Wittgenstein’s Revenge

A short series by Mike Elias

We treat facts like they’re “atoms of truth” — small, indivisible, solid — and if you add them up, you get “big truths.”

But like atoms, facts are mostly empty space, and the closer we examine them, the less solidity we find.

It may be time to graduate from the metaphor of facts completely, to a metaphor that reflects a healthier relationship between truth and people.

Problem 1: Facts do not objectively exist.

We think of ‘facts’ as being somewhat equivalent to data — they’re observations, which in theory anyone could verify. 

But observation is just one of three essential ingredients required to make a Fact.

The other two are:

1) Context omission, and

2) Trust, in the one omitting the context.


Context Omission

In order for a fact to be useful, you need to decide where that fact ends and other facts begin.

To create a fact thus requires curating — that is, omitting — context.

Without omitting nearly all context, there would be no way to distinguish “the temperature in Guam at 12pm Monday” from the state of the rest of the entire universe at the same or any other time!

To establish a Fact, someone must decide what context is relevant to include, and what to exclude.

The trouble is, there is no objective way to decide what context to omit. Context omission is inevitably subjective — it’s whatever the omitter decides is irrelevant.


The necessity of context omission means when we encounter a Fact, we must either review all the context ourselves to decide whether we agree with the context omission decisions (which is impossible, or at least infeasible) — or simply trust the context omitter.

Trust is needed because context omissions can be made dishonestly in order to deceive people. The book How to Lie with Statistics is a crash course in the dishonest omission of context.

All Facts require observation, context omission, and trust. 

Even instruments of data collection have this implicit 3-pillar recipe:

  • Observation — “the sensor reports it’s 78 degrees outside”
  • Context — “a buffalo stepped on the sensor last week”
  • Trust — “the sensor was built by Volkswagen, who lied about their vehicles’ CO2 emissions”

Alternative facts

Without questioning the veracity of observations, if we distrust the context omitter or disagree with their omission decisions, an infinite vista of justifiable debate opens before us.

Neil DeGrasse Tyson will tell you there’s “no evidence for the existence of UFOs,” in the sense that they are spacecraft built and piloted by extraterrestrial beings. If you trust NDT, you might call this a Fact. If you don’t, you might disagree with the context he omits, such as:

  • Objects traveling 5,000 miles per hour and making 90-degree turns without slowing down have been recorded by radar that can only detect physical objects.
  • People have been reporting sightings of flying craft since long before earthly aircraft were invented.
  • Astronauts, generals, admirals, and executives of aerospace and defense companies have said unequivocally — even on film — that extraterrestrial spacecraft do exist, and that we’ve known about them for a long time.

If you decide context like this is important, and you trust it, it creates an “alternative fact,” in a completely genuine and non-insulting sense of the term.

Perhaps not all “science denial” is denial at all — perhaps much of it is honest disagreement about omitted context.

Context Omission and Trust are fundamental ingredients in the creation of facts, and both depend completely on subjective judgment. Facts are not “indivisible atoms of truth” — instead, they seem impossible to define with the objectivity we’ve come to expect from them.

Problem 2: Facts do not persuade.

“Facts inherently persuade” is the myth of our era.

While many of us realize this intellectually, society’s response to fact-skepticism still resembles the redneck response to meeting a non-English speaker: “Just say the same thing again, slower and louder.”

Ironically, “the fact that facts don’t persuade” has yet to really sink in.

The verdict is clear: Research on numeracy, cult behavior, science denial, and scientific suppression shows people tend to be far more concerned with the social-psychological and identity implications of a proposition than with its truth value. (We can discuss why you shouldn’t blame the public or bemoan this tendency in another essay.)

If the future of humanity depends on persuading people to be rational, we must take a rational approach to persuasion, and “just give them the facts — slower, harder, and with mustard” decidedly isn’t it.

The future of humanity does depend on persuading people to be rational. And in the same way nobody cares who first spotted the Titanic’s fateful iceberg, nobody will care who “was right about the facts” if we all end up underwater.

“Being right about the facts” in 2020 is merely a fashion statement — like wearing a mink coat on the Titanic. Who gives a fuck? Steer the god-damned ship.

Problem 3: The metaphor of Facts is epistemically degenerate. 

The metaphor of Facts is anti-Science.

Science holds all knowledge as tentative and uncertain. 

While people tend to accept this in theory, the metaphor of Facts connotes certainty and permanence, and inspires a fervor often indistinguishable from religious fundamentalism.

The metaphor of Facts thus creates a countercurrent to the spirit of science in public discourse and everyday life.

The metaphor of Facts encourages binary thinking: “I’m right, you’re wrong, period.”

  • “Sex is a sin.” No, it’s more complicated than that.
  • “Grains are healthy.” No, it’s more complicated than that.
  • “X is a fact.” No, it’s more complicated than that.

The metaphor of Facts implies a moral obligation to believe.

A moral obligation to believe holds a gun to the superego: “Believe this, or you’re a bad person.” 

Does this sound like an appropriate attitude for benevolent educators to have?

Most educators are benevolent and deserve no criticism — but many have been duped into using this presumptuous, anti-freedom implication via the metaphor of Facts. As a result, they have inadvertently spawned bands of rebels bearing a legitimate grievance.

The metaphor of Facts implies a moral obligation to convince others to believe.

Given Facts imply a moral obligation to believe, every fact is a micro-Bible, demanding its own micro-crusade to carry it forth to trample the heretics.

Note how “science deniers” and “conspiracy theorists” are persecuted. These crusades may be nonviolent, but they are still crusades.

The metaphor of Facts justifies bullying and shaming of doubters.

Facts license social viciousness in the name of epistemic justice.

Recall this warning from Brave New World author Aldous Huxley: 

“To be able to destroy with good conscience, to be able to behave badly and call your bad behavior ‘righteous indignation’ — this is the height of psychological luxury, the most delicious of moral treats.”

The metaphor of Facts is an escape from the responsibility of personal judgment.

The metaphor of Facts is a flight from the existential responsibility of exercising personal judgment. “If there are facts, then I am absolved of the error if I believe them and they turn out to be false! Right or wrong, I am innocent.”

The converse is “If the world goes to hell, it’s the fault of the people who disagreed with me. There is nothing more I could have done.”

With this pair of delusions, fact-crusaders see themselves simultaneously as totally righteous and totally helpless. (This mirrors a definition of narcissism I’ve heard — “wanting to be both the boss and the baby at once.”)

As the presumptive base unit of epistemic reality, the metaphor of Facts seems designed to excuse humanity from any exercise of the rational functions whatsoever.

The metaphor of Facts has been so effective at obstructing public discourse that I’m beginning to wonder what intelligence agency popularized it. It almost seems too deftly counterproductive to have become important by chance. Are “facts” epistemic dog food for the proletariat — unfit for bourgeois consumption?

Problem 4: The metaphor of Facts is incompatible with freedom of speech.

The inevitable result of the belief in facts is censorship.

Like fact-based crusades, fact-based censorship is already widespread on social media platforms. Censorship is the default — the first line of defense — against doubt and uncertainty in a fact-based world.

How humanistic can the notion of Facts be, if it produces censorship at the first sign of stress?

Does “being right” make thought-policing right?

Ultimately we must decide what’s more important — freedom of speech, or the metaphor of Facts.

Revising our expectations

Accepting the failure of the “fact-bombing” approach to consensus means abandoning several oft-proposed solutions:

Fact-checking algorithms, no matter how open-source

Someone has to build the algorithms. Someone has to fix them if they “don’t work.” Someone will say they don’t work, because “if they did, they’d say [my opinion].” So they’ll fork it. And we’re right back where we started — tribes “choosing their own facts.”

The “blockchain of facts”

A growing movement aims to build a “decentralized database of facts,” which will be used to settle debates, perform fact-checks, and prevent would-be tyrants from gaining power by lying about the past.

The intent is noble, but as we explored earlier, the “facts” that can be written to a blockchain are vanishingly narrow in scope, and useless without human interpretation. If this distinction is conveniently forgotten or obscured, a tyrant could easily claim blockchain-level indisputability for propositions that don’t qualify, for reasons only experts understand.

Do we really want the people’s freedom to depend on their willingness to defend the distinction between ‘blockchainable vs unblockchainable facts’ against leaders who would abuse it? 

Trump tweets are a portal into the future

Political risks aside, whatever the blockchain of facts could discover about consensus has already been thoroughly demonstrated by Trump Tweets. Trump Tweets have shown us that access to, obviousness of, and even universal agreement on the facts often achieves nothing for public discourse. Everyone agrees on the exact words he said, but their interpretation remains individual and tribal.

The blockchain of facts may lead humanity to the world’s purest water, but it will not make us thirst.

At best, the blockchain for facts will inspire a great and unfounded hope, followed by a great and inevitable disillusionment (and how many more disillusionments can we endure?).

At worst, it will turn cypherpunk liberators into the Orwellian tyrants they’ve spent their lives fighting against. Seeking truth is great — but mingling truth-seeking with ambitions about consensus is one twitch away from the belief that “forcing my truth upon others is a good thing.”

How precarious!

To be continued.

Get Ribbonfarm in your inbox

Get new post updates by email

New post updates are sent out once a week

About Contributor

Ribbonfarm is a longform blog featuring a variety of themes and perspectives.


  1. possible typo?

    may lead humanity to the world’s purest water, but it will not make us thirst > quench our thirst

  2. Having an immutable data repository WOULD make it harder to revise history though: any history revision starts with burning all the books. It’s SO much easier to build the narrative you want on evidence you make up than it is to explain others’ evidence your way. Of course immutable data won’t make revisions impossible, but so what. All security is about raising the cost of attack, not preventing the attack in all possible worlds.

  3. Why do you think that this idea should be called Wittgenstein’s revenge? Although I find your argument very compelling and will use it in my own work, your references are not so clear. And I am just curious.

    • Thank you, Henry. :)

      In Language Games, Wittgenstein told us that expressing knowledge in language is really only meaningful to the extent we agree on the rules of the “language game” we’re playing.

      Modern society seems to have forgotten Facts are a language game. At some point, we *agreed* to omit context, and to trust fact-finding authorities, because it was effective and convenient to do this.

      Now, we seem to be experiencing the hangover that comes from forgetting this was merely an agreement, not a philosophical reality. We cut corners for a good reason, then forgot there were corners missing, and now that we need corners, we don’t know what to do with ourselves. ;)

      • There was never such an agreement, not even in philosophical faculties practicing philosophy as language analysis. Consent on anything has always already been simulated but good enough for many to hold the belief that consent was possible, that revisions may be kept smaller and smaller.

        Now, we seem to be experiencing the hangover that comes from forgetting this was merely an agreement, not a philosophical reality.

        By now many people seem to understand “framing” and the adversarial use of context. They do get that even the hell is a “mostly peaceful” place depending on the circle, on the seat-guide and what is meant by “mostly”.

        • For what it’s worth, I think the “forgetting” was purely on the side of the experts forgetting to teach successors about the full Straussian bargain upon which their role was forged. The agreement had been developed between the social roles of “layperson” and “authoritative expert,” certainly not between experts. To inhabit the role of a layperson is to take on the stance that one is ignorant of a topic in contrast to the experts, a stance that is becoming less politically popular. Now inter-expert disagreements have become harder to hide for the purpose of presenting a united front to the masses, and the implicit consent to be governed by expert authority is being explicitly revoked.

          • The agreement had been developed between the social roles of “layperson” and “authoritative expert”

            How can I understand the bargaining between “social roles” ( and how is the division between the doctors and the laymen “modern” )?

            Sorry, this sounds like abstract nonsense to me, just wrapped into sociologese – no one expects scrutiny to anything which is prefixed with “socio” or “social”.

      • I see, thank you for the clarifications. On the other hand, one of my main objections to this would be that facts could be considered as a short way of saying that something seems obviously truthful within the understanding that our perception in general is limited and biased. So therefore, we do play the language game, even with the understanding that we could be wrong. We just use the word “fact” to indicate high probabilties. Anyway, I think that most of your analysis still holds for an understanding of “fact” as undeniable truth. It’s just that I have a hard time to get words like this and therefore try to find easier ways to establish connections to similar or parent concepts like perception and knowledge.

  4. This reminded me of Venkat’s thesis in Tempo about calculative vs. narrative rationality. Basically, a(n evolving) story in our heads is the heuristic for setting context boundaries.

    • A “language game” for Wittgenstein has a more precise layout than a “narrative”. As an example W. discusses the “composition of a wooden chair”. One doesn’t need a narrative to shape context in order to see that one has to perform domain design to disambiguate the meaning of “composition”. Composition from parts and composition from atoms are just two obvious types. In case of a chessboard he also discusses composition by colors and shapes. The Philosophical Investigations ( published in 1953 ) are a close contemporary of Category Theory. W. had a keen interest in the foundations of mathematics but I don’t know if he knew about the work of Eilenberg / Mac Lane which goes back to the 1940s and has a similar vibe.

      Narratives are far more memetic creatures which have survived on the “market place of ideas”. You can attack them with facts for sure but those who believe in a narrative fire back and attack facts by (re-)framing them. Adversarial contextualization is the norm as it seems and as I learned last month, not even basic math is immune from it but that’s for another discussion.

  5. I wrote on something very similar here: https://wearenotsaved.com/2020/08/26/justice-mercy-data-evidence-blm-and-qanon/

    It’s long but essentially I had just completed The Master and His Emissary by Iain McGilchrist which posits that the modern world suffers from an over active left hemisphere, which is primarily concerned with grasping things, and breaking things down. Facts perfectly fit into this category and so people are obsessed with facts, but only a few facts, and only the ones that fit their narrative. The key point being (much the same as this post) that emphasizing facts and evidence isn’t the solution to our problems it’s the cause of them…

  6. You can lead a horse to the most accurate facts but you can not make him … something something something. On the other hand, 2 + 2 = 4, I am pretty sure, so your mileage may vary.
    Seriously, yes, Americans particularly need to start building unity and shared context so that we can to a much greater extent share facts. America’s leaders need to build their rhetorical, presentational skills in order to actually *persuade* with facts.
    I’m gonna stop now cause my IQ is probably not within the tolerance zone of this blog.

    • Seriously, yes, Americans particularly need to start building unity and shared context

      Project 1619?

      Maybe a proper multicultural society has to go down the road of embodying of a byzantine egalitarianism, one in which everything is regulated by quota and a strict etiquette which avoids ever mentioning the differences in upbringing, temper and skill. In Americas case, critical whiteness is certainly necessary to get there.

      It might be funny that this should happen to the most violently dynamic, and creative-destructive society in modern history and not to a more ceremonial East Asian state but if America decides to go trans and reinvents herself as a byzantine socialist, no one but herself can prevent her. Europe, Russia and China certainly won’t intervene and press for regime change and color revolution.

  7. Excellent article Mr. Mike Elias. I have been saying in my head for a while now that the truth is dynmaic, and while I’m sure I had ‘connections’ in my head for thinking so I never really solidified my thoughts leading up to that conclusion. Some people will get this article, and some people will deny it, whether it is a lack of want to know, or able to know, we will never know. ;)

  8. Interesting stuff. A lot of the issues highlighted in this article are what happens when there is a lack of trust in the person/technique omitting the context. Facts are part of technical knowledge and are anti-democratic. But the constitution of the world is non-democratic. We definitely don’t want the design of everything (aircraft for instance to be democratic). It is only the constitution of our democracy that can be democratic. Murdoch Media/Fox have run a campaign to deligitimize any form of expertise beyond business expertise so that these groups have the most power in public debate. Wedge and block is the name of the game and has an epistemological dimension and resulting culture war you have so clearly illustrated.

  9. Well, I think I’m going to have to be boring today.

    Science doesn’t hold all knowledge as tentative and uncertain, this is a classic flaw in reporting of science; science defines current levels of certainty on a continuum of uncertainty, that has “bold exposure of the flank of your hypothesis to methods of falsification” as one of it’s criteria underlying broader acceptance.

    This is why, for example, scientists can talk boldly about the lack of mass of the photon, or the existence of climate change, and so on; because they have a broad history of failing to be wrong, and a series of counterfactual consequences, the searches for which have failed to provide any results.

    There is a bravura element to science, and though of course, saying “more research needs to be done” is common, creating an assemblage that seems fragile and yet refuses to break is how falsification, in practice, fits into a profession focused around establishing verified regularities.

    The paradox of science is that the performance of uncertainty is part of the generation of certainty, because by generating “inconvenient ghost facts”, gambits placed before the budding conspiracy theorist, for them to attempt to demonstrate their facts and reveal the procedures by which they can be accessed by others, that if they exist, will bring down the edifice entirely.

    Then a new game begins about fending off these new applicants, showing that neutrinos did not actually travel faster than light under the italian mountains, or that a given quantum information transfer did not actually duplicate an entire quantum state, and so on.

    This is a battle of facts, but it does have a relatively determinate way of proceeding, because you begin with a method of analysing the world, born from a particular set of rules, and focused on looking at deviations from those rules.

    • Thinking about this more, I think your proposal for the three overlapping elements that develop a fact is not quite right.

      Firstly, it’s not a venn diagram, rather it is an argument for three grounds insufficiency of the fact as a means of truth.

      What do I mean by that?

      To say that context and trust are required to make a fact strikes me as incorrect.

      Suppose you notice a mark on your window, you look at it from many angles, and you decide, “yes, I wasn’t imagining it, there really is a mark on my window”.

      Now to say that this mark requires context, of your room, your state of mind, and so on, is false. You can create a proliferation of useless propositions that mark a huge number of details of your immediate surroundings, record them in your diary, and so on.

      These are facts.

      Now these facts might have a lifetime, they might be conditional on certain circumstances, often the circumstances that were present during the process of investigation, such that the proposition was stable, but this is a much more specific thing that just talking about context.

      Because after all, trust is the relational context, observation is the interactive context, and so on.

      Because events have history, and they exist in space, and because we have the capacity to create analogy or imagine strange combinations, you can always create context.

      Two events can become contextualised retroactively by us dreaming about them both at the same time, weeks later. Two people may have never met, but we relate them mentally.

      But the very generally of this version of context means that it means nothing. If we consider all the possible relationships a fact can have, to say that there is no objective way to separate something from the possible web of relationships is to, for example, say that there is no ordering of events in time, that an observation, the distinct fact, cannot exist without us subjectively distinguishing in the moment between “On the 17th of december 2011, John Smith of 33 hampstead lane, london, england, was 55 years old”, and the fact that two years from now, someone you do not know will be trying to remember how old they are. That is context, in the sense that it is a relationship between the things that can be observed and something else, but if there is no objective ordering before human subjective judgement, I will be deciding to exclude context of which I have no knowledge.

      So I will propose that actually that is false, someone does not always need to decide what context to omit, rather an observation, if it appears to us as a distinct fact, presents itself to us in singularity, with the process of observation suggesting to us that it has a certain set of relationships, with other relationships that could potentially be established not appearing relevant or integral to the fact in front of us.

      This is not a subjective decision, it is a part of the process that forms subjectivity in the sense of distinct perspectives. You have a domain of interactions that allows you to verify certain kinds of propositions, you are able to state the facts as you understand them. This may be a consequence of choices you have made, but rarely is a choice made to see a situation a certain way, as by way of contrast, most ambiguous optical illusions attest; (to use a Wittgensteinian example) in that situation, perception is made arbitrary by construction, as the optical illusion is crafted so as to resemble two different familiar objects, and we immediately experience a breaking down of the distinction between those objects, having our vision flip between the two, sometimes with effort, settling on one or the other, but presenting to us as something ambiguous.

      In other words, the duck/rabbit has a context presented to us, made intentionally to have structural similarities to both the image of a duck and a rabbit. The object implies relationships to us because of real measurable mappings that can be made between elements of the image and elements of either image, and indeed similar illusions have been constructed more recently by using software to interpolate between images, or to divide up high and low spatial frequency from different images and mix them etc. We experience these things as moments where we must choose how to percieve, as strange unstable moments where a material object has been constructed so as to present us with a choice born of its ambiguity.

      So if we can construct situations, material objects so that they reliably create ambiguous perception, can we do the reverse? And even if we cannot, shouldn’t the way that this object imposes upon us a certain experience along a scale from ambiguity to clarity suggest that clarity of an observation is not merely in the subject but in the object observed?

      The fact may appear to us suggesting relationships, but not with infinite relatedness, and sometimes extremely distinctly. And it is by exploring this domain of things that appear distinctly that we have entire fields of experimental science.

      So I would say that your account of how facts are constructed from context by subjective choice is just wrong; just as we have learned to make optical illusions we have learned to construct experiments, situations in which it is possible to varify and then record propositions up to certain bounds of accuracy.

      You can then go from there into the social contexts of the archivist, the cataloguer, the wikipedia editor, and so on, and the specific cultural practices that elevate the particular qualities of facts and their indexing, and arrangement in space and time. (My examples of the diary and the person at a specific moment and location were chosen carefully)

      There are certainly problems with such an approach, things they cannot see, things they don’t respond to etc. but these are not born of the inherent unrecognised subjective freedom of choice of creating environmental distinctions, but because of the specific tendencies developed practically while following the desire to create well corroborated and watertight depictions of specific events, the ways that people, their perception and their practices are shaped by the specifics of “the facts”, through how operations related on them have been found to be successful in the past.

      So if this is obviously wrong, this idea of subjective context, why does it appeal? I think it’s a rhetorical or cognitive trick to say that a concept already contains as a pre-requisite those things that it is in opposition to.

      So when talking about facts, you present it as being dependent on observation, and implicitly, opposed to both context and trust, in its idealised form, so that by exploring how it might be said to be based on the last two, you can undermine it conceptually; facts lead us out of the domain of contextual knowledge and trust, offering us the potential to just know a situation, as a stranger, better than those who are immersed in it. This kind of progressively desituating analysis, beginning with data precision and methodological clarity, and ending with facts, has the fact itself as one of the points of attack where this form of knowledge returns to a relational world of local people, and starts demanding change in the name of truth.

      To the extent that people recognise the validity of this particular species of fact, people are doomed to having to destabalise their views and life situations, just as is the purpose of the fact in experimental science; to provide the fly in the ointment to people’s existing grand explanations, and generally that which must be explained.

      So this is one reason why I think the blockchain comparison gets made; because although both facts and cryptocurrencies rely on trust in practice, the ideal to which the practices that create them are directing themselves is the elimination of trust to whatever extent can be currently practically achieved. So the question of observation could become an internal criticism of facts, along the lines of Hume (with a lot more other internal criticisms based on the specifics of fact creation), and trust and context can become external criticisms, of the forms of knowledge that are not immediately amenable to representation within the domain of facts, or of the interactions that result when this returns to a domain of experience formed from local patterns of trust and mutual understanding.

      Finally, I want to observe how disappointing “problem 3” is in your discussion, the subheading of epistemic degeneracy, is a fascinating thing to discuss, but instead you skip that and focus heavily on trying to provoke fear in the idea of facts, that facts violate sacred values of science and responsibility, and lead to conflict,threaten your audience etc.

      That might be true, but that’s boring! Oh no, your target of disapprobation will apparently attack the foundations of your audience’s sense of the world, how conveniently apocalyptic. Tell me about that “epistemic degeneracy” instead.

  10. I’m in for the “to be continued…” I get this but I’m left with nothing. Loved the corner metaphor.

  11. Enjoyable essay, but this seems to be the weakest argument: “Like fact-based crusades, fact-based censorship is already widespread on social media platforms.” Can you give an example? I’ve seen fact-checking additions to privately-owned social media channels – wouldn’t that merely be an expansion of free speech, then? Even deletions of comments from these platforms isn’t quite censorship, is it? If the state apparatus is not involved then the individual speaker may continue speak these thoughts freely and openly elsewhere in the public sphere. To require a private company to publish or disseminate information that it believes will harm it or its users seems to be a free speech sin of its own.

Leave a Comment


This site uses Akismet to reduce spam. Learn how your comment data is processed.