, that is, between the entities with which the theory populates nature and what is “really there.”

There is, I think, no theory-independent way to reconstruct phrases like ‘really there’; the notion of a match between the ontology of a theory and its “real” counterpart in nature now seems to me illusive in principle. Besides, as a historian, I am impressed with the implausability of the view. I do not doubt, for example, that Newton’s mechanics improves on Aristotle’s and that Einstein’s improves on Newton’s as instruments for puzzle-solving. But I can see in their succession no coherent direction of ontological development. On the contrary, in some important respects, though by no means in all, Einstein’s general theory of relativity is closer to Aristotle’s than either of them is to Newton’s. Though the temptation to describe that position as relativistic is understandable, the description seems to me wrong. Conversely, if the position be relativism, I cannot see that the relativist loses anything needed to account for the nature and development of the sciences.

Kuhn is saying two things here, and neither should be all that controversial. What he’s calling “ontology” = Truth-Capital-T, and I’ll use both accordingly. Chill, continentals, “ontology” is what you make of it or whatever:

First, paradigms are incommensurable in terms of “description of reality,” not in terms of number of puzzles.

One of the main issues with paradigms is that data and problems aren’t necessary coherent one to the other. The way one paradigm will interpret a phenomenon – and thus what it finds worthy of investigation – will require the entire paradigm behind it. Theoretically, this makes every paradigm a kind of black-box. In practice, there’s an easy movement between one aspect: one can handle more problems, even if it conceptualizes them differently.

Newtonian physics makes several ontological claims (the universe is corporeal particles), Ptolemaic astronomy the same (circles are fitting for the heavens due to their divinity), etc. Both of these are wrong. Newtonian physics, however, can solve many more puzzles. “Amount of puzzles solved” is commensurable – it carries from one scientific set to another, there’s a quantifiable, comparable idea of progress. The ontologies of the paradigms display no such progress. One might allow you to solve more problems, but whatever is in the background of those problems isn’t building on the former paradigm. It’s totally overhauled it – the first block is blue until you press the second block, at which point they’re both red, in the same way that a new paradigm re-characterizes the old one’s problems. Since there’s a stark division between the puzzle-solving and the ontology-claiming, one can show progress while the other does not. The grid-game tends towards 100, but it doesn’t tend towards red. Or, at least, it requires a very different argument to say that it tends towards red. Click 3 makes more boxes green, but green isn’t “more” than blue. “More what?” Exactly. “Waves are-” It’s a metaphor.

Because of this, the success of a paradigm may not be for the reasons it claims (i.e. accurate description of the world). This is good. Ptolemy believed that circles were required to save the appearances of the heavens, which is wrong, but his results are still accurate within minutes of modern measurement. While this is still “inaccurate” it’s not nearly as inaccurate as “epicyclic geocentric circles.” To be more careful: it’s “wrong” in a very different way. Still, it’s only because of the break between “puzzles solved” and “truth” that we can compare increasing puzzle-solving. If they had a 1:1 comparison, Ptolemy’s measurements should be wrong in the same way as his physical theories are wrong. A certain ontological commitment behind a paradigm doesn’t necessarily tell us much about how good it is at solving problems.

This gets intensified by the idea of paradigm shifts: Science progresses step-wise, not smoothly, in a series of crises and leaps. The only thing we know is that the ontological claims get completely reorganized in a way that little else does. To be able to conceptualize and deal with “more problems,” you need a completely revamped paradigm. Hence, it’s historically unlikely that any current view of reality will remain the same after the next paradigm shift. It’s certainly possible, it just hasn’t happened before, and it would be somewhat weird to expect it to. We can’t say either way. You’re at block 78, everything is red, what color will the next block be? Who knows – it’s unlikely to remain red, but that’s also just an assumption. The point of this is the following: it’s premature to say that science tends “towards the truth” until we’ve solved all the anomalies. Solving them might require [radical departure], or at least history suggest as much.

I’m overstating the case by comparing geocentrism and heliocentrism, but he really is trying to be as strict as possible. It’s provisional, can be falsified but not proven. One can at least imagine a new paradigm where the universe actually does move around the earth. It’s unlikely, but that doesn’t much matter. At stake is progress: neo-geocentrism would still continue the “more puzzles” view of progress, but it’s hard to say whether or not it’s building on the current model towards [future one]. It would certainly feel like a return to an older view of reality. At the very least, it would be different in a way “more astronomical charts” is not different.

This does not allow you to pick and choose whatever you want. Or, well, it does, but you were already free to do so.

Second, there’s no reason to assume that the final description will give us a “Full and Complete” view of reality.

Instead of territory, you get maps; instead of truths, you get predictions. An ambiguous construction of “what’s really going on” may be tacked to that, but it’s neither necessary nor certain. Again, not very controversial.

There are metaphysical commitments that predate the puzzle-solving, meaning that it can’t use its own predictions to test its own predictions. Horseshoe theory in action is discussing philosophy on the internet, so you can try this one yourself: argue with a scientifically minded young empiricist about the nature of mathematics, “How can you be sure that equations are ascertaining something rather than hiding it?” read: “Is science truly objective?” A thousand humanities students just groaned, because they know the exact conversation to which I’m referring. A thousand young STEM students also groaned, for they know the exact conversation to which I’m referring.

In the strictest sense, you can’t determine the “truth” of a paradigm from its success at puzzle-solving, because there’s always a host of pre-paradigmatic intentions. Phrased in a different way, trying to gauge the “true truth” of a theory by number of accurate predictions is affirming the consequent. An obvious and indifferent fact, absolutely shocking in its impact, as prosaic as it gets: Paradigm A -> result B /= result B -> Paradigm A. Do not do this.

This resolves many issues. For instance: multiple ontological commitments can exist within a paradigm, which shouldn’t really be possible if affirming the consequent is acceptable and/or helpful. Interpretations of quantum mechanics is the most famous contemporary example of this, but plenty exist. The equation is not the same as its interpretation, multiple physical models can account for the same predictions. We simply can’t say either way, and trying to requires something other than normal science.

The same is true of certain epistemological commitments. A modern physicist’s minimum paradigmatical commitment is to mathematics as a tool, not anything deeper. She could be a realist or anti-realist re: mathematical objects, provided the first commitment is intact. Interpreting mathematical objects as a Platonist, say, certainly changes how she ought to think about the universe, but it’s not something easily tested, nor should it impact the work she does. (Of course, one should note that math as a tool for physical prediction certainly is a philosophical commitment, and one that Enlightenment thinkers had to defend).

Of course, almost all of this is totally uncontroversial. Bayesians build priors via these processes, but those aren’t meant to be set. Popperians recognize that you can’t positively prove a theory. So on, so forth. Scientism is a by-word for too much faith in Empirical Unerring Truths, but basically no rigorous thinker believes that science heads towards absolute truth. Assuming that requires a serious misunderstanding of the scientific method and/or being drunk and mad at the comparative literature department.


“Is this relativism, though?”

The broad result of this is the following: progress-towards-Truth is wrong, but progress-towards-more-puzzles is correct. My personal feeling is that Kuhn overplays his hand a little bit – scientific achievements incorporate successful metaphysics, i.e. mathematics in physics, but it’s not an easy thing to call “progress” in the way we normally mean that. We also don’t know if there will be a Hegelian end of history (doubtful). I take Kuhn’s point, though: science has very little to say about the deeper sense of truth normally associated with philosophy. Since that’s normally considered to be part of a full true picture of reality, there are problems.

Truth is a binary, and science currently fails. Try plan B: science at least tends towards Absolute Truth. Kuhn’s argument gets rid of that. When Truth-Capital-T is your only value, Kuhn absolutely is a relativist, but you’re going to have a hell of a time arguing for [anything]. You can’t compare the truth values of two paradigms, we don’t even know what that would look like beyond the general assumption that such a thing is impossible. Two things are equally false, then the choice between one or the other must be arbitrary. Any value besides truth is meaningless. In other words: “better puzzle-solving” is arbitrary, it does not matter, pick and choose particles or continuity at a moment’s convenience. Any reasonable rigor vs. empirical science is going to bring you to something closer to Kuhn than not. Rejecting puzzle-solving will then take you to absolute nothingness, i.e. Straw Man Post-Modernism or whatever the kids are calling it.

Note that this isn’t an expression of indifference towards truth, but the opposite. It’s being extremely rigorous about science’s own claims re: truth, taking it at its word, but judging it by a particular value system. Hilariously, the opposite of what we take to be relativism.

Thankfully, truth isn’t the ultimate arbiter.

I’ll gloss relativism as “your call” where “your call”  implies that the judgment is arbitrary for anything besides the individual. By this definition: no, Kuhn is not a relativist. Kuhn gets around charges of relativism in basically the same way:

Taken as a  group or in groups, practitioners of the developed sciences are, I have  argued, fundamentally puzzle-solvers. Though the values that they deploy at times of theory-choice derive from other aspects of their work as well, the demonstrated ability to set up and to solve puzzles presented by nature is, in case of value conflict, the dominant criterion for most members of a scientific group. Like any other value, puzzle-solving ability proves equivocal in application. Two men who share it may nevertheless differ in the judgments they draw from its use. But the behavior of a community which makes it preeminent will be very different from that of one which does not. In the sciences, I believe, the high value accorded to puzzle-solving ability has the following consequences [it is not relativistic].

It’s right to say that science isn’t getting closer to the “truth,” and it’s also probably right to say that one paradigm’s ontology can’t be compared to another’s in terms of Absolute Truth. At the very least, we don’t have tools to do so in a rigorous way. Still, Absolute Truth is not the only variable – in the sciences, puzzle-solving is the primary value, and we absolutely can compare the puzzle-solving powers of one paradigm with another. This is no small thing. Included under the “puzzle-solving” rubric are vaccines and space travel.

You’ll recognize this as “the exact same thing we all say,” and I agree. Given that fact, you might be wondering why I’m bothering to lay out Kuhn’s argument in (I hope) painful detail. It’s an attempt to ward off a confusion. For some reason, the people who ask me “Why do you care about Platonic Truth at all?” are the very same people who ask me, “Why are you so concerned with values?” with the general understanding that such things have been settled by decent society.

The answer is the same in all cases, “We’re already using something else,” which is why the questions should concern you. If truth is so obvious and also so unimportant and also comes from science and also justifies science and also has so little to do with science, then why wince when I say “We use lies to justify ourselves, science is inadequate, truth it too, this is good”? Provided you don’t take any of that seriously anyway, it’s just stating the obvious.

Or, if Kuhn is simply telling us what we already think, why does he have the reputation he has? Why can I search his name and find Errol Morris’s feverish ravings, ten billion hard-nosed “realists” shrieking about the post-modern threat?

What do you want to hear, besides what you already know?

top from Satantango by Béla Tarr

Author: Lou Keep

40 thoughts on “Science Cannot Count to Red. That’s Probably Fine.”

  1. I’m not a philosopher, I don’t know if this has been answered or addressed in the canon. But. Would this be a workable definition of Science?

    Science is what enables engineering efficiency.

    It, at least, addresses the notion of working towards something, there is a thermodynamic limit to peak efficiency of basically anything. It’s (with a little hand-waving) applicable to all the branches of science I can think of: for example, research in biology improves efficiency of diagnosis and cure; research in psychology improves (in theory) efficiency of dealing with mental state, which could on a meta level improve processes that drive (other) scientific research…

    At the very least, can someone recommend further reading on this topic?


    1. Hmmm. I think you’d need at least some extra bit devoted to conscious behavior. It certainly seems that increasingly efficient engineering happens from non-scientific behavior, e.g. agricultural planning from religious festivals, dams and waterways from water temples in Bali, etc. I’m sure there are better examples, these are just ones that I’ve used before (read: overuse).

      Kuhn’s really skeptical of any coherent definition of science. He kind of avoids the question by just looking at all the things that have historically been called “science.” Better, for him, to think in the concrete (physics, biology, etc.), all of which have paradigms. The other big names are Popper (falsification), Lakatos (research programs and novel facts), and Feyerabend (there is no science at all).

      Philosophy of science isn’t my field, and those are all, like, stars from some time ago. I’m sure that no one has defined science in a way that’s pleased everyone, but unsure where the current debates are.

      Liked by 1 person

      1. Good things to think about, thanks.

        On the religious festival front: I want to object that the failure to understand the underlying causes of, say, increased harvest when praying to X god with Y ritual inherently limits the amount of total production. To at least a local minimum, say. But I read your/SSC’s writing on Seeing Like a State, which is the obvious rebuttal to that…

        I’ve seen Popper’s name often enough, I’ll look into the others.


  2. I don’t see why there’s necessarily a connection between the incommensurability of paradigms and progress…

    Instead of a series of better-maps-of-reality, you have a series of better predictions

    What’s the difference?

    It seems like this Truth thing is creating all sorts of conceptual problems. Why not just ignore it? Giving it up doesn’t necessitate relativism, and the whole “we use lies to justify ourselves” doesn’t need to follow…As long as we keep making better predictions, who cares if we are progressing toward some telos? I feel that evolutionary epistemology solves a lot of these issues in a nice, pragmatic way by just ignoring all the weird conceptual crap. As long as you have what Land calls a “social mechanism for the elimination of failure, based on extra-rational criteria, inaccessible to cultural capture”, everything’s swell. And if you don’t, the Outside will impose itself sooner or later anyway. You can only be a Lysenkoist for so long before you start starving.

    As Constantin said, in the long run only existence proofs matter: we can apply this in epistemology as well.


    1. Whoops, nice catch on the quote. I must have merged two sentences and mangled both – it should have been paired territory/map and predictions/truth. Will edit.

      Kuhn’s argument is that incommensurability severs the connection between paradigms and progress towards truth, but allows for progress towards more puzzles-solved. That’s definitely progress, just of a different kind – I’ll add a sentence to make that more clear.

      For your broader point on Truth: we’re (mostly) in agreement, and that’s more or less the motivation to write this. I suppose it could be either “lies” or “untruths,” and perhaps I should be slightly more careful there. Sort of a balancing act between the historical form of the argument (which is lies) and how we tend to think of that now (something more like “less-than-truths”).

      At stake is how to deal with relativism and/or society. If “Truth” as traditionally conceived is inaccessible (at least scientifically – Kuhn leaves other possibilities), then the way to escape relativism isn’t hunting-for-truths, but focusing on other kinds of values. Hence the turn towards analysis of valuation, rather than analysis of truth. This isn’t original (it’s where Nietzsche starts), and I’ve said it before (in Everything is Going According to Plan, which was about Nietzsche), but a really, really common response to that line of thought is something like “Well, science is value-neutral, so we can use that to escape relativism.” Which is, you know, wrong – it’s affirmed by a non-truth value, not truth itself.

      It opens up the door for a different kind of analysis, essentially. It also means that there may be things that are supremely false but helpful, and (if we’re using other values) then the “This is a lie” argument against those does not work. Since that’s the go-to argument against, say, mythology, it’s helpful to have this argument in one place. Notable that in Land’s technical work, he goes deep into Kant (and thus pretty academic questions of truth and the limits of knowledge) to deal with these issues. To get there, you need to deal with the traditional definition, show why it doesn’t work, and then show that whatever you do have negates the “this is true” argument.

      Agreed, re: existence proofs in epistemology. Then again, that’s going to get heavily into phenomenology. I suppose there’s a reason that Constantin has been reading Heidegger.


      1. Is this what Jordan Peterson talks about when he says that “reality is that which selects”? Anyways, let’s back up a bit. Surely, no scientific field makes ontological claims except fundamental physics? “What is capital-T Truth” then is irrelevant question in every other field, and “mythology is untrue” has different meaning. Mythology is a lie in other senses: for example, it does make ontological claims and is therefore dishonest for reasons you’ve outlined above. But it doesn’t have to! If tomorrow Meditations on Moloch and Nick Land’s writings and Robo Basilisk get compiled into one book, the resulting mythology will be understood to be a theory on how to best model complicated arrangements of underlying structures supplied to us by our physics theories.(so like any other scientific field, like psychology or economy and so on).

        You ask “If Truth is so unimportant, why do we have it as value?” One ..facet.. of the answer is that we desire to be coherent (as in no compartilization)(see Robin Hanson’s work and social costs of revealed hypocrisy). Traditional mythology is then undesirable (and, separately, untrue) because it is incoherent to have two distinct ontologies – either you have gods or you have atoms. There – i’ve rejected many claims about reality on grounds other then truth and subordinated Truth to other value – logical consistency. Maybe value of One Truth is what you get when you intersect desire for coherency and desire to incorporate all of the available data into your model.


      2. Is this what Jordan Peterson talks about when he says that “reality is that which selects”? Anyways, let’s back up a bit. Surely, no scientific field makes ontological claims except fundamental physics? “What is capital-T Truth” then is irrelevant question in every other field, and “mythology is untrue” has different meaning. Mythology is a lie in other senses: for example, it does make ontological claims and is therefore dishonest for reasons you’ve outlined above. But it doesn’t have to! If tomorrow Meditations on Moloch and Nick Land’s writings and Robo Basilisk get compiled into one book, the resulting mythology will be understood to be a theory on how to best model complicated arrangements of underlying structures supplied to us by our physics theories.(so like any other scientific field, like psychology or economy and so on).

        You ask “If Truth is so unimportant, why do we have it as value?” One facet of the answer is that we desire to be coherent (as in no compartmentalization)(see Robin Hanson’s work and social costs of revealed hypocrisy). Traditional mythology is then undesirable (and, separately, untrue) because it is incoherent to have two distinct ontologies – either you have gods or you have atoms. There – i’ve rejected many claims about reality on grounds other then truth and subordinated Truth to other value – logical consistency. Maybe value of One Truth is what you get when you intersect desire for coherency and desire to incorporate all of the available data into your model.


  3. The first part of this seems like a lot of effort to say something very simple:
    There are infinite noncontradictory realities which have the exact same behavior at every level you could possibly interact with, but nonetheless are following different rules. It is utterly impossible to know which one you are in.

    I have a long winded rant on why this isn’t very important, but it isn’t very original, and I am on a phone.


    1. That’s a related argument, but it doesn’t really address Kuhn’s view of progress. Or, well, Kuhn is more interested in looking at historical changes than hypothetical interpretations, anyway. Kuhn is often accused of relativism, so I wanted to show that his own argument runs counter to that. It wasn’t my own view, and I was pretty clear that I disagree with more extreme interpretations.

      I dpn’t think it’s particularly important as a fact in itself. Depends on what you’re trying to do with it, or whatever argument it’s used against.


    2. And occasionally you gain the ability to interact with new levels, thereby falsifying some models that previously seemed sound.


    1. I agree that science doesn’t progress towards Truth, or at least not towards the kind of absolute truth that philosophy worries about. I don’t think you really need Kuhn for that, though.

      I generally think that paradigms are a good way to understand science, that the “worlds” that different paradigms inhabit are incommensurable, and that science (like most things) is affirmed not by its “truth” but by some other value system. That may be puzzle-solving or something else.

      Things I’m iffy on: how seriously to take the claim that the ontologies (his word) are always different. It’s a somewhat unfortunate phrase for him to use, because it’s super broad. Certain core components paradigms definitely carry through in pretty profound ways, e.g. it’s hard to imagine a post-enlightenment physics that rejects the application of mathematics to physical reality. It’s certainly possible to imagine a future philosophy that does, but he’s pretty strict about the science/phil distinction. I guess: it’s unclear to me quite what counts as “puzzle-solving” for him, and I can see ways that I’d agree and ways I’d disagree. It is probably true that fewer aspects of a paradigm carry across than the popular conception, though.

      I think more broadly, I’m interested in how Kuhn would conceptualize mathematics and physics. Math concerns itself with absolute truths, physics (by Kuhn) cannot. He might need a bit more Kant in his system, I dunno.

      Hopefully that’s a somewhat satisfying answer for all its uncertainty.

      Liked by 1 person

      1. I have not read Kuhn; I’ve only gleaned things about him. For example fro your kind of explanatory post.
        I think I may have read some excerpts by him or scanned through parts of his book that had to do with paradigms in science and things I can’t remember was a long time ago.

        I have a similar idea about things ; I put it in terms I have “communication through a category”. I talk about how there is no communication across a common category. And I speak of a difference between truth and reality.

        It sounds like he is saying something similar.

        Liked by 1 person

      2. This post was about 500 words long when wordpress ate it. But I shall not let uncooperative technology win, and I will write this whole post again just to spite it.

        I disagree with your first three points.

        Science does progress towards capital T Truth, paradigms are commensurable, and science is affirmed by Truth, at least partly.

        Science doesn’t progress toward Truth in the same way as philosophy or maths. Picture the set of possible candidates being considered for Truth. The role of science is shift probabilities in Truth Possibilities Space (TPS), and occasionally add new areas to it. Science can’t ever properly reduce the mass of probabilities over any given region to zero, but it can effectively make the probability so low that no one considers them worth taking seriously any more.

        It also can’t raise the probability mass of any region of TPS to one, or shift probabilities between empirically indistinguishable options. But if science were to ever place 99.9% of the probability mass above a tiny region of empirically indistinguishable possibilities, that is a massive leg up compared to where we are today. I think we agree that scenario is within the theoretical bounds of science, the “every puzzle solved”, scenario.

        I think we also agree that it would be disingenuous to say that we weren’t any closer to the truth in that scenario. Every puzzle solved is a constraint on what the Truth could possibly be, and constraints shift the distribution of probability in TPS.

        Paradigms are commensurable. The purpose of a scientific paradigm is to allow for testable theories which shift around probability mass. Paradigms are better or worse depending on how well they allow this, AND how easy it is to tell when you paradigm predicted something incorrectly, and needs to be replaced.

        For example, physics has an excellent paradigm. It had a large number of tests and values that are well connected by strong rules. Small changes in any one of those values can mean huge things. Slight variations in the light output of a star mean a planet, and the light variations can tell you surprisingly detailed information about the planet, like how big it is, how close to the star it is, and sometimes even what its atmosphere is made of.

        For the second leg, a single repeatable way to observe anything at all going faster than c would get the whole paradigm thrown out.

        Economics has a terrible paradigm, because even after decades of trying they still can’t agree on what minimum wage does. In their defense, physics cheated by being simpler, better funded, and much less politically charged. But reality doesn’t grade on a curve, which means that physics took us to the moon while economics has to hope it doesn’t crash the thing it should be studying again.

        To compare paradigms within a field, Einstein beats Newton not just because he predicted starlight bending during an eclipse, but also because Newton wouldn’t know what to do with that information even if he had it.

        Finally, science is at least partially affirmed by Truth. Just ask them, “Why do you want to become a scientist?” Most of them will answer something that translates to curiosity, an interest in truth. Most of them will be lying of course, because “money/status/I didn’t but my parents made me” doesn’t play well in interviews, but at least some of them are telling the truth, and that’s enough for “partly”.


        1. I appreciate you taking the time. Sorry that wordpress has a terrible comment system.

          We agree on a fair amount, but I think we’re talking past each other here. Part of the issue is, likely, over the way that incommensurability should be interpreted.

          When I say “Capital-T-Truth” (or, more importantly, when Kuhn does) I only mean mathematical and philosophical truth. He’s responding to a branch of scientific realists that were arguing that.

          “Newton wouldn’t know what to do with that information even if he had it” is part of what Kuhn means. It’s not merely a theory that you can plug data into, but something that determines what kind of information is even meaningful, the kinds of questions you ask about it, what an appropriate answer would look like, and what it ought to look like. Of those final two, Kuhn is saying: our current model tells us that “appropriate answer” should look [thus], but since we have no good idea what capital-t-truth is to test that against, it becomes a mere bias of the paradigm. This does not mean that the choice from one to another is arbitrary: it means that we’re trapped, in a certain sense, and all the physical models offered should be taken with a grain of salt. (I think we may also disagree on the use of rules in science – paradigms are much more akin to patterns followed – but that’s a bit of a different discussion.)

          In its place, he offers “amount of puzzles solved,” which is similar but not quite the same. I think the kind of truth you’re talking about works here and falls under the more puzzle-solving rubric. Increasing probability and accuracy and prediction and interpretation is certainly what Kuhn means by that. That’s distinct, however, from the physical interpretation of it – there are many physical models that fit our predictions, all equally fit the predictions. Some may well be outside of it. In the comments to my last piece on Kuhn, someone brought up a quite helpful example: imagine a conscious automaton in the (computer program) Game of Life. Based on prediction and observation, a proper scientific truth for it (the Full Theory) would only tell it the program’s rules of updating. That may or may not be enough (depends on your desires, I suppose), but it certainly isn’t “reality” in the way we normally or philosophically use the term.

          Allowing for 99.9%, and assuming it only admits of one physical model (which I do actually doubt), you’re going to have to accept certain metaphysical positions to get there, and those will fall outside of the standard scientific framework. Modern physics requires that mathematical models work in a way that no one really understands, but for which a variety of theories are all fine (Tegmark vs. Kant, etc.).

          I’ll admit that science is at least partially affirmed by truth, or at least the hunt for it. I’m more interested in how to affirm other things with philosophical truth out of the game. I should also say that I agree with (what I interpret as) part of your skepticism – as I said to landzek and in the piece, it does seem to me that certain paradigms have achievements that carry through. Math is a big one, but certain tools, practices, and assumptions also seem to be there.

          I’m writing about Plato at the moment, which should (hopefully) clear up some of the issues re: truth. I’ll think about how to better address some of this there.

          Thanks for taking the time to engage, even if we’re at odds with parts of this.


  4. To summarize (and hopefully not butcher). Science is great, lets you solve all the puzzles. And it hypothesizes capital-T Truths, through “paradigms.” It swaps them out every now and then, like changing prescriptions, to find the perspective which best reveals the solution to the latest puzzle–but critically, there’s no “progress” in ontology, just “progress” in puzzle-solving, by experimenting with ontology. So if you’re Newton, you change the paradigm, lots of puzzles get solved, those solutions will survive; but your paradigm change won’t necessarily be preserved by the next people to take up the reins. Science progresses, able to use nominalism to sidestep the metaphysical speculations of the past. Chill.

    But problems in ethics don’t seem so conveniently metaphysics-invariant. If you’re a utilitarian, you think we should maximize pleasure and minimize pain. So you’d better know what pain is, and it had better be real. Really-out-there realism real. Ontologically real. And whatever it’s defined by/supervenes on had better be real, too. (i.e. if “pleasure” is “desire satisfaction”, “desires” and their “satisfaction” had better be ontologically real too.) And in the domain of ethics, you can’t nominalism problems away; the answer to the question “is a fetus a person?” cannot just be “it depends what you call a person”, for obvious reasons. Ethical points of view depend on the Truth of very specific ontological commitments. Brush aside the metaphysics, and deny the consequent of the ethics. You can’t do utilitarianism if pleasure doesn’t exist.

    To oversimplify, Kuhn makes a good case that science doesn’t need metaphysics, only scientists do; and this undermines indispensability arguments. But it seems like, in contrast, maybe ethics really actually needs metaphysics. What to do about this?

    In opposition, you could say that just like we have scientific paradigms, we have moral ones. And a future paradigm might do better at ethics by shifting its ontological outlook, but this can be a constructive/progressive change, just like a scientific revolution. Is that the grand analogy being set up here (cf. Samuel)?

    If so, that doesn’t strike me as obviously false. But you can see why adopting that kind of position would be called “relativist.” Abandoning capital-T Truth in the ethical domain is a big ask. Removing Truth from science leaves you with “puzzle-solving,” which is all scientists care about anyway, thankfully. But what does removing Truth from ethics leave you with? Constructivism? “Values”? I think the answer is probably “nothing.” Nonetheless, I’m interested to see your answer, if that’s where you’re going with all of this.

    (Relatedly, I recently left a comment on “Platonism without Plato.” Would you mind taking a look?)

    Liked by 1 person

    1. I think there is a Truth. But we can never reach it, and we are always reaching it. So there is an incommensurablity there. I mean. What would it mea. If we found the Truth of it all? Would the whole thing dissolve and we get a massive “universe” payout? And yet. There are Truth we find, for example, escape velocity and things like that. Or water boils when you heat it.


    2. I agree pretty strongly. For whatever it’s worth, Kuhn himself writes something like, “This stance if applied to morality or politics would unquestionably be relativistic, but in the sciences it is not.” He mostly things that science has a global enough constellation of commitments in its values – and that puzzle-solving is reliable enough of an indicator – that he can escape the charge. Everything else is…

      I am heading towards that, but it’s going to take me a while to get there. This, the issues with math, and [several essays worth of political arguments] all together are the grounding for virtue ethics. Or, at least, that’s what they’ve pointed me to.

      As a side note, the closest system I know to successive-paradigms-that-change-ontology is actually Kant’s metaphysical arguments for deontology. Or, well, that’s what the arguments imply from certain readings. For Hegel, it’s clearly continuous and commensurable (at least, how Kuhn is using those terms), but Kant does seem to think that moral achievements might function step-wise in a way analogous to the categories (he makes that analogy several times, each time more and more literally). I’ve always been pretty skeptical about that. Also: definitely a non-standard reading, and someone with more grounding in the philosophical tradition is probably going to come at me now.


  5. Kuhn argues that, while science doesn’t progress toward capital-t Truth (whatever that is), it does progress in that new paradigms solve more puzzles than old ones. But isn’t what counts as a “puzzle,” and what counts as a “solution,” itself relative to the paradigm you follow? So I’m not sure if he can escape from relativism after all.

    I do basically agree with the case you’re laying out here, though I think you (general “you” here) can make the case more comprehensive by showing that there are many incommensurate systems of logic that you can’t really judge between.


    1. I think Kuhn would say that you’re certainly correct that puzzles are relative to paradigm, but part of normal science is the translation of previous puzzles and phenomena into new paradigms. That is to say, Newtonian physics recodes Ptolemaic data/puzzles-from-the-data into a heliocentric model with gravity, but it does not absorb/translate any part of the Ptolemaic physical model. That totally gets tossed aside. I was trying to motion to that with the “colors change all previously filled blocks” thing, but I probably should have been more explicit. Perhaps I should edit in a more thorough description.

      The result of this is… well, it definitely is relativistic in any strict sense. It’s only from the standpoint of the current paradigm that puzzle-quantities advance, and he can only avoid relativism by defaulting to other values. If you want any certainty, you need to postulate a complete-puzzle-solving-system. That’s a big driver of his emphasis on “progress from” rather than “progress towards, but… well, your mileage might vary with that one. ”

      I’m not really sure how satisfying that is, and I’d agree it’s one of the weaker parts of his argument.

      I’d agree re: systems of logic, and I think that’s a good suggestion. I wanted to be relatively strict here by using Kuhn’s own case, so I (mostly) avoided strengthening any parts of it I found unsatisfying. Eh. Unsure if that was a good call or not.


      1. We could imagine a new paradigm that rejects all or most of our current puzzles/solutions as not being “real” puzzles/solutions. Empirically scientists adopt the previous paradigm’s puzzles and solutions into the new paradigm (mostly), but theoretically there’s no need to do that. And, of course, prizing puzzle solving is itself a value; someone who doesn’t care about that wouldn’t be at all convinced by Kuhn’s argument. But maybe this is the point where the distinction between objectivity and subjectivity starts to collapse.

        I think you did a good job summarizing Kuhn’s argument and not significantly strengthening it was a good call imo (that can be for future essays perhaps); I just personally came to much the same conclusions by learning about paraconsistent logic and the like.


  6. I am perhaps overindexing on the toy example with the colours, but isn’t the equivalent of “Does Science progress towards truth” in the square-colouring model “Do the squares progress towards a colour”, rather than “do the squares progress towards red” specifically?

    To claim the squares progress towards red you have to know what that final Truth is; to say they progress towards a colour you only need to see that so far the box has been increasingly filled up with a single colour.

    For me, a better metaphor for science in the same category looks like this:

    Imagine a 10×10 grid of squares, some filled with different colours. The squares with colours are things that you know answers to; each particular colour is a theory/paradigm/model, that may be answering multiple questions. Empty squares are things you haven’t thought of / don’t have answers to.

    You start clicking. On each click, some squares shift colours; maybe a new square gets filled with a new colour; maybe two colours merge together; maybe even some coloured squares become white again, because you thought you knew something but it is proven false.

    And yet, as you keep clicking, you start noticing a pattern. It seems that as you click, more and more of the squares are filled, and the number of distinct colours keeps decreasing. You started with 64 different colours in a 10×10 grid, and now you have four.

    Does this game progress towards the grid being filled with a single colour? You can’t be certain, but it does seem like it.

    And maybe when you get to 99 squares filled with one colour your mouse breaks / the game hangs, and you can’t actually ever get to the 100%. But that doesn’t mean you weren’t getting closer to it when you were clicking.


    I get a similar feel towards the rest of the article as I do to the toy example, which is why I wrote it up. It feels like I’m missing a concept. In particular even in the opening statement:

    To say that science is advancing “towards” truth presupposes that we know there will be a stopping point, that it’s on the right track to get there, that there’s an interpretation of the end point we’ll agree on.

    I don’t think you need this terminating condition to say you advance towards something. The series 1/2, 3/4, 7/8, 15/16, etc progresses towards 1, even though it never gets there nor is there an end point you can agree on.

    So maybe we just have a definition conflict, and I need to read up on Kuhn’s terminology?

    (apologies if this is double-posted; wordpress asked me to log in then seemingly discarded the submission)


  7. Kuhn’s points may be relevant to non-scientific social revolutions as well. For example, we could think of candidates like Hillary and Jeb? as engaged in normal politics using the accepted paradigm, with Trump, Sanders, identitarians, alt-right, and various marginal political viewpoints (monarchists? really?) trying to cause a paradigm shift in their favor.

    And then Trump won. Hillary’s base was really, really upset by the prospect that politics is progressing away from the Clintonian model. The new paradigm may be quickly overturned by something else, or it may grow deep roots. It’s impossible to say. Which seems a lot like saying that politics cannot count to red.


  8. Since I know this is leading into a discussion of values, it looks like you are searching (if I can be bluntly reductive) for a schema of useful things outside the realm of truth. So maybe “the meaningful unverifiable.” Which is maybe not quite the same thing as “the meaningful falsifiable” which is the thing that dudes like Peterson dude that pisses me off. Its one thing to come up with a “just-so” story of how we ought to live that fits all the data and can’t be disproved, but that’s not quite the same thing as a meaningful paradigm shift of the sort that a dude like Nietzsche seems to want. Also, any appeals to “ancient wisdom” seem perilously close the sort of “retreat to a past paradigm” that we want to avoid, although you make a good case that sometimes the past paradigm is closer by analogy to our present understanding of things than what we have traversed through in the meantime, so I guess there’s room to argue there. Anyway, my point is that the values we seek have to be useful, at least in the sense of “solving more puzzles,” and by analogy to Kuhn I am finally starting to get some sense of what that must look like. Something like faith, but not blind faith. A friend of mine once remarked that “when I say I have faith in someone, that doesn’t mean I believe in them despite any and all evidence to the contrary, but rather that a reflection on their past performance leads me to believe they will be showing up with the solution any second now.” That fine line between “difficult to justify” and “reliable” so…metis? I realize I’m getting way ahead of you, I just wanted to let you know how I’m following you here.


  9. I think the assumption behind the ‘relativist’ position may be more like: Yes, science works at giving us vaccines and space travel, but should we WANT vaccines and space travel? , Is it worth the effort? Maybe we could just learn to meditate better and we wouldn’t care that we didn’t have vaccines and space travel .. so that the decision to invest time and attention into the paradigms that enable vaccines and space travel is seen as questionable, arbitrary..


  10. This was a great, high quality post and lays out Kuhn as I understand him from reading Structures. I think the massive freight-load pulled by the term “puzzle-solving” often goes unappreciated. All of our values, the teloi (meaningfullness generators) of our individual biologies and group cultures and influential intellectuals, all of our living and concept making apparatus, is some type of attempt to solve certain HUMAN CONCERNS not metaphysical realities. If that is correct, then a method for doing metaphysics seems difficult, although not impossible to construct (cf. William James’ lectures on Pragmatism, or some of Richard Rorty’s work for more agnostic accounts of metaphysics).

    Perhaps the Kuhnian starting point would be answering what is and what is not a puzzle-solving mechanism? What is the ontological status of puzzle-solving, puzzle-solvers, and puzzles?


  11. I think the strongest counterpoint is something like Asimov’s Relativity of Wrong. “When people thought the earth was flat, they were wrong. When people thought the earth was spherical, they were wrong. But if you think that thinking the earth is spherical is just as wrong as thinking the earth is flat, then your view is wronger than both of them put together.”

    If you have any sort of notion of how different some ontologies are – if they’re not realio trulio incommensurable – then one can consistently talk about them “converging.” This is pretty obvious in the case of the shape of the earth, but is somewhat trickier in the case of the progression from geocentrism to heliocentrsim to acentrism, or the progression from continuous matter to particles to fields. Many scientists will claim that human judgment provides such a notion of difference, but we might be suspicious that this judgment is too influenced by hindsight – and similar with seemingly ad-hoc metrics we might deploy.

    I do think there is in some sense a human-natural way to think about differences between ontologies, though. Atoms versus fields as ontological building blocks is a dispute that only matters at small scales – and so the smaller the scale it matters at, the smaller the difference. The location of the center of the cosmos only matters at large scales, so the larger the scale, the smaller the difference. This is completely human-centric – the difference in scale is the difference from the everyday human scale. But we’re humans, and human-centric knowledge is all we’ve got, so it makes a sort of sense.

    There might be something even better – just like how Turing machines gave us a mathematical way to enumerate puzzle-solving methods (at the cost of restricting ourselves to “computable” methods), we might imagine that there’s some way to enumerate a large enumerable class of ontologies that also tells you how different they are (probably up to a constant – these sorts of things are always only up to a constant). But this is just a pipe dream.


  12. Note that paradigm change is not throwing away all models.

    For example in move from geocentric to heliocentric view planets are still spheres, separate from Earth and following some simple laws of motion, not in influenced by people.

    Compare it to idea that Sun and Moon are good and evil good fighting for their kingdoms and they are influenced by rituals that must be properly hosted.


  13. “Phrased in a different way, trying to gauge the “true truth” of a theory by number of accurate predictions is affirming the consequent. An obvious and indifferent fact, absolutely shocking in its impact, as prosaic as it gets: Paradigm A -> result B /= result B -> Paradigm A. Do not do this.”
    — I don’t see how this makes any logical sense. Trying to gauge the “truth” of a theory by number of accurate predictions would mean that you had a fixed set of antecedents, and tested their predictions under competing theories, then gave each theory a score proportional to the fraction it got correct. “Paradigm A -> result B /= result B -> Paradigm A” doesn’t represent this situation at all. It’s a fallacy because it doesn’t work with numbers. But choosing the paradigm that maximizes the posterior probability of your validation set isn’t a fallacy.


      1. Probably “there’s always a host of pre-paradigmatic intentions” is the important part. But I think this notion comes from philosophical foundationalism–the idea that you have to make some starting assumptions before you can make any inferences. This is true only in classical logic, in which infinitesimals don’t exist. Everyone who uses differential equations knows foundationalism is false in general. In most systems of equations of interest, most or all starting points or distributions converge on the same or similar solutions.


  14. Waited almost 5 years to read this since I hadn’t finished reading Kuhn yet. Still not sure what I think on this subject, since it’s far from anything I could even try to call myself an expert in, but I feel confident at least that this post has helped me understand Kuhn’s work better.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: