The Same Machine

Every job I have been paid for comes down to the same operation. Something takes eleven steps; make it take two. Something needs a person awake at three in the morning; make it not need one. I run servers that other people's businesses sit on, I write software, and for the past few years I have written a fair amount of it alongside a model that finishes my sentences. There is nothing unusual in that. It is most of my industry, and my industry is a machine for removing friction.

So when I say I have spent a while with the people who think this is how you ruin a civilisation, take it with the appropriate discount. I have a stake in the answer, and it is on the supply side.

The argument I want to deal with goes like this: technical progress keeps producing new ways to sin, therefore progress is not neutral, therefore it is in some deep sense an evil. I have heard it from religious friends in a soft version and read it in a hard one. I think the conclusion is wrong and the accounting behind it is not close. I also think that somewhere inside it there is an observation my own side has never answered, and that our failure to answer it is not an oversight. It is a missing organ.

The version worth arguing with

First I have to take away the easy target, because beating it proves nothing.

"Christianity holds that technology is evil" is not a thing Christianity holds. The tradition's dominant instinct runs the other way, and it runs that way early. Augustine puts the first sin in the will and nowhere else: in City of God XIV.13 our first parents "fell into open disobedience because already they were secretly corrupted; for the evil act had never been done had not an evil will preceded it." The corruption is upstream of the act, and the act is upstream of any object. He is arguing against people who wanted evil to be a substance you could point at.

The same Augustine, eight books later, writes one of the more startled passages about human capability in late antiquity. In XXII.24 he exclaims at the "stupefying" advances of human industry "in the arts of weaving and building, of agriculture and navigation," then keeps going through pottery, painting, sculpture, medicine, writing, music. He is not a booster. He lists poisons and weapons in the same breath, and he caps the whole inventory with a line that should be quoted alongside it: all of these are "but the solace of the wretched and condemned, not the rewards of the blessed." His point is that even the consolation prize is this astonishing. Still, this is not a man who thinks the loom is the problem.

Aquinas tidies it up. Art, meaning craft, is "the right reason about certain works to be made," and its excellence sits in the object, not in the maker's soul: a craftsman "is commendable, not for the will with which he does a work, but for the quality of the work." The moral question is relocated to the will, which is where Augustine had already put it. The Catechism keeps the anti-Manichaean guard rail explicit, describing human nature after the fall as wounded and "inclined to sin" while insisting it "has not been totally corrupted."

Modern institutional Christianity is, if anything, more emphatic. Laudato si' is filed in most people's memory as an anti-technology document, and its actual chapter on the subject says that it is right to rejoice in technological advance, that "technology has remedied countless evils which used to harm and limit human beings," and asks who could deny the beauty of an aircraft. Francis names his target: not technology but a "technocratic paradigm," a single one-dimensional method that gets applied to domains where it does not belong. In 2024 he stood in front of the G7 at Borgo Egnazia and told the assembled heads of government that artificial intelligence "arises precisely from the use of this God-given creative potential." His successor took the name Leo mainly because Leo XIII had addressed the social question of the first industrial revolution, and AI looked like the second. The Vatican's 2025 note on artificial intelligence runs to 117 paragraphs, files science and technology under the "collaboration of man and woman with God in perfecting the visible creation," and reserves its hard line for autonomous weapons. Nearly seventy American evangelical leaders signed a statement in 2019 affirming that AI development demonstrates "the unique creative abilities of human beings."

None of that is a religion telling you to put down the tools.

The serious critique is elsewhere, and it is not primarily about sin at all. Jacques Ellul, a Reformed layman and law professor at Bordeaux, defined his subject as method rather than machinery: technique is "the totality of methods, rationally arrived at and having absolute efficiency (for a given stage of development) in every field of human activity." His claim was that this drive becomes self-augmenting and stops being something you choose. He is regularly called a determinist and regularly denied it; Samuel Matlack argues in The New Atlantis that the sweeping language is deliberately hyperbolic, meant to expose an unchecked fascination, not to deny that anyone can act. The Stanford Encyclopedia is politely brutal about the limits of the genre, noting that Ellul "was not so much interested in investigating this relationship as in emphasizing and denouncing the cultural and moral consequences as he saw them."

Neil Postman put the usable part in a sentence: "Technology giveth and technology taketh away," with the advantages and disadvantages never distributed evenly, and change that is "not additive; it is ecological." That is a claim you can actually test.

What happens when you make something cheaper

So test it.

In 2018 the American Supreme Court struck down the federal statute that had kept states from authorising sports betting. Thirty-eight states legalised some form of it, and by 2023 nearly $300 billion had been wagered through the new markets, most of it through online channels. This is close to a natural experiment on friction, because the underlying appetite did not change in 2018. Only the number of steps between the appetite and the bet changed.

The results are consistent, though the authors' own word for the damage is "modest." Hollenbeck, Larsen and Proserpio, working with credit-bureau records for roughly seven million consumers, find that legalisation cuts average credit scores by about 0.8 points, and that where online and mobile access arrives the fall is 2.75 points, alongside rising bankruptcy, collections and auto-loan delinquency. Note which variable does the work. The phone, not the legality.

A separate team at the NBER, using household transaction data, finds that betting does not displace other gambling or consumption. It comes out of savings, with risky bets crowding out positive-expected-value investment, and the damage concentrates in financially constrained households where credit-card debt rises and overdrafts multiply. Diagnosed gambling disorder in legalising states rose from 3.0 to 4.8 per 100,000 adults with a qualifying clinical visit between 2018 and 2026, while falling in states that did not legalise. That last figure needs a warning label, since it counts diagnoses and not prevalence and legalisation itself drives screening, but the direction is consistent with everything else.

Here is the sentence the religious critic wants me to write next: technology multiplies sin. I am not going to write it, because the same evidence base refuses it.

Take the identical mechanism and point it somewhere else. Madrian and Shea studied what happened when one large employer switched its retirement plan from opt-in to automatic enrolment. Participation among new hires rose sharply, and a large share of the automatically enrolled stayed at the default contribution rate and the default fund, a combination almost none of the previous cohort had chosen for themselves. It is among the most replicated findings in behavioural economics. A later study using the US Army's automatic enrolment of civilian hires found that four years on it had caused no significant change in credit scores and no increase in debt, which kills the obvious objection that people were simply borrowing to fund the saving. Johnson and Goldstein's organ-donation work found consent at 42 per cent under an opt-in default and 82 per cent under opt-out in a randomised online experiment, where the act consented to was hypothetical.

Nobody who was auto-enrolled decided to be prudent. They filled in a form, or rather they failed to fill one in. The step that used to sit between a person and a sensible decision was deleted by the same engineering that deleted the step between a person and a bad one. Richard Thaler even named the deliberate reverse operation, sludge: friction added on purpose to stop you.

The machine does not know which way it is pointed. Reduce the cost of an action and the action becomes more frequent. That is the whole law, and it is morally blind.

The residue

If friction removal works on everything, then the moral direction of a technological civilisation is set by what we choose to point it at, and we do not choose evenly. Appetite has customers. Somebody makes money every time a bet is placed, and that somebody will pay engineers to shave milliseconds off the path between a bored man on a sofa and a losing wager. Prudence has no equivalent sales force. Nobody sells you automatic enrolment. Where it exists it exists because an employer or a legislature decided it should, and no one's revenue depends on your taking it up.

That asymmetry is one residue, and it sits in the incentives, which are ours.

And there is a second residue, harder to dismiss, which is the part of the religious complaint I cannot make go away. When you deliver the outcome of a virtue by deleting the choice, you get the outcome and not the virtue. The auto-enrolled saver is not thrifty. He is enrolled. He has acquired no capacity he did not have before, and if the default is switched off he will revert, because nothing in him changed. A civilisation can run for a long time on good defaults and unformed people. Whether that is a triumph or a hollowing out is a disagreement about what a person is for, and I do not think either side gets to win it by citing a study.

My own answer is that I will take the outcomes, and I will take them without much agonising. But I want to be precise about what I am taking them instead of.

Half of all children

Start where it hurts. For most of human history, around one child in two died before the end of puberty. Not in a bad year. As the baseline. By 2020 the global under-five mortality rate was 4.3 per cent. Global average life expectancy was 32 years in 1900 and 73 in 2023, and before anyone reaches for the familiar correction, Our World in Data addresses it directly: it is a misconception that the gain came only from falling child mortality. Survival improved at every age. Fifteen-year-olds in France in 1816 could expect to reach 58; by 2021 the figure was 83.

Smallpox killed an estimated 300 million people in the twentieth century on Donald Henderson's widely cited reckoning, a figure nobody counted and everybody arrives at by applying case-fatality rates to assumed incidence. The last natural infection was recorded in Somalia in October 1977, and the World Health Assembly certified eradication in May 1980. There is a detail in that story I find difficult to shake. The final smallpox death was not caused by the disease escaping our reach but by our reach escaping us: in 1978 the virus escaped a laboratory at the University of Birmingham and killed Janet Parker, a medical photographer. Henry Bedson, who ran the department, took his own life five days before she died. Even the triumph has a technological body count, and it belongs in the total.

Vaccination through the Expanded Programme on Immunization has averted an estimated 154 million deaths since 1974, 146 million of them children under five, with measles alone accounting for more than 60 per cent. In 1800, in Sweden and Finland, the only two countries whose records are good enough to measure, about 900 mothers died for every 100,000 live births. The global figure in 2023 was 197.

Some of these numbers deserve to be handled carefully. The pre-1900 figures are reconstructions, not measurements. The extreme-poverty series is genuinely contested: Jason Hickel argues that the threshold is set too low to mean anything and that the pre-1981 and post-1981 data measure different things and should not be spliced into one line, and while I think the direction survives his objection, the objection is not silly. The twentieth century's worst famines were political products, not agricultural ones, which complicates any story where technology defeats hunger; Our World in Data says as much, describing famine as increasingly attributable to political causes including non-democratic government and conflict.

Adjust for all of it. The shape does not move. Any argument that technical progress is on balance an evil has to explain why it was worth burying half of all children, and it does not have an answer, because there isn't one.

The trial that everyone remembers wrong

There is a set-piece my side likes to tell here, and going back to the primary documents cost me the version I had been carrying for years.

In November 1847 James Young Simpson introduced chloroform for childbirth in Edinburgh, and shortly afterwards published a pamphlet with the unambiguous title Answer to the Religious Objections Advanced Against the Employment of Anaesthetic Agents in Midwifery and Surgery. The objection he was answering came from Genesis 3:16, where the curse laid on Eve is that in sorrow she will bring forth children. His reply worked on two levels. Philologically, he argued that the Hebrew nouns translated "sorrow" denote toil and effort, not bodily agony, and that where scripture means the pain of labour it reaches for different words. Scripturally, he pointed at Genesis 2:21, God casting Adam into a deep sleep before taking the rib, which he called the first surgical operation ever performed on man, with Calvin's commentary in support: Adam was sunk into a profound sleep so that he might feel no pain.

That much I had right. Here is what I had wrong.

Read the pamphlet and you find that Simpson names not one clergyman as an opponent of anaesthesia. The people objecting are patients, "not a few medical men," and an unnamed rival medical school where his conduct had been "publicly denounced ex cathedra." The clergy who appear in his pages are mostly on his side. Simpson has it at one remove, from his colleague James Miller: when Miller told Thomas Chalmers, the most eminent churchman in Scotland, that people were objecting on grounds of the primeval curse, Chalmers first said he saw no theological question in it at all, and then, on being pressed, remarked that if some "small theologians" really took that ground, Miller should not heed them. The story's shape is the reverse of the one I knew. Chalmers had to be persuaded that anybody was making the religious argument.

And the one dated, quotable objection from 1847 that actually invokes Genesis 3:16? Simpson reproduces it in a footnote. It was printed in the Edinburgh Medical and Surgical Journal for July that year, in an anonymous article on the injurious effects of ether: "In the lying-in chamber, nothing is more true than this; pain is the mother's safety, its absence her destruction. Yet, there are those bold enough to administer the vapour of Ether, even at this critical juncture, forgetting it has been ordered, that 'in sorrow shall she bring forth.'"

That is a medical journal. The scripture was coming from inside the profession.

The historians got there long before I did. A. D. Farr, reviewing the episode in Anaesthesia in 1980, concluded that opposition rested mainly on the belief that pain served a physiological purpose, and that "religious opposition is no more than a myth of historiography." He made the case at length for historians three years later in Annals of Science. The claim is now standard enough to have its own chapter in Ronald Numbers's collection of science-and-religion myths. It is not unanimous, and I should say so: A. G. McKenzie, working from an Edinburgh maternity case-book, found a marked rise in obstetric anaesthesia in the eleven months after the pamphlet compared with the eleven before, which he reads as evidence that there was real opposition and that Simpson overcame it. But even McKenzie's own summary is that the topic "was gradually embellished in historical articles."

We know who did most of the embellishing. Andrew Dickson White, in his 1896 History of the Warfare of Science with Theology, wrote that "from pulpit after pulpit Simpson's use of chloroform was denounced as impious and contrary to Holy Writ." He names no preacher, no church, no sermon, no date. He also converts Simpson's written argument into a spoken one, giving him a quotation no one has traced to him, beginning "My opponents forget the twenty-first verse of the second chapter of Genesis." The idea was Simpson's. The scene was White's.

I have told the sloppy version of this story at dinner tables. It was a good story, it flattered me, and its most vivid details were manufactured in 1896 by a man with a thesis.

One press, two books

The dual-use point is usually made with the printing press, and the press repays the attention because it is where the argument gets hard. Within fifty years of Gutenberg, European presses had produced more than 28,000 separate editions that still survive in at least one copy. Luther understood the machine better than his opponents did; between 1521 and 1525, on Andrew Pettegree's count, he and his supporters outpublished them nine to one, and his 1518 sermon on indulgences went through two Wittenberg editions, four reprints in Leipzig and two each in Nuremberg, Augsburg and Basel inside a single year. A Christian reader will call that a good outcome.

The same machine printed the Malleus Maleficarum, Heinrich Kramer's witch-hunting manual, first published at Speyer in 1487. A 2024 study in Theory and Society examined trials and manuals across 553 Central European cities between 1400 and 1679 and found that the publication of each new edition was followed by an increase in witch trials, spreading along networks between neighbouring cities. The hunts ran for nearly three centuries and produced roughly 90,000 prosecutions and close to 45,000 executions in that region.

So far this is the standard lesson: neutral tool, two uses, draw your own conclusion.

Keep going, because the second half is the part that decides the argument. The case against the trials also came off a press. That half of the story is intellectual history, with nothing like the 553-city measurement standing behind it. In 1631 a German Jesuit, Friedrich von Spee, published the Cautio Criminalis anonymously, and his case against witch hunting was not a claim about what devils can do. It was procedural: under the methods in use, no accused person could possibly be cleared. Johann Weyer had printed his own attack at Basel in 1563. Christian Thomasius condemned the trials in his Theses de crimine magiae in 1701 and died in 1728, the year the last person charged with witchcraft was executed in Prussia.

The antibody travels on the same rails as the infection, and it arrives later. That gap is where all the bodies are. But there is no configuration of the world in which you get Spee without also getting Kramer, because they are the same invention.

One more thing came off that press, and it is aimed at me, not at the believers. Ask most educated people how many witches were killed and a good number will say nine million. That figure was manufactured in the mid-1780s by Gottfried Christian Voigt, a municipal attorney in Quedlinburg, who counted thirty trials in his own town, derived a rate, and extrapolated it across 71 million Europeans over 650 years. He published it in the Berlinische Monatsschrift, a journal for readers who thought of themselves as the Enlightenment. It overstates the executions by a factor of roughly two hundred.

The number that anti-clerical Europe has used for two centuries to indict religion was invented by the Enlightenment, in an Enlightenment journal, using arithmetic that would not survive a first-year seminar. My side is not the side that checks. My side is the side that has better press.

The people who actually decide

The one group in the Western world with an answer to the question I cannot answer is the Amish, and most of what I thought I knew about how they decide turns out to be wrong. I had believed, and said out loud, that the Amish run controlled trials: one household takes a new technology, the community watches what it does to them, and then everyone votes. I cannot support that. Nothing in the scholarship I could reach documents a designed experiment with a fixed trial period. What Erik Wesner, who writes the Amish America site, describes is messier and, on reflection, more interesting. New technologies drift in at the edges, get taken up by more progressive members, and eventually the district has to rule on them; the question is typically raised by the bishop and put to a congregational vote. The Ordnung, the body of rules, is mostly unwritten and passed on by practice, and members reaffirm their own district's version twice a year, before the spring and autumn communion services. Breaking the vow you made at baptism carries church discipline and possibly excommunication.

The criterion is not that machines are wicked. The Young Center states it plainly: "The Amish do not consider technology evil in itself but they believe that technology, if left untamed, will undermine worthy traditions and accelerate assimilation into the surrounding society." The objection to the car is social geometry rather than mechanics, that greater mobility would pull the community apart. Which is why the rules look arbitrary until you see what they are protecting. Televisions, radios and personal computers are rejected outright; batteries are fine, and run buggy lights, calculators, fans, cash registers, copy machines and typewriters. Solar panels charge them. Most groups forbid drawing electricity from public utility lines, which is a rule about dependence, not about electrons. The telephone is not banned so much as evicted: many districts use a shared shanty by the roadside, a phone that exists but does not live in your house. Businesses run on diesel generators, propane refrigerators, hydraulics and a stripped-down word processor called the Classic, which is built by Old Order Mennonites.

Whatever else this is, it is a civilisation with a procedure.

I am not recommending it, and I am not going to sentimentalise it. Wisconsin v. Yoder in 1972 established that Amish children need not be compelled into formal schooling beyond the eighth grade, which is a real cost borne by people who did not vote for it. The founder effect has left a documented and ongoing burden of genetic disease, serious enough that the NIH maintains a chapter listing the variants. The population, an estimated 410,955 across North America in 2025, roughly doubles every twenty years, so whatever they are doing is not a failure at continuity, but continuity is not the only thing worth measuring.

The point is narrower than admiration. They have a mechanism for saying no. It costs them to use it and they use it anyway.

We have nothing. There is no room I can walk into, no vote, no twice-yearly meeting at which anyone asks whether the last thing we shipped left people better off. I have deployed systems that made a slow thing fast for tens of thousands of people, and not once has anyone required me to say what the speed was for. The question does not exist in the profession. We have an accelerator and a marketing department and no brake, and we have got so used to this that the absence does not read as an absence. It reads as freedom.

This is what my side won, and I think we won it too completely. Two centuries of arguing against people who wanted to ban the loom, the anaesthetic, the novel, the comic book and the video game left us with a reflex: every objection is superstition, every objector is a small theologian, and the answer to a bad technology is always more technology. Usually that reflex is right. It has been right so often that we stopped keeping the machinery for the cases where it isn't, and now we could not deliberate if we wanted to, because we never built the thing you deliberate with.

So: no, progress is not evil, and the people who say it is have to walk past a very large pile of children's graves to say so. The mechanism they blame is not moral at all. It removes friction from whatever it touches, and it will happily remove it from your pension and your bet with the same indifference.

But they have noticed that a machine with no capacity for refusal is running the century, and they are not wrong about that, and I help build it. When someone finally asks me to justify a thing I made, the honest answer will be that nobody ever asked, and that I never volunteered.