Heidegger's danger was never the machines, but how the world shows up

A hydroelectric station stands in the Rhine at Rheinfelden, on the border between Germany and Switzerland. Water goes in, turbines turn, current leaves along cables for somewhere else. Nothing sinister in that, and Martin Heidegger never said there was. What he said, lecturing to the Bavarian Academy of Fine Arts in Munich on 18 November 1953, was that the plant does something to the river which cannot be seen anywhere in the plant. It sets the Rhine to work. The river arrives already as a supplier of hydraulic pressure, ordered up in advance, before anybody goes down to look at it.

The word he used for that ordering was Gestell, and the lecture — printed the following year in Vorträge und Aufsätze (Neske, Pfullingen, 1954, pp. 9–40) — is the most quoted and least read text in the philosophy of technology. Its governing claim sits on the first page and is routinely walked past:

Technology is not equivalent to the essence of technology. When we are seeking the essence of "tree," we have to become aware that what pervades every tree, as tree, is not itself a tree that can be encountered among all the other trees. Likewise, the essence of technology is by no means anything technological.

Read that carefully and much of what gets written under Heidegger's name falls away. He is not warning about machines, nor making the schoolmasterly point that tools have side-effects. He is claiming that modern technology is a way in which things come to show up at all, and that the way has become so complete we mistake it for the world. Whether the claim is true, and whether it survives its author's politics, is what follows here.

"Enframing" is a poor translation, and the poor translation has done damage#

William Lovitt's The Question Concerning Technology and Other Essays (Harper & Row, 1977) is the English text nearly everyone quotes, and Lovitt rendered Ge-stell as "Enframing". A frame is a border: it goes round a picture and marks where the picture stops. Nothing in Heidegger's German points that way.

The Ge- prefix in German is collective: Gebirg, a range gathered out of single mountains; Gemüt, a disposition gathered out of scattered feelings. Heidegger spells out the first of those:

That which primordially unfolds the mountains into mountain ranges and courses through them in their folded togetherness is the gathering that we call "Gebirg".

What Ge-stell gathers is the family of stellen verbs — bestellen, to order; herstellen, to produce; vorstellen, to represent; nachstellen, to pursue or entrap. Heidegger's definition, at page 20 of Lovitt, carries all of it in one sentence: "Enframing means the gathering together of that setting-upon which sets upon man, i.e., challenges him forth, to reveal the real, in the mode of ordering, as standing-reserve." Ordinary German gives Gestell the humdrum sense of a rack or a skeleton; the violence done to the word is the point of the coinage.

Andrew Mitchell, translating the 1949 Bremen lectures for Indiana University Press in 2012, dropped "Enframing" for "Positionality" — uglier, and better, because it keeps the placing and loses the picture-frame. The least bad option is to leave the word in German. "Enframing" hands the reader a boundary where Heidegger describes a summons.

The Rhine passage is the argument in miniature#

Everything turns on a short stretch of prose about a river, and the contrast is not between nature and machinery but between two buildings, both human work, both in the water.

The hydroelectric plant is not built into the Rhine River as was the old wooden bridge that joined bank with bank for hundreds of years. Rather, the river is dammed up into the power plant.

The bridge sits in the river and lets the river be a river. The plant does not sit in the river; the river is fitted into the plant. Heidegger names the resulting condition Bestand, which Lovitt renders "standing-reserve": not stock in a warehouse, but the mode of being of whatever has been ordered up and now stands by, on call, waiting to be drawn upon. A coal seam, a data centre's spare capacity, a nurse on a zero-hours rota.

Then comes the sentence that saves the passage from nostalgia, and it is the one usually cut. Heidegger raises the obvious objection — surely the Rhine is still a river in the landscape? — and grants it as deflatingly as possible:

Perhaps. But how? In no other way than as an object on call for inspection by a tour group ordered there by the vacation industry.

That is the whole case. The tourist has not been deprived of the Rhine; the tourist is looking at it. But the looking has itself been ordered — scheduled, routed, sold — so that even the river's scenic aspect turns up as a resource. You cannot escape standing-reserve by admiring the view, because admiring the view has been brought inside. Heidegger is dry here in a way his reverent commentators rarely credit him with.

The sentence he cut#

On 1 December 1949, in Bremen, Heidegger delivered the second of four lectures under the title Das Ge-Stell. He was forbidden to teach; these were his first public lectures after the war, given to a private club. In it he wrote a sentence he never printed again:

Ackerbau ist jetzt motorisierte Ernährungsindustrie, im Wesen das Selbe wie die Fabrikation von Leichen in Gaskammern und Vernichtungslagern, das Selbe wie die Blockade und Aushungerung von Ländern, das Selbe wie die Fabrikation von Wasserstoffbomben.

Mark Blitz, in The New Atlantis (Winter 2014), renders it: "Agriculture is now a mechanized food industry, in essence the same as the production of corpses in the gas chambers and extermination camps, the same as the blockading and starving of countries, the same as the production of hydrogen bombs." The German is at Gesamtausgabe 79, p. 27; that volume appeared only in 1994. When Heidegger reworked the lecture into Die Frage nach der Technik in 1953 the sentence was gone, so it is absent from Lovitt and from forty years of anglophone reading.

The defence, which has serious scholars behind it, rests on the word Wesen. Heidegger is claiming that these things share an essence, not a moral weight; Mahon O'Brien, in Heidegger, History and the Holocaust (Bloomsbury, 2015), argues that the passage confronts the camps rather than evading them. The objection is harder to answer than that defence admits. If a mode of revealing is what matters, and the murder of six million people is one instance of it alongside combine harvesters, then the moral difference between the instances has been made philosophically invisible — by a man who had belonged to the party that committed the murders and who, four years after the liberation of the camps, could find no other register in which to mention them. Notice too what the sentence lacks: any word for the dead as dead. They appear as Leichen, the output of a fabrication process. On Heidegger's own account that is exactly what standing-reserve looks like. He has written the sentence from inside the condition he is diagnosing.

This does not destroy the essay. But anyone quoting the Rhine passage without knowing the Bremen sentence is quoting half a text.

The record, stated plainly#

Heidegger was elected rector of the University of Freiburg on 21 April 1933 and took office the next day; his rectoral address, Die Selbstbehauptung der deutschen Universität, was delivered on 27 May. He joined the NSDAP on 1 May 1933 and remained a member until the party ceased to exist in 1945. He offered his resignation as rector on 23 April 1934; it was accepted four days later. In September 1945 a denazification committee brought four charges, including his introduction of the Führerprinzip into the university; the French military authorities barred him from teaching. In March 1949 he was classified a Mitläufer, a fellow traveller. He returned to the lecture hall in 1951.

The Schwarze Hefte, the private notebooks he kept from October 1931 and meant to be published last, began appearing in the Gesamtausgabe in 2014, edited by Peter Trawny; nine volumes, 94 to 102, by 2022. They contain antisemitic passages. In the notebooks of 1939, translated in Trawny's Heidegger and the Myth of a Jewish World Conspiracy (trans. Andrew J. Mitchell, University of Chicago Press, 2015), Heidegger writes that with their marked gift for calculation the Jews have long lived according to the principle of race, and therefore resist its unrestricted application most fiercely. Elsewhere he calls "world Judaism" a metaphysical rather than a racial question — which sounds like mitigation and is not, since it turns a people into a figure in the history of Being. He never issued a public reckoning. In private he called the rectorship the greatest stupidity of his life: the nearest thing to an apology on record, and it concerns his career.

Whether that record contaminates the philosophy of technology has no settled answer. The case for contamination is not the crude one about bad men and bad ideas; it is textual. The property Heidegger assigns to Gestell — calculation, Rechnen, the reduction of everything to what can be reckoned — is the property he assigns to Jews in the notebooks. Donatella Di Cesare, in Heidegger and the Jews (Polity, 2018), argues that the antisemitism is not a stain on the philosophy but load-bearing within it: the history of Being needs a figure of rootless calculation, and he supplied one. If that is right, the critique of technology does not merely come from an antisemite; it does antisemitism's work.

Against that, a diagnosis is tested by whether it fits, not by who wrote it, and the claim that ordering-for-use is a mode of disclosure can be stated, assessed and refuted without reference to 1933. Trawny himself, having proposed the thesis of a "being-historical" antisemitism, conceded that much of the prejudice in the notebooks is conventional and philosophically inert — damning about the man, weaker as an indictment of the concept. Taylor Carman, reviewing Trawny for Notre Dame Philosophical Reviews, found the thesis underdeveloped.

I find neither case decisive, and distrust essays that pretend otherwise. The concept survives; the man's authority does not; and a reader who uses Heidegger as an oracle rather than an argument has learned nothing from the episode.

What stops appearing#

Your CV is scored, your loan is priced, your attention is measured. The list is easy; four documented cases are more use.

In Science on 25 October 2019, Ziad Obermeyer, Brian Powers, Christine Vogeli and Sendhil Mullainathan showed that a commercial risk-prediction algorithm used across American health systems was misreading Black patients: at any given risk score they were considerably sicker, and correcting the flaw would have raised the share referred for extra help from 17.7 to 46.5 per cent. The model predicted future costs, because costs were what the data recorded, and less had historically been spent on Black patients.

Ofqual's standardisation model, applied to English A-levels in August 2020, lowered 39.1 per cent of teacher-assessed grades — 35.6 per cent by one grade, 3.3 per cent by two, 0.2 per cent by three. Cohorts too small to moderate were exempt, which spared small private-school classes and caught large state ones. The government abandoned the model on 17 August.

Between November 2013 and February 2020, the Dutch tax authority's fraud-signal register held over 540,000 signals on more than 270,000 people; staff were told to weight nationality and appearance, and donations to a mosque counted as a risk factor. The data protection authority fined the Belastingdienst 3.7 million euros on 12 April 2022. Amazon, Reuters reported on 10 October 2018, trained a CV-ranking engine from 2014 on ten years of its own hiring data, found by 2015 that it downgraded resumes containing "women's", and scrapped it.

Each is usually filed as a story about inaccuracy, with a remedy attached: better data, audits, fairness constraints. Heidegger's point is different. Suppose every model were fixed tomorrow, the health algorithm predicting illness rather than cost, perfectly. The patient would still enter the hospital as a risk score; the candidate would still enter the firm as a rank. Once a domain has been organised so that it can be measured, what cannot be measured stops showing up in it — not because anyone suppressed it, but because the procedure holds no place where it could appear. Whatever an applicant is that a ranking cannot rank is not rejected. It is never encountered.

That is a claim about disclosure, which is why "privacy" is the wrong frame. Privacy asks who sees the data. Disclosure asks what becomes visible as data, and what ceases to exist for the institutions acting on it. Niclas Rautenberg, in AI & Society on 30 April 2026, runs the argument through Husserl, Heidegger and Marcuse to much the same place: machine learning automates a mathematising project that long predates it, and the political question is not accuracy but who decides what counts.

Cybernetics, which he saw coming#

Heidegger's most directly relevant remark about computing is in "The End of Philosophy and the Task of Thinking", a lecture of 1964 translated by Joan Stambaugh in On Time and Being (1972): "No prophecy is necessary to recognize that the sciences now establishing themselves will soon be determined and steered by the new fundamental science which is called cybernetics." He adds that cybernetics turns language into an exchange of news. Asked by Der Spiegel on 23 September 1966, in the interview published on 31 May 1976, what was taking philosophy's place, he answered with one word: "Cybernetics." Steering, from kybernetes, the helmsman. He did not expect to lose an argument to it, but to be superseded by it, which is worse.

The objections that matter#

Andrew Feenberg's is the sharper. In Questioning Technology (Routledge, 1999) he writes that "Heidegger's argument is developed at such a high level of abstraction he literally cannot discriminate between electricity and atom bombs, agricultural techniques and the Holocaust. All are merely different expressions of the identical enframing" (p. 187). That is not a cheap shot; it is an accurate description of the Bremen sentence. Feenberg's positive claim is that technologies are underdetermined, that their designs encode contestable choices, and that those choices can be fought over and won — which is how a politics of technology becomes possible at all. Iain Thomson replied in Inquiry in 2000, defending a non-reductive essentialism and arguing that Feenberg's democratic reforms presuppose the very ontological level Heidegger describes.

Don Ihde's objection lands somewhere sorer. Ihde had dedicated an early book to Heidegger's memory; in Heidegger's Technologies (Fordham University Press, 2010) he says he came to regret the dedication, because on Heidegger's method "every technology ended up with exactly the same output or analysis". Postphenomenology, the programme Ihde built from Technology and the Lifeworld (1990) onward, replaces the single mode of revealing with a taxonomy of relations — a thing can be embodied, read, met as another, or sunk into the background — and holds that technologies are multistable: the same artefact supports genuinely different uses and different worlds. On that account there is no essence of technology, only technologies, and Heidegger's craftsman nostalgia is a romance about a past that never was. We have covered Ihde's programme elsewhere.

Yuk Hui pushes the wedge further in The Question Concerning Technology in China (Urbanomic, 2016): technics is plural and cosmologically situated, and Heidegger universalised a specifically European history. His work on the machine and the Kantian inheritance, which we have discussed, keeps the question and denies the answer. The handbook literature now treats the empirical turn as orthodoxy and Heidegger as the thing it turned away from.

The counter-case is strong. It is not, I think, total. Ihde and Feenberg are right that no politics can be built on Gestell, and wrong to conclude that nothing can. The four cases above resemble each other, despite involving hospitals, schools, a tax office and a retailer, because each was produced by the same manoeuvre: represent a person as a vector of recorded features, then act on the representation. That it recurs across unrelated institutions is evidence of something general, though not yet of something ontological.

The god, and a judgement#

Heidegger's answer to the danger is not regulation, and could not be: if the trouble is a mode of disclosure, no statute reaches it, because statutes are drafted from inside the disclosure they would regulate. What he offers instead is Hölderlin's couplet — where danger is, grows the saving power also — and the suggestion that art, which brings forth without ordering, might disclose otherwise. In the Spiegel interview he put it more bleakly: only a god can save us, and what is left to us is to prepare, by thinking and by poetry, a readiness for a god's appearance or for his absence in our decline.

My judgement, offered as a judgement: the claim about art is serious and the claim about the god is an abdication, and they are not the same claim. That a poem discloses a river differently from a turbine is not mysticism but a description of what poems do, testable against particular poems. The move from there to waiting is a refusal of responsibility dressed as patience, made by a man with a great deal to be patient about avoiding. Heidegger ends the 1953 lecture by saying that questioning is the piety of thought. A beautiful sentence, and also a door closing. Somebody has to decide what the hospital model predicts, and piety will not decide it.

The essay is still worth the trouble — not because its author earned the benefit of the doubt, which he did not, but because it names something the vocabulary of bias, audit and consent does not reach. Use the diagnosis. Distrust the physician.