Wednesday, November 04, 2020

Guestpost: "Launch of the Initiative 'For a Smarter Science'" by Daniel Moreno

[This post was written by Daniel Moreno]

I remember when I first told a professor at my home University that I wanted to do a PhD and become a researcher. I was expecting him to react with enthusiasm and excitement, yet his response was a warning. That was the first clue I received about the academic world's inefficiencies, although I did not realize to what extent until many years later.

I was a postdoctoral researcher for six years after finishing my PhD in 2013, working on areas such as holographic QCD, numerical General Relativity and gravity dualities. It all came to an end last year, and I decided to rekindle an idea I developed with Sabine back in 2016, one that never came to fruition, but which has now turned into my current project, under the name 'For a Smarter Science'.

The precariousness of our scientific system is a topic of common discussion among informal circles in academic events and institutions. It should be familiar to readers of this blog, as well as it is to researchers from the many fields of Science. During lunch with colleagues, dinners with invited speakers, conference coffee breaks… conversation commonly drifts toward some version of it.

Increasing numbers of publications of decreasing relevance. Increasing numbers of temporary contracts of decreasing stability. Acceptance of bad scientific practice to the benefit of bare productivity. These are some of the criticisms and complaints, prompting warnings to young researchers.

Our initiative brings no original solution to any of these problems. It actually brings questions: What would happen if there existed a formal platform to discuss the current state of academic culture? What if there was a scientific method to approach issues about the way Science is done today?

If unsupervised, human structures tend to evolve in a predictable way. Issues such as the ones mentioned above are sociological issues, they form a sociological trend. This can be, and is, studied by experts who have been writing on the topic for decades. Educated analyses published in peer-reviewed journals.

These studies largely go unnoticed and/or dismissed by the people involved in the very scientific fields they talk about, for a variety of reasons. The assumption that social topics are naturally subjects of informal conversation only, the belief that intelligent people are not affected by cognitive biases, or the selection of like-minded people by the academic system itself, leading to communal reinforcement.

And so, the academic wheel continues running along the railway already set in front of it, with no one there to steer its course. Short-term thinking permeates research. Researchers go from one application deadline to the next. Academic metrics go unquestioned.

Our proposal is simple. One conference. One formal event bringing together experts from a specific subfield of Science (high energy theoretical physics) and from Sociology of Science, to elevate these questions to well-informed discussion. We trust that such an event has the potential to trigger a positive change in the historical development of all of Science.

Of course, not everyone necessarily agrees with the idea that scientific progress is being discouraged by the current peculiarities of the academic culture. This is why we find it fitting to fund the event by means of a crowd-funding call. Its success will be an effective measure of how worthy of consideration the scientific community considers these discussions. Are they just light topics for chit-chat during coffee breaks, or is it time to pause and actually examine the social evolution that scientific careers have been following?

This initiative is for all those who have been waiting for a chance to do something about the state of modern scientific research. Writing posts and discussing with our colleagues can only take us so far. If you believe in the place of Science as a shining light for human progress, make your voice be heard. #ForaSmarterScience



  1. Smart science seems a pleonasm to me. Smarter science? I guess that what is meant is more apt, more usefull science.

    1. That would be a more accurate description, yes. We just decided to prioritize 'catchiness' when choosing the name. I like to think of it as an aphorism.

    2. OK, in that case I would prioritize positiveness and choose the name For a Brighter Science.

      Interesting post anyway.

  2. You wrote that sociological issues regarding science have been "studied by experts who have been writing on the topic for decades. Educated analyses published in peer-reviewed journals."

    But you didn't talk about *any* of the results. Before talking about a crowd-funded conference, let's talk about the results published so far. What do we know already?

    1. We know, among other things, that job insecurity is increasing, and that job insecurity decreases the willingness to take risks, which decreases the willingness of researchers to embark on new topics. We know that academics routinely lie in grant applications. We know that citation and publication counts can and are being gamed, and we know that they are nevertheless still in use. We know that the current organization of the academic system rewards "bad science" because bad science tends to be more productive. I have references for tall this in my book.

    2. Thanks, Sabine,

      How about a few posts that review some of these results in detail -- one major result per post? I think that would be very useful.

    3. I understand that creating such posts requires a fair amount of work. If the material is all in your book (which I regret to admit I don't have), wouldn't it be fairly straightforward to extract and post it? How about Daniel Moreno? Since this is his initiative, he would seem to be the obvious person to do it.

    4. (a) content of my book is copyrighted (b) as the book is about physics in the first place I do little more than quoting the relevant papers

    5. Gah... I truly do not want to come off as "that guy" but you both have very good points. 1.) writing about a subject takes a lot of time and energy and 2.) the modern internet has a new and improved version of the Gish Gallop argument where increasing amounts of time and energy are demanded of writers to put everything into one place for the sake of the readers, who also do not have the time to look everything up for themselves.

      This low-key argument/debate is exactly why I suggested the series of technological solutions I placed in my comment blow this.

      Allow whomever has the time to share the information. Allow technology to collate. Do not think of a body of work as complete, think of it as growing collaboratively until it has reached its potential. Allow all contributors to be collected and credited by the collaboration system. Etc...

  3. Academia has produced an overabundance of scientists. With so many billionaires in the world (and soon to be the first trillionaire), they are a ready source of funding to save our planet. That effort is the first step in saving science from the wrecking ball of authoritarianism and nationalism. We need more than crowd-funding for physicists and social scientists.

  4. Bring together "experts from a specific subfield of Science (high energy theoretical physics) and from Sociology of Science". Please add experts from Financing of Science.

  5. For the purposes of keeping this comment short and to the point I wish to skip the arguments already presented in the blog post above and in multiple articles and publications concerning the problems surrounding Academia. Instead, I would like to jump straight to a proposed solution taken from the tech industry and data and information sciences: that of publication validation using automated systems and algorithms.

    The simple truth is that Academia is already a process and if the past century has proven anything it is that no social process is immune to the depredations of political change or human failure.

    Imagine for a moment being able to get real-time asynchronous semantic field research that includes sentiment analysis happening behind the scenes while you type out each section of your academic article.

    Consider adding sentiment analysis based on jargon-sensitive topic and a contextually-aware scope walking algorithm pointing out potential points of rebuttal or disputation and assisting you in the process of exploring and documenting your consideration of these issues.

    Imagine that after you address each issue, it is added to the ancillary data included with your publication and highlighted and used to flag rebuttals as they are being written, thus saving you the time and frustration of negative reviews being published by hasty, incompetent or disingenuous researchers competing for funding.

    All of this automatically sorted and added to the footnotes and citations as you work. All of it available for visualization within or appended to the research notes of the publication.

    When you are finally ready to publish, your work is a rich, dynamic, real-time and interactive tool able to assist in its own defense with your peers.

    Consider having the ability to allow anyone, even those without an accredited degree, the opportunity to contribute through the use of automated tools and publication submission to something akin to, a VCS applied to academic publication with automated peer review, popularity monitoring (stars and reviews), metadata analytics and to aid search parameterization.

    Between machine learning algorithms (MLA), NN<->Neurology inferrence allowing researchers to get glimpses inside the true empirical meanings of words like 'wisdom', 'experience' and 'intuition', and complex statistical analysis of evolving population groups in real-time.

    It's past time that academia modernize its work flows and everything I've outlined here already exists in functional form (if scattered and not at all optimized for Academic work, more for development operations.

    Take the monolithic human institution out, put in automated systems, and watch science take off on digital wings. Now that would be worth crowd funding.

    1. Sky Leach, I’m sure many agree with you that's the correct direction. The question however is, how do we make this work?

      PS: best wish to Moreno et al. for this initiative, but imho crowd-funding is premature without a clear plan of what one want to do with the money and why.

    2. That's the easy part TBH.

      While it might makse sense to form a formal working group, forming an open source commission or foundation with an associated publication journal and best-in-class suite of tools using pre-existing technologies would absolutely work.

      Most bibliographic and deep search tools, especially semantic search tools, use open source technologies and suites. Here are the major issues that would face an open source developer:

      #### Initial setup hurdles

      * Virtualization costs - Setting up a serious AWS stack or BigQuery or ElasticSearch stack that can handle the volume of academic publications is a university-level task. Universities haven't done it because it's a threat to an income stream. It's a multi-project proposition that has only ever been handled by corporations that are selling SAS (Software as Service) model business plans to research groups and making a lot of money doing it.
      * Journal Access and costs - Very few except educational institutions or corporations can afford access to journals. There are alternatives that provide free access, but they are legally grey areas and most FOSS developers and foundations will avoid such things because they have no legal budget other than the EFF. This means that the foundation needs to obtain and maintain some sort of subscription access until the overwhelming majority of publications use or agree to an open system of sharing information.
      * Academic credibility - Those that have the technical knowledge to build the system do not have the academic credibility. Those that have the academic credibility do not have the technical knowledge. In both cases, overspecialization has lead to an inability to ask the right questions or research the correct information. In addition, large services with both the philosophical agreement and the infrastructure to commit to such a project require a contact and working group with whom to plan the service. In my comment I mentioned something like Github has two service models at present: enterprise and public. Public projects are always available to be viewed and forked at any time by anyone. Enterprise projects are restricted and controlled by an enterprise of some kind. This initiative would be something different. Researchers would need the option to keep their projects closed until or unless it becomes published. There are other details to be worked in here as well, such as double-blind data access for IP-restricted peer review and public vetting.

      There will doubtless be more, but those are the large ones I see immediately. None are insurmountable by any means. I would see the software itself looking like or working like a more formalized and extended version of Jupyter Labs. Each project would be a self-contained notebook combined together with some kind of configuration file specifying publication parameters that would build out a virtualized instance.

      All of the previously discussed items are, however procedural. As for social hurdles, I firmly believe that the utility of best-in-class tools will encourage adoption en-masse provided there are no massive blockers like forcing open licensing that would stand in the way of industrial patents and IP stakeholders.

      Presently, all academic tools of this kind are cumbersome proprietary tools that depend on high cost rather than volume sales. Any kind of well designed and publicly supported tools that provided superior or even near-parity solutions would be of immediate utility.

      Finally, I see the tools as coming in multiple scope-based forms. Asynchronous linting (real-time analysis), deployment linting and bibliographic cross-reference, and then pre-publication lock-in build-out where all changes are frozen in preparation for publication. Changes after publication would be a new release, just as with software.

    3. Abstract:
      Won't happen and even if by some miracle it did it wouldn't work.

      Even as I switch myself into optimist mode and assume this is a genuine attempt to get science out of slump and not clever scheme to secure decent to well payed job for you and select few buddies for the next 30 years there are institutional and psychological hurdles. Former possibly surmountable, latter I'm afraid not.

      Let me start on purely conceptual footing though, you are talking about half birth. To my limited knowledge on the subject, no female has ever managed to do it.

      On institutional front; you are talking about squeezing thousands out of billions while creating yet another entity that will be same kind of a beast. Different colour of its fur won't make a difference. Because to win this battle you will have to play by existing rules, market rules. In doing so, even if it all starts far more idealistic than I suspect it would, it'd morph into more of the same sooner rather than later. But like I said, this might be somehow steered true.

      However, psych will not. You seem to have fair knowledge on how open software works so I'm surprised and/or skeptical about your apparent blind spot on this particularity. One of its foundation traits is near absolute absence of ego. True, implementation on the field is not seldom enough so far away from this idea that one checks more than once if one is on the right forum or mailing list during discussions but in the end, only product survives. When I start an open source office suite I have absolutely no idea who wrote that. For quite some time I wasn't even aware of existence of foundation behind it. Do you honestly believe scientific community is ready, willing and able to relinquish fame and prestige?

  6. It sounds like you are describing something like the Workshop you were in. I found the link because my professor was in the audience. Perhaps something similar could be organized over Zoom.

  7. "If unsupervised, human structures tend to evolve in a predictable way."

    I hope it is not the idea that physics research becomes "supervised" by some central authority. That would be a disaster and contrary to the scientific method. The mechanism by which research should be "supervised" is the comparison with experimental observations. I agree that there are systemic problems with how research is being funded, but the diversity in approaches is actually the strength of scientific research.

    1. Absolutely not, that would be detestable. That implication was not meant by the text.

      Your comparison is good, the idea is to frame research funding within a scientific context. Our initiative is not created to support one direction or another when it comes to solving the problems of academia, it simply supports creating a framework for informed discussion. We believe these are serious matters whose discussion belongs in a scientific context as mentioned above, as opposed to relegating it to blogs and social media.

  8. In thinking about this there really is no such thing as science. There are the sciences. Methods of analysis, experimentation and data work vary quite widely. We are not in the time after Bacon where we have the easily a single philosophy of science, or just plain science.

  9. Sooner or later (and it has already been all of the time since 1859, when "The Origin of Species" was published), you will have to do the REAL work, of changing your fundamental assumptions, and your beloved-but-ugly-lie theories. The underlying paradigm for today's scientific research is false, and all fields of science are tattered remnants of hopeful, but entirely false, speculations. In short, get are engaging (as always now) in obsessive and deluded avoidance behavior.

    And sociology has NO place in physical science (hint-hint!). Period.

  10. The problems have been known for 60 years and discussed to death. We don't need more discussion. What MIGHT help is a better mechanism for crowdfunding independent research.

    I've been thinking about the ancient institution of Trinity House in England, founded by Henry VIII, which has managed to retain its original purpose for 500 years. It was designed to replace false lighthouses (breakers) with true beacons, and to fund research in better lights, better electric power for lights, better foghorns, better communication, and so on. Marconi and Lodge, among others, benefitted from Trinity research funding.

    It's constructed as a Mutual Assurance Society, and funded by a small fee on every ship that docks in Britain. How has Trinity House stayed on its OWN beacon for 500 years while protecting shipping from actual false beacons? Every other organization and agency succumbs to Parkinson's Law sooner or later, losing its purpose and turning to triviality or crime.

    There's a lesson to be learned, perhaps by talking to insiders.

  11. posts by scientists for non-scientists are why i read this Blog. Thank you for hosting this post Bee.


COMMENTS ON THIS BLOG ARE PERMANENTLY CLOSED. You can join the discussion on Patreon.

Note: Only a member of this blog may post a comment.