What happens when we can’t test scientific theories?

Just got back from a wonderful trip to Chile, where the weather was perfect for watching the solar eclipse from the beach at La Serena.

While I was away, the Guardian Science Weekly podcast I participated in before leaving for Chile went online and is available here. Thanks to Ian Sample, Graihagh Jackson, and the others at Science Weekly who put this together, I think they did a great job.

The issues David Berman, Eleanor Knox and I discussed in the podcast will be familiar to readers of this blog. Comparing to the arguments over string theory that took place 10-15 years ago, one thing that strikes me is that we’re no longer hearing any claims of near term tests of the theory. Instead the argument is now often made, by Berman and others, that it may take centuries to understand and test string theory. This brings into focus the crucial question here: how do you evaluate a highly speculative and very technical research program like this one? Given the all too human nature of researchers, those invested in it cannot be relied upon to provide an unbiased evaluation of progress. So, absent experimental results providing some sort of definitive judgment, where will such an evaluation come from?

This entry was posted in Uncategorized. Bookmark the permalink.

12 Responses to What happens when we can’t test scientific theories?

  1. Jon Orloff says:

    Another relevant point, how does the science community expect taxpayers to pay for a program whose outcome may be uncertain for hundreds of years?

  2. Robert says:

    Same as in other fields. Mathematics comes to mind. No real experiments there, formal correctness of proof does not say anything about relevance of result and peers who can judge are also typically personally invested in the subject.

  3. tulpoeid says:


    formal correctness of proof does not say anything about relevance of result? That’s a new one to me. It’d be nice to hear more about what the implied problem is.

    (I’m not a mathematician and I’ve been having the impression that things are pretty smooth in the proof department.)

  4. Low Math, Meekly Interacting says:

    “(H)ow do you evaluate a highly speculative and very technical research program like this one?”

    If you’re talking about science, to me the simplest answer is “You can’t.” It’s also the only correct one, in my opinion.

    It shouldn’t stop people from working on it, necessarily, but if one so chooses, one must therefore forgo classifying the activity as “scientific”. People should be entitled to their opinions. I think they should even be afforded the opportunity to seek funding to further cultivate those opinions in an academic setting. Other disciplines do this. But divorced from experiment indefinitely (conceivably even eternally), speculation is just that: opinion, however rigorous or well-informed.

  5. Robin says:

    It comes down to costs and benefits. The cost of doing any theory is essentially independent of the type of theory that is being worked on – it is mainly the salaries of theoreticians plus overheads, and the cost is incurred as the work is undertaken. Benefits that might not be realised for some time in the future are worth less than comparable benefits that come sooner, for example by undertaking different endeavours. Berman mentions that he sees benefits delayed by perhaps a thousand generations, so these are of very little value indeed. This mismatch of timing looks fatal for the endeavour. Feel pity for the students and postdocs who are sucked in.

  6. Low Math, Meekly Interacting says:

    Suggestion: Include a photo of one of the eclipses you’ve observed in your banner rotation. It’s obviously a passion of yours, and may have a certain aesthetic harmony with the the M87* data for some.

  7. Maik says:

    Well, I think it’s a very delicate question, as it goes right to the heart of what freedom of scientific inquiry should and should not mean. Where does science end, which questions are worthy of scientific pursuit and where is one just wasting taxpayer’s money for playing one’s own little games?

    Even though I am very critical of string theory myself, I hesitate to answer that problem by calling for a restriction to this freedom in the sense of “only work on what can directly be tested”. This may achieve the opposite of what one aims for, namely scientific progress. Nonetheless, I think there are ways to subject scientific work(s) to scrutiny without profoundly restricting that freedom: methodological and institutional ones (at least).

    Regarding the former, philosopher of science Paul Feyerabend was arguing very strongly that it is a bad idea to nail down scientists to a particular methodology (science as a kind of anarchy), but he did not question the need for methodology itself. Rather, which methodologies are appropriate or not is a question that needs to be evaluated and discussed by the scientists in the field and adjacent ones. With respect to fundamental physics, I can say that much of it has essentially become a mathematical subject and as such works therein should adhere to mathematical rigor and conceptual coherence. Even if this is not without problems, as Robert suggests, mathematical culture has evolved like this because it is all too easy to go wrong otherwise.

    On an institutional level, I think it is important to create opportunities for this kind of discussion, to keep the scientists engaged (not just in a circle of specialists), to challenge their ideas, be willing to listen both as a layman and a scholar with a healthy amount of skepticism. The power structure of academia is not always, but in my experience all too often an obstacle to this kind of `open science’. Furthermore, as I view it, Alexander von Humboldt rightly postulated the `unity of research and teaching’, for in the context of this question teaching can also be considered a mechanism of `checks and balances’.

    The problem we should be worried about is not that decade-old research programs turn out to be in vain (it’s difficult and we’re all human!), but that we create a system and culture that prevents actual progress via the illusory, implicit claim that “science is truth”.

  8. Mongo says:

    Sounds like a very common problem many areas of science have that claim the proof is in the future and just 100 years or more away. Is it still science? Does this apply to all areas, hard and soft sciences? Is this a string theory only problem?

  9. Lennie says:

    As opposed to physics, mathematics is not supposed to have any relevance. Its relevance is a bi-product. and nobody knows in advance what is going to be relevant, and relevant to what. Of course, there is always a question of how many mathematicians the taxpayers are supposed to support. Recently some kind of experimental mathematics came to life, though one can say that it is not quite a recent phenomenon: that is what a big part of mathematics was before the 19th century. Maybe one should treat string theory as a new kind of physics, speculative physics, that, if one looks back, is also not quite a new phenomenon. It has already influenced mathematics in a positive way. Maybe this is enough to justify its existence?

  10. Peter Woit says:

    Jon Orloff/Robin,
    The costs of paying theorists doing speculative physics theory are not very large, and in many cases aren’t coming from taxpayers (e.g. the theorists at the IAS are mostly being paid from private money). I don’t see any reason society shouldn’t support at a modest level some small group of people to do speculative work on trying to find a better fundamental theory of physical reality, even if the time-scale for expected success is very long. The problem is how to evaluate this work as it is being done.

    I’ve often argued that part of the answer to the question of evaluation of speculative physical theory is to look to the culture of mathematics, which doesn’t have experiment to keep it honest. Part of the culture of mathematics is to insist on clear and unambiguous statements and arguments, and to my mind speculative physics is very much in need of more emphasis on this. Robert is pointing to the much more subtle problem of evaluating the difference between interesting and uninteresting correct results.

    Much of our current understanding of things like qft is based on the work of mathematicians and physicists who developed new ideas following research programs that had little to nothing to do with experiment. You don’t want to cut off this kind of activity. In any case, if you insist on this, you’ll just have people going on about how their extra-dimensional research has something to do with LIGO, or how they will make predictions, but it will take 100 years or more.

    What you really want is progress towards new understanding, and while this is something hard to evaluate, it is very real. Some work does deepen our understanding, other work doesn’t, and expertise is a necessary but by no means sufficient requirement for evaluating this.

  11. Robert says:


    You can easily write a computer program that derives an infinite series of proven “theorems” from a list of axioms in a formal way (see gödel Escher Bach) with zero creativity or usefulness. On the other hand, many extremely fruitful contributions to mathematics came with incorrect proves at first or totally lacking those (as conjectures, programs, frameworks etc). I am just saying, things are not black and white once you look closer.

  12. DrDave says:

    I found the podcast to be eerily prescient. The universal thread that seems to bind the various mutations in ST is that whatever comes up, ST will adapt. No evidence? Just wait. No predictions? No worries. Until forever.
    We can assume at this point that ST will have an answer to any challenge. There’s no evidence that long term proponents will change their minds, in any case. From the comments in the podcast, we could be looking at support for speculative science to continue for decades or even centuries before anyone feels the need to change course.
    It also seems clear that any breakthrough or discovery in Physics will be absorbed, Borg-like, into the theory.
    In addition, and perhaps most importantly, the structure of academic institutions is simply not equipped to deal with this issue.
    At this point, the only recourse would seem to be new, well financed institutions that provide a complete pathway from graduate student to professor, with a system in place to remain independent. A massive undertaking.

Comments are closed.