Is Science Broken?

By this I mean just what you think I mean.

Is science dysfunctional (i.e. functioning against its stated purpose) and could it be fixed?  I will leave it to you to determine what science’s stated purpose is, though by any standardly accepted definition, I claim that science is broken.  I’d like to run an experiment here to try to either change my belief or solidify it.

In the comments below, I invite you use the Like buttons to vote on what you believe.  You have only three boxes to choose from: Broken, Not Broken, and Undecided.  I respectfully ask you to first use the appropriate Like button and only then add your arguments/comments/questions if you have them.  Also, please categorize your arguments/comments/questions by making them replies to of one of the three top-level boxes (if you “think outside the boxes” I will delete your comment; sorry it’s my experiment :-)

In order to begin the debate, I will refer you to two blog entries which make my argument for why science is broken:

Let the games begin!

  • Rafe Furst


    • Romina

      Broken -though, it was never whole to begin with. Science, is ever evolving and assuming/expecting/hoping in it’s (in)fallibility is…

    • Natalia Vega-Berry


    • Matt

      I don’t know whether broken is quite the right term. The tools of science give us a lot to work with, but also seem to make scientists blinders to other realms of reality that might need new tools.

      • Matt

        give scientists blinders… sorry.

  • Rafe Furst


    • hbonwit

      We don’t have a Grand Unified Theory (yet) and so it might not all make complete sense to us yet. But what we do have is solid ground to stand on, moving forward.

      What I think you might be referring to (though I’m honestly not sure) is that we’re hitting some pretty rough patches with respect to 1) the level of science research is beyond what a person can verify in their garage, leading to #2) we’re not communicating our real science findings to the average consumer. Things are getting very messy and political, with respect to ‘science.’

      • Rafe Furst

        i’m actually positing a much deeper hypothesis: we will never have a “Grand Unified Theory”. reality is infinitely, fractally complex, and unfolding every moment to be more so. as soon as a theory is hypothesized it becomes instantly out of date; change is constant. as your sadhguru suggests, “embrace ‘i don’t know'” :-)

        • TheQuickBrownFox

          Indeed, a grand unified theory may be fundamentally impossible. It would depend on reality actually reducing down to something that can be expressed as less than the whole of reality itself (i.e. a set of laws). We don’t know for sure that this is the case.

          • Dan

            Godel’s Incompleteness Theorems.

    • Anonymous

      To me, the goal of science is to generate models of reality that enable us to create more useful artifacts and predict events more accurately. By that measure, science is clearly working.

      I certainly don’t think science is optimal at the moment. I acknowledge the possibility that the rate of new model generation and incremental improvement to existing models may have decreased recently. But it’s hard to specify a good metric and it could be a natural consequence of diminishing marginal returns.

      • Rafe Furst

        what if the the unfolding/increasing complexity of “reality” is such that it
        has a higher ordinality (or cardinality) than the pursuit of science can
        generate? if that were the case then as time goes on, the gap between
        what’s true — i.e. predictable and usefully-createable in your words —
        what can be known scientifically would grow (at least) exponentially. there
        would be a point (you might call this a singularity) at which it becomes
        meaningless to pursue science due to increasing futility. no matter how
        highly you value science, surely there would be other values you hold that
        would not suffer as great a futility factor….

        • Anonymous

          That would certainly be an interesting state of affairs. Because science is recursive, based on tools created in reality, I would naturally assume that its ability to generate models will increase as the perceived/accessible complexity of reality increases.

          Now, this ability may increase more slowly than the complexity of reality itself. However, if this is because reality’s complexity is growing at an exponential rate and science’s is growing at a geometric rate, that’s still a lot of capability that science is generating!

          Even if reality is growing linearly and science is growing logarithmically, the marginal product may well be worth it if amortized across a large enough total capital base.

          So I think the possibility space occupied by the world you posit is smaller in volume than the ones I can imagine where science still pays off. Unless you have evidence to the contrary.

          • Rafe Furst

            My evidence is circumstantial and can was summarized in the two links above.

            Also, what’s to say that reality and science are not already different cardinalities in which case the “gap” is, was, and always will be infinite, just more so in the future and more obviously so all the time.

            • Anonymous

              Uh, given our argument in the New Scientific Enlightenment post, I obviously don’t think that evidence is very strong. The other is too narrow for me to consider informative on this topic.

              On “cardinalities”, I’m not concerned about the gap. I’m concerned about whether science is breaking significant new ground. As I argued above.

              • Rafe Furst

                Ah, but my argument is that “breaking new ground” is isomorphic to “reducing the gap”. Thus you should should be deeply concerned about cardinalities.

              • Anonymous

                This seems to go against your normal argument about models != reality. Cardinality is a model. Who says it’s a good model in this case? In fact, the “reality” of infinite sets is a matter of some debate, even where we have evidence that it seems to work (e.g., physics). Trying to apply it casually to vaguely defined concepts like science and reality seems ill-advised.

                If you can define a complexity model and show that that the cardinality of science’s complexity and reality’s complexity are different, I’ll be darned impressed. Until then, I’m taking your argument about cardinality as simply throwing up a semantic stopsign. It tells us nothing about science being broken or not.

              • Rafe Furst

                Correct, it tells us nothing about science being broken or not. What it
                does is semantically reduce the question of whether science is broken to
                being one about relative cardinalities (or ordinalities).

                Put another way, intuitively it seems obvious to me that the universe itself
                is increasing in complexity really fast; Technology is on a double
                exponential, and this is just in our small corner of the universe. The
                question is, can the models being generated (to explain/predict the
                universe) keep pace with the universe.

                If it can’t, then the gap widens. How quickly is just a wild-ass-guess.
                But it sure does seem that the socio-technical systems we know about are
                peeling away — in complexity measure — from the models being created of
                them (whether those models are human-generated or technology generated).

                Now, I grant you that maybe there’s just a lag until some transhuman AI is
                able to start closing the gap. After all, it’s not a fair computational
                fight at this point. But I feel there’s something more fundamental about
                the situation that precludes this, and that is reflexivity.

                Assume a system, S. A theory of S is created called T. If S contains a
                subsystem, s, which is in fact the creator of T, then the positive feedback
                loop begins and we are in trouble. Because for T to improve to T’ (i.e.
                narrow the gap), T’ must include at least S+T:

                T” > S+T+T’
                T”’ > S+T+T’+T”

                The claim that science is broken is simply an extension of Godel’s
                Incompleteness Theorem. In particular, it’s a realization that
                incompleteness is accelerating just like — because of — The Singularity.

              • Anonymous

                Intuition doesn’t work very well with transfinite sets. The integers and the rational numbers are of the same cardinality, even though it seems like the rationals should increase much faster.

                But this is already misplaced concreteness. Again, I object to your _model_ of this problem on two grounds. First, you haven’t provided any justification that using transfinite sets is a good model.

                Second, even if transfinite sets apply, and even if reality has a higher cardinality than science, that doesn’t mean science is broken! What we can explain will still increase very rapidly, just not as rapidly as what we can’t explain. But so what?

                If social utility is monotonically increasing in explanatory power, then the social utility of science remains quite positive. Therefore, not broken. QED.

              • Rafe Furst

                My claim is that social utility depends on explanatory power and if the gap widens that power diminishes. So while the utility may be ever increasing, it will be with diminishing returns. And at some point, that utility will be dwarfed by many if not most other social endeavors.

                It doesn’t matter whether you call this “broken” or not, but the net effect is that science as practiced today becomes like astrology, a socially marginalized activity with little explanatory power.

              • Anonymous

                No. Diminishing returns does not follow from your argument. This is precisely what I’m saying.

                Assume that you are correct about cardinalities. For concreteness, say the universe is increasing complexity at a double exponential and science is increasing its explanatory power at an exponential. The gap is widening but the marginal return of science is _increasing_!

                Recall that I don’t think your model of cardinalities is at all applicable here. But even if it is, it tells you _nothing_ about the marginal benefit of science.

                This is a classic glass half full / glass half empty situation. You say the gap increasing is bad. I say the volume of explanation increasing is good. Again, even assuming your cardinalities argument, which I think there’s no evdience for.

              • Rafe Furst

                I see the disconnect now. My definition of explanatory power is essentially how close science and reality are to one another. Thus the size of the gap is the inverse of utility, by definition.

                I have no idea what your explanatory power is, but EP(kev) is way different than EP(rafe).

              • Anonymous

                Let’s not confuse what utility means too. I’m using it in the economic (von Neumann-Morgenstern) sense.

                EP(kev)=number of phenomena explained. Evidently, EP(rafe)=fraction of phenomena explained. I claim EP(kev) is more relevant to standard of living because if you can explain more phenomena, you can build more gizmos, means you can do more stuff with less effort, means a higher standard of living.

                To me, it seems your assertion that science is broken requires two extremely tenuous assumptions. First, that your model of complexity cardinality is applicable. Second, that somehow the fraction of explanatory power is more important than the gross amount of explanatory power. I don’t care to argue either of these with you because my priors on either of us changing our minds are extremely low. However, I’m happy to argue about whether two assumptions are in fact what define our difference of opinion.

              • Anonymous

                Branched back up near the top for readability…

              • Rafe Furst

                Yep, we are in agreement about what the assumptions are.

        • Anonymous

          I’m branching back up here because we now understand each other better on this topic and it’s getting hard to read the other branch with all the indentation.

          I propose the following thought experiment. In 2012, physicists fire up the LHC and make an astounding discovery. This discovery has two immediate implications: (1) we now know how to create an abundant, cheap, clean energy source and (2) we realize that fundamental phsyics is much more complicated than we thought.

          As far as I can tell, these facts perfectly fit Rafe’s assertion that the complexity of reality is increasing fundamentally faster than that of science. Moreover, we see that this discovery has actually decreased the fraction of what we think we know about reality. Nevertheless, I claim it’s a tremendous positive advancement attributable to science.

          If this would meet your definition of “broken” Rafe, I’m quite satisfied with broken then.

      • Paul

        I think it’s more accurate to say that engineering is clearly working. We’re making better gizmos all the time. But the pure sciences are in a lot of trouble.

        • Rafe Furst

          How would you define “better gizmos”?

          • Paul

            What Kevin said: semiconductors, storage, materials and nano. I don’t enough to agree or disagree with bio. There’s a fun PBS series running right now, hosted by David Pogue.

            The pure sciences that are in touch with engineering are healthy. To the degree that a pure science has no engineering counterpart, it doesn’t just stagnate, it meanders with no compass.

            • Paul

              “I don’t enough” => “I don’t know enough”

        • Anonymous

          I don’t think the engineering/science divide is nearly as clear as you do. I talk to a lot of people in semiconductors, storage, materials, bio, and nano who think they are doing science. But it leads pretty directly to “gizmos”.

          I’m not sure what “basic” sciences covers. The bio guys are making a ton of progress as are the materials/surface chemistry guys. That leaves physics, which I agree seems to be stagnating.

          But I would be surprised a priori if you couldn’t find some area of science that was stagnating.

    • Dan

      It is not science that is broken, but rather peoples’ interpretations of it. Let’s start with what science is not, as far as I am concerned:
      - Science is NOT: “Studies show…”.
      - Science is NOT: “Statistics say…”.
      CORRELATION DOES NOT IMPLY CAUSATION! This is one of the most fundamental ideas in Statistics (applied math, also not science), yet it is ignored by people every single day. A statistic does nothing to imply the cause of something and under no circumstances can it ever prove something!

      Science is bound by this same limitation: The scientific method, by definition, can never PROVE anything: It can only show correlations. PROOF is the fruit MATHEMATICS, which is fundamentally different from SCIENCE.

  • Rafe Furst


  • Pingback: Tweets that mention Is Science Broken? - The Emergent Fool --

  • Natalia Vega-Berry

    I can’t agree more with you. I have the same thinking after 3 years of research. Thank you for this debate.

    • Rafe Furst

      Natalia, please re-enter your comment as a reply to one of the three comments mentioned in the post. It’s not clear what you agree with since you didn’t follow the directions :-)

  • Pingback: Scientific Singularity? - The Emergent Fool()

  • We have been taught a slightly idealized version of the Scientific Method, which emphasizes Reductionist (Model Based) Methods. But for decades, starting in Life Sciences and spreading, we have increasingly been forced to use Holistic (Model Free) Methods. Genomics and Drug Discovery are prime examples; the term Model Free Methods was coined by Lionel S. Penrose, a famous pioneer geneticist (and incidentally father of Sir Roger Penrose) in a paper in 1935 (!).

    I discuss this at some length and even provide both a zoo of Model Free Methods an an example of them in action (the NetFlix Challenge) in my talk which is available at by the name “Science Beyond Reductionism”. It starts slow and loose… stay with it :-)

    My other talks there discuss how Artificial Intelligence research could start making major progress the moment we re-classify it NOT as a programming problem but as a Life Science. For more on this aspect, Google for Artificial Intuition and my name.

    BTW, we unabashedly use the word “Holism” to mean context-supported . This is not the “holistic” stuff often associated with crystals and aromatherapy, this is the hardcore epistemological holism that has been debated in the Epistemology and Philosophy of Science community every 20 years or so ever since Smuts (1926) and Schrödinger (1946).

  • Sarahsenergy

    What is science, a simple quest for things misunderstood or not known. New data always being found that was always there to begin with but not seen. It takes many ways of being and thinking to get whatever we are searching for, science can be broken only if you think that it is broken-what is there to be broken about science anyways, I did not realize that science could be broken. Just answers that do not make sense or SIMPLY something that is incomplete.

  • s miller

    Not broken. My first day in graduate school, around the lunch table with my major professor and several of his new students, he asked, “What’s the purpose of science?” We went around the table and each of us gave fairly straight forward answers - exploration, finding truth, learning, etc.  After each answer my professor’s headed tilted more and more down. We were clearly disappointing him. When it was his turn, he said simply, “The purpose of science is to publish.”

    What is broken, in my opinion, is the fact that scientists communicate so poorly.

    The public has little understanding of what scientists do and why it’s important. We communicate okay among ourselves, usually. But outside the classroom there are few scientists who are trusted by the public to speak the truth.  Part of this is related to the relatively recent media obsession to treat controversial subjects as equal, giving both sides the same print space no matter where the consensus or truth lies. And part of this is related to the “for every PhD there is an equal and opposite PhD” theory, which is a product of uncertainty in science and how it can be exploited to support political or personal agendas. 

    I really like your idea about investing in people, but would add a twist that some of the investment needs to be for gains that are about more than profit.

    For example, invest in scientists, with the specific aim to enhance their sex appeal, not their research. We need scientists with rock star appeal. The investment would be for things like sending them to acting school, makeovers (especially for females, but not limited to just females), teaching them how to make films (that is, how to tell a good story), rather than teaching them how to better answer interviewer questions. Get them out of their “heads” in terms of how they deal with the public, so they speak and communicate from their heart, gut, or even lower regions (sex).  Of course, because they are scientists they would still be constrained by the truth.

    Serious investment is needed. As you know, start-ups are risky and they often fail. Same would happen here, but invest in 10 scientists and if you get one big win, then you have succeeded.

    Best regards, A fellow poker player (amateur) and scientist (marine biologist). 

  • Thomasee73

    Not broken.

    The blog “Cancer’s Inconvenient Truths” main claim is that there are many inconvenient truths about the epidemiology of cancer, and too much
    cancer science funding is going to an inappropriate area, genetics.

    However - whether or not too much science funding is going to genetics
    irrelevant to the question of whether scientific practice is broken. It
    be relevant to the question of whether  institutions that make science
    funding decisions are broken, but not whether science is broken. Furthermore, the
    main reason that we know about all of the inconvenient truths as listed
    on the blog is primarily because of (scientific) research. Science isn’t
    broken, the implicit expectation that there are simple answers is

    The main claim of “The New Scientific Enlightenment” is that new
    developments in the philosophy of science are suggesting that earlier ideas in
    the philosophy of science have significant limitations, and that new scientific
    methods need to be developed to cope with emerging problems, and emerging recognition of the limitations of earlier practices.

    However, the new scientific methods that need to be developed and are being developed right now ARE science. Furthermore, the reason that we know that earlier ideas in the philosophy of science have limitations is because of the practice of science. Science as currently practiced has limitations, but it is not broken. Similarly, the more general nature of Einstein’s theory of relativity
    does not imply that Newton’s laws are broken. The way that we know that
    Newton’s laws have limitations is because of the practice of science. The idea that science produces certainty is broken.

  • Enkhbat Volodya

    Yes I agree with you. In my opinion Science 2.0 should provide a cybernetic explanation of digital
    physics in frame of three kinds of control. Perceptually controlled systems –
    living systems, none perceptually controlled systems – physical systems, nor
    perceptual systems – subatomic systems. Of course it should explain also that all
    these three kinds of systems exist in cybernetic interdependence and their
    physical and chemical interactions can be easily described in realm of communication
    and computational processes. The explanatory power of Science 2.0 should reach
    from brane (string and M theory) to brain.

  • Enkhbat Volodya

    I forget to say that scientific method as currently practiced is incomplete. Incomplete doesn’t mean broken. It is like a children case of intelligence. Even they do wrong in our view point it doesn’t mean that their perception is broken, but incomplete.