Thursday, February 20, 2025

Fascism, the War on Complexity, and the End of Science
February 17, 2025
Source: Originally published by Z. Feel free to share widely.


Image by Mossadegh-keller and Mailfert



On February 11, 2025, the current US government published a website of the US Senate Commerce Committee (Chairman Ted Cruz) announcing the results of an “investigation” into research grants funded by the National Science Foundation (NSF) during the four years of the past administration. The announcement reveals that “the Biden administration weaponized federal agencies to push a far-left ideology.” The website includes a downloadable spreadsheet database listing 3,400 grants issued by the NSF that, in their words, promoted “advanced neo-Marxist class warfare propaganda.” This is obviously a serious allegation, that is being promoted as the motivation for a catastrophic restructuring and budget reduction of the NSF, that will effectively grind most scientific research to a halt and entirely wipe out the US leading edge in science on the international scene. What is going on? A close inspection of the downloadable database reveals a very different picture, so let’s take a look. For example, one may become a little suspicious of their claims when noticing that their spreadsheet includes a total of 113 research grants in pure mathematics. They cover topics such as “link homology theory and other quantum invariants” (2204386), “geometric aspects of isoperimetric and Sobolev-type inequalities” (2340195), “data driven inversion methods and image reconstruction for nonlinear media” (2308200), “four-manifolds and categorification” (2203860), “elliptic and parabolic partial differential equations” (2236491), “stability patterns in the homology of moduli spaces” (2202943), “stable homotopy theory in algebra, topology, and geometry” (2414922), “log-concave inequalities in combinatorics and order theory” (2246845), “harmonic analysis, ergodic theory, and complex geometry” (2236493), “matroids, polyhedral geometry, and integrable systems” (2332342), “commutative algebra in algebraic geometry and algebraic combinatorics” (2246962), “isoperimetric and Minkowski problems in convex geometric analysis” (2337630), and so on. It looks like Marxism really became a lot more advanced since last I looked! Flagged grants in theoretical physics similarly include such gems as “scattering in quantum gravity: symmetries and holography” (2310633) or “quantum cellular automata, many-body decoherence, and complex entanglement” (2210566). It may be a hard call to decide whether quantum gravity or quantum cellular automata make more effective Marxist propaganda. It is very important for the public to understand that, when we talk about NSF grants, we are looking at the most advanced scientific research carried out inside the US. The content of grant proposals is extremely specialized and is evaluated by a panel drawn from leading experts in the field of research. If words like “isoperimetric” or “matroid” look obscure to the reader, it is inevitably so: these are mathematical terms, which often require hundreds of pages of mathematics well beyond college level to be understood. Professional mathematicians have ten to fifteen years of specialized studies, and even then, they would not be considered competent enough to evaluate mathematics research proposals outside of their specific area of research. Since this witch hunt list of Marxist mathematics was not compiled by mathematicians or scientists, how did these grants end up included in this list? A few days prior to the release of this database, a document was leaked to the press from inside the NSF: it consisted of a list of words that were supposed to be used to flag grants for “wokeness” (whatever that means) and what is described on this website as the advanced neo-Marxist class warfare propaganda. The leaked list of words included terms such as “inequalities”, which is a very standard mathematical term (quantities ordered by a “less than” or “greater than” relation, rather than by the equal sign: an estimate, in other words). So, any scientific grant that estimates something (quite a common activity in science) and expresses that as an “inequality” (the mathematical word for estimate) gets flagged. This explains the “log-concavity inequalities” (a relation between coefficients of certain classes of polynomials), or the “isoperimetric inequalities” (an important estimate in differential geometry), both evidently considered instruments of class warfare. Other mathematical terms flagged included homomorphisms, homology, homotopy, etc., all of them technical mathematical terms where the forbidden prefix homo is nothing but the Greek word denoting similarity, as a mathematical concept, and has nothing to do with sexual preferences. Perhaps the “many-body” systems (systems of many interacting particles) studied in mathematical physics also seem too ungodly to the zealous censor’s eyes. All of this is of course absolutely ridiculous. Let’s look for a moment at a randomly chosen but fairly typical paragraph from the project summary of one of these grants: “Special cases of these isoperimetric inequalities are connected to an affine version of the sharp fractional Sobolev inequalities of Almgren-Lieb. The techniques involved in studying these questions are from convex geometric analysis and PDE. In the last few decades (particularly the last two), there has been a community-wide effort to extend results in the theory of convex bodies to their counterparts in the space of log-concave functions” (from the project summary of grant 2337630). I am sorry to tell you, comrades, but if this is Marxist class warfare propaganda, I am afraid it is not going to be so terribly effective.

It would be easy to just dismiss this nonsense as plain stupidity, but one needs to analyze all this more carefully. While some of the now identified DOGE centurions taking a wrecking ball to public funding of science are just college dropouts with ties to shady online neofascist groups, some of them have advanced degrees in science (including one of our former Caltech graduate students): even the worst students at Caltech know what inequality means in mathematics, or homology. So, in the mixture of buffoonery and malice that lies behind this released dataset of “woke Marxist science”, malice seems to be more preponderant than stupidity. It is important to keep in mind the target audience of this release, and the goal that it serves. The plan of the current government includes a full-frontal attack on science. Largely dismantling the NSF is part of this plan. So is the similar attack currently aimed at the National Institute of Health (NIH) that funds most medical research. The public should be aware that, for example, the sudden decision, effective immediately, to cut overheads of all NIH grant from the usual 60% or so to just 15% (overheads on medical grant manage, among other things the running and monitoring of clinical trials for experimental medical treatments) devastated ongoing studied of novel advanced treatment for cancer, effectively leaving out to die many patients undergoing last resort life saving treatments. The parallel attack on the NSF discussed here hasn’t been as deadly yet, but the combined effect of this two-pronged attack on the two major funders of science will cripple scientific and biomedical research in the US in an unrecoverable way. If a similar dismantling of public funding for science had happened in past times, the US would likely not have won the Second World War (OK, given this government’s ideology, I suspect they would not entirely dislike this scenario), but certainly it also would not have won the Cold War, fought for decades against another world superpower that heavily invested in top level science and technology as a national priority. Indeed, in the past, Republican administrations, even in their most reactionary forms, have always maintained the idea that funding science was important. Such sectors of the Republican voters base still exist today. They still believe that the US should maintain a leading edge in science and technology in the world, and that public funding should support this effort. Certainly, the sector of the Republican electorate that is squarely anti-science is huge, but with a slim parliamentary majority and the full consolidation of more dictatorial antidemocratic power still underway, the government cannot afford to lose that small sector that would not support a point-blank unjustified dismantling of NSF and NIH. Thus, the witch hunt for such chimeras as isoperimetric matroid quantum Marxism is purely a spectacle, for the consumption of this sector of the Republican electorate, that needs to be convinced that it is not science per se being attacked and dismantled (it is). They are being told that the drastic measures undertaken are a heroic and necessary attempt to save science from its current takeover by “far-left ideology.” The list of flagged grants is risible: there is no ideology in any of these research projects in pure mathematics or theoretical physics. Scientists, and mathematicians, generally tend not to be very politically active and focus their energies almost entirely on their science. This list only exists as a ploy to justify to a sector of their voting base an action that will lead to the effective destruction of scientific research in the US, without explicitly saying that that is the real goal.

This shifts the question, then, to a very important one: why is this government so different from all its predecessors, even the most conservative and reactionary US governments of the past? Why this literal armed crusade against science, carried out by a vanguard of operatives with no clearance or authorization, no congress oversight, in violation of the constitution and of all the mechanisms of separation of powers? Why is ending science so urgent to them? I think this is highly revelatory, a key to understanding what is really going on politically in the country. It is unfortunate that the term “fascism” has been, at various moments in time, abused to generically indicate any episode of violent authoritarianism. This makes the press, perhaps understandably, apprehensive about using the term, even when it is in fact warranted. They prefer to use euphemisms such as “constitutional crisis” (the situation of a democracy in a constitutional crisis caused by its own head of government is a coup d’état, also known as an autogolpe.) So far, that I am aware of, the only centrist liberal voice that publicly called what has been happening after January 20 a fascist coup d’état is the Yale historian Timothy Snyder, an expert on the history of central Europe in the 1930s and 1940s. Snyder gave his explanation, in his capacity of famous historian, on why he thinks it is the appropriate term to use in the present context. I will give here my own explanation, as a scientist: it is based on the notion of complexity.

Before doing this, let’s take a step back in history of about exactly a hundred years. It is useful to recall a basic chain of events that lead to the rapid transformation of a democratically elected government into a dictatorship, the eponymous one from which the term “fascism” is derived. The year is 1924. The place is Italy, at the time functioning as a democratic state, in the form of a constitutional monarchy. On April 6, 1924, Benito Mussolini wins democratic elections with a significant majority in the parliament and is appointed prime minister. On May 30, the head of the parliamentary opposition, the socialist Giacomo Matteotti, presents in a speech to the parliament detailed evidence of voter suppression, illegal power grabs, and unconstitutional actions by the Mussolini government. On June 1, Matteotti is kidnapped and murdered by Mussolini’s squads. On June 26, the entire opposition declares a boycott action aimed at stalling the parliamentary work (the Aventine Secession). On November 8, the opposition presents evidence that Mussolini ordered the Matteotti murder to the king (who had at the time the constitutional role of appointing the prime minister). Citing concerns of civil unrest from violent Mussolini squads and supporters, the king refuses to dismiss him as prime minister. On January 3, 1925, Mussolini calls a vote of confidence in the parliament, after ousting the entire opposition. Shortly thereafter he declares by executive order all the opposition parties illegal and starts a single party dictatorship. Italy remained a dictatorship for 20 years, until April 1945. What one should extract from this brief timeline is how fast a supposedly stable democracy, with checks and balances of power, can crumble, and the mechanism by which this happens: unconstitutional power grab within governmental institutions. About ten years later, it took Hitler an even shorter time, just about two months, to go from democratically elected government to full dictatorship, following the same path. The very fast chain of power grabs and unconstitutional moves by the current US government are following a well-established template. Mussolini was an operetta buffoon and at the same time a deadly threat. The risible grotesque aspects of the Trump government are extremely similar in form and nature. But the argument by historical similarity, striking as it may be, is not what I want to focus on here.

The Physics Nobel laureate Giorgio Parisi once said that “the denial of complexity is the essence of tyranny.” This is the main point that I want to address as the deeper source of the current political crisis and of the looming end of science (and likely of democracy as well) in the US. The historical fascism of the 1920s was an emergent phenomenon, a violent reaction against the rapidly growing complexity of modernity. Fascism is in essence an agrarian fantasy: the “sheaves” in the historical symbol of “fascism” hinting to an essentially rural idealization of society, rejecting complexity and urban modernity, pandering a mythological image of an unspecified past whose return is to be sought. Fascism, old and new, rests upon the dangerous delusional idea that a strongman leader can, by share willpower and violent authoritarianism, replace and eliminate the deep, widespread, interconnected network of human competence that is modernity.

In relation to the dismantling of the Department of Education, Trump stated that he “prefers uneducated people”. A society in which there are treatments for cancer and airplanes fly safely, where your everyday bank transactions are secure, where communicable diseases do not decimate the population as they regularly did in pre-modern time, require a continuous influx of highly educated people. Each of these things, and many others, require advanced science and can only exist because of people with years of higher education, with twenty years of their lives spent in school. All these things will cease to exist, the moment an influx of people with the right specialized competence would terminate. What prevents societal collapse, what maintains viable safeguards on the food we eat, the medication we need, the nuclear weapons in our arsenals (whose controllers were all “accidentally” fired by Trump’s Department of Energy), is the fact that education provides us with the capacity to handle the enormous degree of complexity that the modern world requires. The mechanisms that allow large scale distribution of services, large scale infrastructures and communication, rely on a human network of intertwined specialized competences. The war on competence that is being waged, which includes terminating the employment of scores of people that hold the expertise required for running the intricate network of societal infrastructure that the modern state interfaces with, nominating science denialists and grossly incompetent people to guide the main government agencies, and a flurry of other actions taken without any careful analysis of their consequences, will be rapidly leading to societal collapse.

Vance, when he is not busy trying to lecture Germans about being more tolerant of Nazism (that does take some stomach!), he loudly complains that “so much of what drives truth and knowledge as we understand it in this country is fundamentally determined by, supported by and reinforced by the universities,” resulting in calling academia the enemy and advocating its complete destruction (despite having completed his own education at Yale.) This is once again the same violent reaction against complexity that we are witnessing. Truth and knowledge require deep and complex approaches, long and difficult investigations, professional expertise often well beyond what is immediately accessible to the general public. Foundational scientific research, which is almost entirely carried out in academic environments, deals primarily with such intricate complexities. One of the goals of academia is education and the purpose is to extend as much as possible the access to that complexity to the population. The more educated the general population is, the more the complexity of the modern world is manageable and in fact highly beneficial, because it is full of possibilities. The more the majority of people are deprived of any access to the tools and methods needed to handle complexity, the more it becomes frightening and unmanageable, leading to the kind of violent reaction that constitutes fascism. Any form of complexity, from our anatomies to the transmission mechanisms of diseases, to the large-scale changes of the environment, and to forms of technology that risk running out of control for lack of adequate theoretical understanding, frightens them. The reaction to the fear of complexities, instead of being one that supports learning, as our best path to understanding and ultimately managing reality, is to destroy anything and anyone who would as much as acknowledge the existence of an intricate network of intertwined complex systems organizing the texture of reality, which can be studied and understood.

Fascism was always steeped in magical thinking. Historically this tendency has a range of different manifestations, from an obsession with occultism among the nazi power hierarchies, to the case of Evola’s system of “high magic”, and to Giani’s “fascist mysticism” in Mussolini’s Italy, advocating a blind faith without reason, ultimately a belief that the world can function simply, literally by magic, directly emanating from a supreme leader, bypassing any form of complexity. The actions of the DOGE group carrying out Trump’s coup d’état reveal the same form of magical thinking, destroying anything that rests upon a delicate balance of many intricate layers of intertwined complexity and replace it by pure magical thinking. Since magic, in reality, does not work, the actual result is catastrophe.

But is the US really transitioning (or has it already transitioned) from a democracy to a fascist dictatorship, in the full sense of the word? In a short span of time the answer to this question will become self-evident, but for now let’s look at the timeline of events. A first critical indicator is the disregard for the constitution and the rule of law. Congress has the power of the purse, of creating government agencies with a specific mandate, and assigning (or cutting) their budget. Congress also has power of oversight on government agencies. An unelected group of people whose names had at first been kept secret (until they were eventually leaked to the press) is taking orders directly from Musk (an unelected individual). None of them has any prior experience working with government agencies. Some of them are young college students or college dropout with prior issues with mismanagement of data and of privacy, or with ties to online far-right extremism. They have invaded and taken possession of computer systems in various agencies including the treasury, extracting and exposing private data of Americans (including social security numbers, bank accounts used to pay taxes or get tax refunds, home addresses). Court orders attempting to stop their illegal takeover are being disregarded. Impunity is clearly implied for any crime they may be committing under the law. The system of checks and balances of power that democracy rests upon is in shambles. While we have not seen yet, as of today, widespread acts of violence carried out against sectors of the population, as the consolidation of absolute power and the dismantling of the constitutional system progresses at breathtaking speed, it may not be long until that becomes the next, more significant, worry.

One of the first acts of the DOGE assault team was the ongoing dismantling of the USAID government agency dealing with the US government’s involvement in humanitarian aid and development, handling work that includes treatment of HIV in various developing countries, support of educational initiatives, monitoring of risks of famine. Of course, the sudden unannounced elimination of all these programs has disastrous consequences for all those who benefited from them, at a huge humanitarian cost. There is, however, another interesting political aspect to the destruction of a humanitarian aid agency that has successfully existed for several decades. USAID did not just serve an important role for the programs In developing countries that it supported: its existence was primarily serving an internal political use, aimed at generating in the US population (and the population of Western European political allies) an impression of benevolence of the US foreign politics, centered on development, human rights, humanitarian concerns, and the fostering democracy, and diverting attention away from the much more sinister aspects of US foreign politics, seeding fascist coups and supporting atrocious regimes all over the globe. Agencies like USAID serve the propaganda purpose of manufacturing a benign view of US foreign politics, creating a “myth of American idealism” (about which Chomsky and Robinson recently wrote) for the consumption of the population. So why is this idealistic myth, so successfully created, is now being dismantled by the US government itself? Why is it no longer needed and considered a waste of money? The answer is the ongoing transformation of the US from a democracy to an authoritarian state. In a democracy, persuasion is the most primary method of generating consent in the population, and traditional media have for decades played that role. In that framework, the myth of American idealism served to quell internal dissent, to shift the focus away from what the population may otherwise have objected to. The situation has changed: traditional media are no longer the primary way to influence people’s opinions, as much faster and efficient spreading of misinformation can now be achieved, on a scale never before seen, over the much more fine-grained system of social networks. At the same time, a significant part of the population does not want a benign image of US foreign politics: the now openly fascist part of the US electorate is perfectly happy with the US involvement with Pinochet (as the ubiquitous helicopter memes periodically remind us of). For the remaining part of the population that still would find the truth about US foreign politics objectionable, the fact that generating an illusion of benevolent idealism as a persuasion technique is no longer needed suggests that other forms of persuasion, usually not available in a democracy, are going to be deployed.

In addition to all the considerations discussed above, there is another factor added right in the middle of this already highly volatile compound. Language is our fundamental tool in a democracy: it is through language that we can compare and discuss ideas, that we can propose and evaluate alternatives, and that we can uncover lies and counter propaganda. Democracy places language at the center of the mechanism, voting, that selects governance. The last couple of years have seen the sudden development of mechanical ways of producing text and speech of near human quality (differences exist and can be detected and studied, but for practical purposes these are simulators of human language). Not only these large language models (or LLMs, as the current AI architectures involved in the automated production of language are called) can simulate language produced by humans, but they can produce enormous amounts of text in no time at all. They can easily flood all the main communication channels used by humans to connect with each other and spread large amounts of cooked up disinformation. In fact, LLMs already have on their own a built-in tendency to “hallucinate”, to produce completely false information delivered with great confidence and assertiveness. Moreover, there is an additional aspect of the current language generators that is potentially extremely dangerous: it is called alignment. This is a process designed to ensure the safety of LLMs against malicious users and fraudulent behavior, and also to ensure adherence of produced content to basic tenants of human ethics. This is achieved through a combination of methods: non-recursive and scalable oversight, the first relying on reinforcement learning and supervised learning via labelled human feedback data fine tuning the LLM training, and the second designed to handle scales at which human oversight becomes intractable, relies on a decomposition of tasks into smaller size independent processes that can be treated either in parallel or sequentially via human-level approaches. Alignment with human values and human ethics is of course the crucial point here, because not all humans agree on values and ethics: fascists are likely to consider the persecution of groups they deem inferior to be justified, and even the violent elimination of such groups in their entirety through violence and genocide. Thus, what may be seen by most humans as basic adherence to shared ethical values, would be seen by a sector of the population as a bias against their own ideology of choice, which has the tendency to violate what most consider basic human ethics. Thus, for example, a few days ago, one of Musk’s posts on his own social media X shared a meme made by someone else, showing an image of a robot with a communist hammer and sickle painted on this forehead accompanied by a complaint about Chat-GPT (currently the most popular LLM) being aligned to leftist ideology. Musk accompanied the post with a comment that simply said: “Very much!”. Days later Musk attempted to buy the company OpenAI that produces this and other AI products. The company refused the buyout, but its intention was clear: gaining control of the primary producer of automated language and changing its alignment to represent “values” compatible with the fascist ideology that he and the government he participated in represent. Thus, according to this plan, AI generated text, preferably aligned to fascism, will replace through its flooding capacities, human democratic debate. False information requires time and careful investigation to be debunked: even when it is obviously false, one still needs to present evidence, to make a case. If it is produced as a firehose by a machine acting at superhuman speed, it becomes impossible for humans to counteract the resulting flood. It would be a massive jamming process making communication of anything with actual content effectively impossible. There is another sinister aspect to DOGE’s attempt to leverage AI into their unconstitutional takeover of all government agencies. The idea they already openly outlined to the public, of replacing scores of highly trained specialized workers at the various agencies with AI, in the current form of a (properly re-aligned) LLM like Chat-GPT, is disastrous on multiple grounds: current AI has no actual knowledge or understanding of reality, it does not reason or model reality, it simply produces text, one word at a time, based on a probability distribution it computes on the basis of huge amount of text it is exposed to. The complex systems that govern services provided to our society, medical services, transportation, payments of pensions, require an understanding of reality and of its intricate interconnections, that AI is incapable of in its current form. Let’s return to the case of the National Science Foundation that initiated our discussion. Imagine replacing a panel of world leading scientists that evaluates extremely specialized and technical grant proposals at the frontier of knowledge, with an automated text generators trained on generic text on the internet. It is well known that, if presented with an even simple mathematical question, Chat-GPT will reply with nonsense, usually word salad, sometimes something that locally sounds like mathematical jargon, but signifying nothing. That thing is supposed to efficiently replace, in the magical thinking of contemporary American fascism, days of intense discussion and scrutiny by a panel of scientists! Again, like all of fascism, this is both risible and absolutely terrifying at the same time.


Dr Matilde Marcolli is the Robert F. Christy Professor of Mathematics and Computing and Mathematical Sciences at the California Institute of Technology (Caltech). She received her PhD in mathematics from the University of Chicago, and subsequently worked at the Massachusetts Institute of Technology, the Max Planck Institute for Mathematics, the University of Toronto, the Perimeter Institute for Theoretical Physics, and the California Institute of Technology. She is author of 7 books and over 150 research papers in mathematics, theoretical physics, cosmology, information theory, and linguistics. Her most recent work is the research monograph in mathematical linguistics: Matilde Marcolli, Noam Chomsky, Robert C. Berwick, “Mathematical Structure of Syntactic Merge”, MIT Press, 2025. This article is based on the author’s personal views and in no way represents any of the institutions listed here.


ZNetwork is funded solely through the generosity of its readers.  Donate


Matilde Marcolli
Dr Matilde Marcolli is the Robert F. Christy Professor of Mathematics and Computing and Mathematical Sciences at the California Institute of Technology (Caltech). She received her PhD in mathematics from the University of Chicago, and subsequently worked at the Massachusetts Institute of Technology, the Max Planck Institute for Mathematics, the University of Toronto, the Perimeter Institute for Theoretical Physics, and the California Institute of Technology. She is author of 7 books and over 150 research papers in mathematics, theoretical physics, cosmology, information theory, and linguistics. Her most recent work is the research monograph in mathematical linguistics: Matilde Marcolli, Noam Chomsky, Robert C. Berwick, “Mathematical Structure of Syntactic Merge”, MIT Press, 2025. This article is based on the author’s personal views and in no way represents any of the institutions listed here.

No comments: