Probability And Humanities: A Need For Better Language.

29 Comments


The account of how History is presented in popular culture and even in academic works is poor.This is also true in general culture of news reporting as well.The problem with humanities has always been its presentation of facts without working out on the proper language first. By comparison, science and maths figured out a better language in which facts find their place . This is so because facts are abducted into certain agendas or sometimes certain facts are denied their due relevance. This also makes one question for example, how historians have come to conclusions one way or other. Are some of their results a product of certain personal prejudice or out of some desire for creating a larger national project?. All this is due to lack of numerical values while presenting their work . It is also the case that a lot of our understanding is also subject to discovery of new evidence as well. This again makes people suspicious . A better way to present data and further the understanding in popular culture and even in academia is to present all theories,facts within a Bayesian paradigm and people be taught the alternative theories, the facts as they stood a hundred years ago and how the facts have changed over time and different theories have been losing strength . This takes one away from confrontation between different ideologies to one of analysis.
Presenting things this way helps further everyone’s understanding and sidestepping ideological faultlines in academia and popular culture as well. The other thing that could be added to is the costs of persisting with certain ideas or ideologies. The unique claim that only certain ideas led to violence etc also leads to counter charges and hence forth. Instead one could present the costs of all ideas and ideologies .The costs of taking certain course of action and the potential repercussions. Only when one forces a non partisan language couched in numbers on all do we change the way one sees things. If everyone presented their views in this way, of assigning permutations and combinations of various happenings and then to whittle them down based on data available does it lead to non ideological acceptance of various ideas. The benefit of this method is that it forces all opponents to give a probability score to not only their idea but to ideas of others as well. This is the advantage of science, where scientists give out the various possible explanations to something and then subsequent data is used to figure things out and even change models.
The world is a place where low probability events happen fairly regularly. And the costs of each of these events would be different. And one can do a cost analysis of these events and whether they are worth pursuing. And one more advantage of this idea is that it leads to tests of one kind or other to see whether an idea is working out as their ideologues claim or otherwise. This brings forth people to come forward to have their ideas be put to test In order for people to show that their ideas are indeed working, they need to show evidence on a timely basis. This would remove much need for bickering. Every group needs to show some evidence for both .That their ideas are working and also that the costs of their ideas are limited. A few examples would be to present our understanding of Hellenic knowledge of mechanics had Antikythera mechanism not been discovered. Or how much our knowledge would have been lost about Indian history had Arthashastra been lost.And the number of available independent sources of Arthashastra surviving to present day.Or how divergent would our understanding of Indian mathematics be if Bakhshali manuscript did not survive.All this would provide an empirical probabilistic view of looking at our history and also our present place in our world.And that hope is that it would create a more intrinsically informative understanding of our world. 

0

29 Replies to “Probability And Humanities: A Need For Better Language.”

  1. @phyecon1
    \The account of how History is presented in popular culture and even in academic works is poor.\
    I don’t agree with this assessment at all
    There has been a number excellent TV history programs. TV progs have an advantage over books is that you see esp for modern history sights and sounds of places and events. Many times you can hear those actually involved in history making decisions. Some of the top class history progs I have watched
    Great War in BBC 1966 26 parts . Very iconic TV series About 1st WW Must watch
    (https://www.youtube.com/watch?v=8WhqVgXz1UE&list=PLU5V45-JEC-Cpx2vz_l75e5DQMPY_c0sf)
    World at War – BBC 2nd WW
    Breakup of Yugoslavia 6part
    Cold War 20 part
    Arab-Israeli Conflict – can’t remenber the series name
    Many series about anceint world esp Greence and Rome. For example Mary Beard’s programs on Rome.
    Books have an advantage that they can give many more details and supporting facts and documents.
    I have also visited great Musuems in Athens, Tunisia, Rome, Berlin, etc .

    Something like Berlin Musuem on Nazi Terror – called Topography of Terror- is interesting and instructive

  2. // world is a place where low probability events happen fairly regularly //

    You are hedging when you can make a far more strong statement. Every classical event that actually happens had an infinitesimal probability of happening, because classical density functions are never truly discrete. Not true of quantum ones though.

  3. On the general topic, I think humanities get an unfair rap on the knuckles. The comparison with Science does not help. Many reasons, but mainly because Science is a study of stupid things, unlike Humanities.

    Though there’s an autochthonous Science/Physics envy in the Humanities, and that makes matters worse. Chomsky holding forth brilliantly on the topic (with specific reference to post-modernist movement):

    youtu.be/OzrHwDOlTt8

  4. I am happy this interested you. Yes, I watched that critique of post modernism. Its not just science. Another example is markets,business where one naturally thinks in terms of opportunities lost, costs, probability, taking long position or being short on the market . What is useful is to get everyone to admit that there is a probability for their ideas to come true, probability for their opponents ideas to come true and go about trying to put a number to that. Also admit that there are costs associated for their ideas.To commit themselves to admitting to those costs & Bayesian probability. If we can just do this much, I think most of the ideological bickering might become secondary. And the funny thing is that it already exists in the world of business. And the physics envy is not same as imitation and use of probability. The example I have given is for Bhakshali manuscript, had it not been found, I guess many historians of math would have concluded that lot of math came to India via china and greece and before that there wasnt much happening. The fact that it survived and has come to us and has significantly changed our view of ancient Indian math should get historians to accept that their ability to understand history is contingent on survival of these copies. Which must mean that history must incorporate probability .There must be some humility.

    Please explain this further with an example .” Every classical event that actually happens had an infinitesimal probability of happening, because classical density functions are never truly discrete. Not true of quantum ones though.” . Everything is quantum, decoherence is what happens, hence a lot of things are predictable and fairly common.

  5. @phyecon
    My earlier reply has disappeared. Something like this
    /istory is presented in popular culture and even in academic works is poor./
    I completely disagree with this. There are great TV serials on many historical themes.
    Great War – Iconic BBC 1966 production on 1st ww. Must watch 26 parts
    World at War for 2nd ww
    Breakup of Yugoslavia
    Cold war.
    Arab- Isreali wars can’t remember the name

    Mary Beard’s series on Rome. Poignant tones
    Great museums in Athens, Berlin, Tunisia, Rome.

    1. @vijayvan, I dont know how you can completely disagree by pointing to few good works.Are you saying there has been no bad work on anything at all?. Its a subjective view. I for example would like to know and feel the uncertainty of how things are. Where one isnt sure of what comes next.

      1. @Phyecon1

        you say \ dont know how you can completely disagree by pointing to few good works\

        You have not even pointed 1 history book/program which you have read to come to the conclusion “how History is presented in popular culture and even in academic works is poor.”

        Take something like History of 2nd WW . Can you give a list of books you have read , after which you arrived at your conclusion ? You gripe with history books seems to be \ All this is due to lack of numerical values while presenting their work \
        History can’t be reduced to statistics . In history, you have particular narrative, a particular model of events which is the subject of that history. You may try to make the model as close to reality as possible with the given level of knowledge on that subject. Or you may be writing from an ideological angle – Have you read Trotsky’s History of the Russian Revolution . Within his ideological framework, that is a great history.

        There are dozens of very good history TV programs and public oriented thinsg like museums.

        Of course, there are good histories and bad histories , either in print or any medium.

        Let me take the example of ‘Great War’ in 26 parts made by BBC in 1966. History , always, always gives only partisan view. But given that undeniable fact, the series gives justice to the subject matter. It gives the reasons without any -or much – British jingoism.

        You can critique a particular presentation of history and come to a conclusion good, or bad, or so so , but you cannot conclude reasonably what you did.

        1. I gave an example of Bhakshali manuscript for this reason. Its discovery is an accident without which our understanding of Indian mathematics would be different. History depends on the survival of sources. That in itself is the critique. Probability must therefore become part of it. Another example I give is the nationalist history project of Indian academics. Where the religious nature of violence in Pre Islamic or non Islamic India is also not given due value. Hindutva folks only look to Islamic violence and see that as wrong. I see the nationalist project as an issue.
          Another use is in Historiography, A history of how history is written over time.
          https://en.wikipedia.org/wiki/Historiography

          1. \I see the nationalist project as an issue\
            I don’t understand what your issue is. Is it that the presentaion of history for popular understanding itself is poor or that you have a gripe with nationalist historians . Any nationalist or any self-consciously ideological history slants facts . That applies not only to Indian nationalist historians ( which ones can you be more specific??) , but also to Chinese or Russian or French or US historians . I was watching a French TV program on French colonial experience in Indo-China, and that was going on about how France brought modern civilization like railways to Cambodia. They won’t spend much time on Dien Bien Phu or Algerian atrocities. Vietnamese histories would go about the Heroic struggles against French/US Colonialism with no mention of thousands of assasinations done by the vietnamese communists Communiost histories – ironic since Communists claimed History to be science – had to tow the latest Party Line and keep General Secretary happy, on the pain of Gulag or much worse.
            Within nationalist historians , there are some who write very good histories. In an earlier generation people like Jadunath Sarkar wrote Indian nationalist history. Neelakanta Shastri wrote excelnnet history books of Cholas and ancient South India , but he is accused of giving a brahminical slant to it. People like Romila Thapar also fairly capable , even though her own ideological slant can be easily discerned from what she says or does not say.

            I watched a multi-part Russian TV production of the Nazi invasion of USSR till 1945. It was very good giving a ground-up point of view, without going too much into “top-level” decisions made by Stalin and Politburo. Becuase that will raise hornet’s nest. But within their remit, it was good history

  6. I don’t think that actually works in Humanities. To be Bayesian, one needs priors. Where do they come from? The overwhelming majority of the most acrimonious debates in Humanities are around disagreement on the priors (or biases).

    // hence a lot of things are predictable //

    Are they? That is just a function of how willing you are to approximate. Saying that a train will arrive at Zurich bahnhof at 10:30am sharp is replete with assumptions about what it means to arrive (to stop at the same point, but how same?) or sharp (how exact must you keep time?)

    Each of the specific values that was actually realised, say, 0.00000145cm from the stop mark or 0.0000225s before 10:30am on an atomic clock, have infinitesimal probability of occurrence. But because our language implicitly assumes error tolerance (Swiss ones less than others, but they’re still human!), we can get on with our lives. And talk about probability.

    1. Lets assume that some say the reason x is done is because of religion, other says, its because of economics etc. Except, if its about politics and economics, one could see how the situation evolves where one could either move towards a secular polity or one continues to pick choices while maintaining a religious polity. One seeks to consider all possibilities, keep options open,while the other avoids certain choices. One seeks to expand one’s mind and allow diverse opinion to flourish, other seeks to close that down. If one keeps insisting its politics nonetheless, then by putting up data , one could show that the political system only occupies certain region of space of possibilities among a large space of possibilities. And, I admit I am not smart enough to answer all acrimonious debates. I just think that our aim is to get them to talk in probabilities, costs for their own ideas and ideas of others. Over a period of time, this becomes the new common sense of how problems are stated and addressed. This could cut out some of the errors. To me progress is about cutting down the errors and stating things clearly and in diverse ways as well. Over time, this can do more than most can comprehend.
      A probability of die falling on any number is 1/6, a coin toss leads to two possibilities mostly or 3 possibilities if it falls in between or 4 possibilities if someone catches it . quantum mechanics leads to classical physics in our world.

  7. The thoughts are interesting, but there is an ideology inhering in your assumption that “data” or what you call “numbers” is free of ideology. Because science has for ages been used as an ideological state apparatus, we have a scientific discourse, both in the East and the West, that’s pretty much predicated on knowledge muscled into mainstream discourse as “truth,” while other forms of knowledge have been discredited or simply evanesced out of history. Check out Dipesh Chakravarty’s “Provincializing Europe” for a more theoretical perspective on this. Your blog carries a certain apriori bias against the humanities on grounds that it’s largely a vehicle of ideology. This is controversial and can be argued against. And the assertion that numbers don’t lie is highly arguable.

      1. Took me a while to understand this, No. Absolutely nothing against jews. Only trying to bring math into what is description of things in words. If any , more jews the better.

          1. Google gave me hit of anti semitic works. In anycase, by poor one only meant that it is not placed with same rigor as in math and can do better, not that it is not of value .I see this as analogous to math before discovery of algebra.

            have no idea of what that means.

    1. Thank you for your interest. My ideology and bias is that everybody either lies or presents data to fit an ideology. In science and markets, one uses numbers instead to cut down various options and force things to obey certain constraints. Yes, some lies will persist. But it is about cutting the errors and having a better standard than what one already has that interests me. And it has other benefits. If this becomes the new common sense. Everyone will put themselves into the situation of realizing that all ideas bring their own costs. It gives us better understanding of our current everyday world and also the past. It can cut through the discord. Because everyone who claims something will be subject to tests and forced to admit to costs. Every idea has costs. Every single one of them.A less rhetorical world and a more introspective world.

  8. lol @ “world is a place where low probability events happen fairly regularly”.
    Should make you think how good our probability estimates are in the first place, if this is the case.

    1. We are in a pandemic, quarantined. Some like nntaleb who keeps betting on these things make money. Others lose. And it therefore becomes an issue of estimation.

      1. Current pandemic isn’t a rare event. There were many leading indicators which many people ignored. Because most people don’t think in long game. People who think long game keep winning.

        People who study catastrophic events quickly see most of them were cascading events. Current quarantine is one of those cascading events. Again, not a low probability if you didn’t average it over time but acknowledged the time dependent nature of occurrence rate and potential sources of escalating consequence.

        Estimation is useless if the requirements are not precise. I can predict an earthquake occurring tomorrow. It is correct but useless because there is almost always a M3 or M4 earthquake recorded somewhere in the world. The requirement of intensity, a location and a duration are important if we are going to quantify probability. But probability by itself doesn’t matter depending on available control of occurrence and the consequence.

        I thought we were talking history though and not prediction.

        1. \Current pandemic isn’t a rare event.\
          In UK govt planning, there is a contingency plan for Pandemics- this for the past 15 years. Even planners did not take it seriously , even though it was in their books
          As they say, before pandemic hits , whatever you do is alarmist and too much ; once pandemic hits what you do is not enough.
          US govt under Bush Jr made plans for pandemics, looks like Trump cut the funds for it after he came to power.

          I am sure China is having a laugh. They brought COVID, and they were the first to control it fairly efficently – now rest of the world is strugglling. It is an unintended (or perhaps intended) cold run for a biological war. It also serves a propoganda coup for the Chinese system of governance which is a devil’s bag of tricks in western books.

          1. Yeah, like Nate Silver notes in his book Signal and Noise, it takes some expertise to extract signal from noise. For any large scale event, false positives have cost consequences and false negatives have safety consequences.

            People aren’t rational enough to do cost-safety conversions intuitively because of we (at group level) are biased to do a lot of future discounting.

  9. I am talking of humanities, history is a small example And I do not contest the value of history as it is presently written or its utility. It clealy has value. And one is not contesting that all of history is some conspiracy or anything like that. It would be better to instead have numbers assigned to studies in humanities.That is all.

    1. My point is that quantifying epistemic uncertainty for complex systems is difficult. If you try to quantify uncertainty with simplifying assumptions and without precise definitions, whatever estimation will be next to useless after so much effort.

      Slapstik has a better example in his comments.

      Data gathering is expensive in physical world. Nobody is willing to bear the expense when predictions from historical data aren’t very reliable. (Past isn’t a representative sample of future, particularly under incomplete system definition – this is true for physical systems like geology too). Optimizing for Value of Information (VoI) will tell you, the resources aren’t worth it.

      Personally, I don’t think probability framework is useful for quantifying epistemic uncertainty arising from ambiguity, incompleteness and imprecision. Although I am happy enough that regular folks are at least willing to acknowledge uncertainty and try to grapple with it, I don’t think trying a Bayesian framework for history (or humanities) is an useful exercise for the reasons Slapstik already mentioned.

  10. had trouble with format. was scared when google gave me hit of “https://en.wikipedia.org/wiki/On_the_Jews_and_Their_Lies” for jees.

  11. Your blog carries a certain apriori bias against the humanities on grounds that it’s largely a vehicle of ideology

    The devil is in the priors. Reverend Bayes cannot save you from it 🙂

    1. Ok. One could start off by looking at evidence of the materials that have survived from each time period. Out of which the amount that has survived in math/science etc. Do it for all literate societies of the past. And have some sort of score for that.Or as Middle Lion pointed out. The utility can be for the present though. Where one can test claims of ideologues etc. I believe experimental economics with randomized controlled trials is going down that path ( am not aware of the field at all). I am more interested in testing claims and cutting down errors. It might reveal some hidden information too.

  12. Phyecon1 makes some reasonable propositions. The time constant proposed is too long for an initial starting point. I would reduce the historical time constant to a month or a year and get going a probabilistic approach. This would be a useful analytics engine for development and a business tool for historians.

Comments are closed.