I once attended a banquet which the host insisted on calling ‘dinner’, and the entire night I kept thinking that he had confused two ideas which are really quite distinct.

Dungeon Crawler Carl is suffering from the same miscategorisation. Damage figures. Levels gained. Achievements unlocked. Sponsorship offers. Bulletins from an administrator who is losing patience with you. Eight volumes of it now, and I am told this is the appeal rather than the price of admission.

I should say at the outset that the book is not bad. It is, in fact, extremely good at the thing it is doing, but the thing it is doing is not what its readers keep calling it.

What it has actually done

Matt Dinniman began posting the story a chapter at a time on Royal Road, a website where serial fiction is published to an audience that answers back, in late 2019. The first volume went on sale in September 2020. By July of this year the first eight books had sold more than fourteen million copies between them, and the audiobooks, narrated by Jeff Hays, had been listened to for something over 140 million hours. Dinniman licensed the printed editions to Ace at Penguin Random House in 2024 and kept the digital rights himself, a very shrewd decision for a novelist to make.

Book eight, A Parade of Horribles, arrived in May. Peacock ordered a live-action series in June. A tabletop adaptation raised something in the region of ten million dollars this spring. This is not a cult. This is an industry, and it was built by a man who was, until the pandemic closed the cat shows he used to sell his artwork at, not a full-time novelist at all.

I should also concede, before anybody writes in, that this newspaper’s own chart has A Parade of Horribles at number five this fortnight. That chart is the word of the industry, and is precisely why the question of what the book is has become worth asking out loud.

The props are all fantasy. The laws are a rules set with an owner, an administrator and a complaints procedure. The laws win, and they always do.

Nobody agrees where the line is, so let us at least draw one

The usual way to settle a question of genre is to look at the shelf it is on. This will not do, and I can prove it with a single sentence written by the man who compiled the standard reference.

John Clute’s entry on Mervyn Peake in the Encyclopedia of Science Fiction observes of the Gormenghast books that “although couched in a language which might point towards Fantasy, it contains no fantasy elements,” and concludes that the sequence “must be thought of as sui generis.” Peake, in other words, wrote three novels about a vast castle, its hereditary rituals and the boy who inherits them, and there is not one spell, one dragon or one supernatural event in the whole of it. The most authoritative taxonomist in the field says it is not fantasy. Every bookshop in the world shelves it as fantasy, and every fantasy reader I know claims it without hesitation.

So we have a book with no magic that everyone treats as fantasy, and we have Dungeon Crawler Carl, which is made of nothing but magic, and whose membership is argued about constantly. Magic is therefore neither necessary nor sufficient. Shelving is not an argument. If the question is worth asking at all, we need a test.

The test

As it happens, this paper already proposed a test in our very own pages about a rather smaller world. A world belongs to the genre of its laws, not the genre of its props. Pokémon’s Kanto is stuffed to the rafters with machinery and governed entirely by the doctrine of correspondences, which is to say by alchemy, and so, by our reasoning, Kanto is fantasy wearing a laboratory coat.

Dungeon Crawler Carl is the case that runs the other way.

Look at the props. Dungeons, eighteen levels of them. Goblins, mana, loot, boss fights, enchanted equipment, a talking cat. Every object in the inventory could have been lifted from a fantasy novel of the 1980s, and a good deal of it was, knowingly and with affection.

Then, look at the laws. The dungeon is owned by a corporation. It is administered by an artificial intelligence that makes announcements. Its rules are published, amended and patched. Its difficulty is tuned. It sells advertising. There are sponsors, and the sponsors have preferences, and a crawler who is entertaining enough will be given better equipment for it. Every mystery in the book has an answer, and the answer is always the same answer: somebody decided this, for a reason, and the reason is money.

That is not a cosmos. That is a business with a customer-facing interface.

Tolkien named this, and named it precisely

The distinction I am reaching for is not mine, and the man who made it made it better.

In On Fairy-Stories Tolkien separates two things that ordinary usage runs together. Enchantment, he says, “produces a Secondary World into which both designer and spectator can enter, to the satisfaction of their senses while they are inside; but in its purity it is artistic in desire and purpose.” Magic, by contrast, “produces, or pretends to produce, an alteration in the Primary World.” One makes a place. The other operates a lever.

He went further in a letter of 1951 to his publisher’s reader, where he set out what his own legendarium was actually about. Magic there is bound up with what he calls the Machine: “our more obvious modern form though more closely related to Magic than is usually recognized,” by which he meant the use of apparatus to obtain power over the world rather than the development of anything within oneself. The Machine is a means. It has a manual, and anyone holding the manual can work it.

The system in Dungeon Crawler Carl is the Machine in its purest recorded form. It is apparatus, it is documented, it is patched when it misbehaves, and the entire pleasure of reading the book is the pleasure of watching a clever man read the documentation more carefully than the people who wrote it. That is a real and considerable pleasure. It is not enchantment. Enchantment requires that something in the world be older than you, larger than you and none of your business, and there is nothing in this dungeon that is none of Carl’s business, because the terms of service are appended to everything he meets.

Stating the other case at its strongest

In the name of fairness I will now give the defence, too, which is better than the objection usually credits.

Laura Miller made it in Slate in February, and she is not a critic to be waved at. Her argument is that the mechanics are not decoration but weaponry: the series is “a frothy, funny tribute to gaming” that ends up scrutinising “the cruelty at the root of so much humor and the perils of treating real life like a game.” The achievements are obscene because achievements are obscene when the experience points come off a corpse. The sponsorship is grotesque because the audience watching is real, and cheering, and buying. The exploitation of the dungeon’s permanent residents, the ones who live there and did not volunteer, becomes the moral spine of the later books. Dinniman is not naive about his own props. He is using them to make an argument about spectacle, and the argument lands because it is delivered in the grammar of the thing it is attacking.

I think that is all true, and I think it wins the case for the book’s quality outright. I do not think it wins the case for the genre. It loses it, in fact, and rather decisively, because an author who turns the rules into the villain has conceded that the rules have an author. A world you can file a complaint about is not a world. It is a venue.

Two juries have already ruled

Here the argument stops being mine, which is a relief, as I dislike carrying things.

This Inevitable Ruin took Best Science Fiction Novel at Dragon Con in 2025. In June of this year the first volume was shortlisted for the Arthur C. Clarke Award, which is the most serious prize in Britain for science fiction and is not given for anything else. It did not win; E.J. Swift’s When There Are Wolves Again took the fortieth-anniversary award on the twelfth of August. But it was there, on a list of six, judged by people whose entire function is to decide what science fiction is.

And Dinniman himself is admirably clear. Asked what LitRPG is, he declines the word genre altogether: “many people call it a genre, they’re not, technically, a genre, either.” It is a style, he says, defined by how a book is written rather than what it is about, and it “can be, and has been, applied to several different genres such as sci-fi, horror, western, and even romance.”

He is right, and his readers are the ones disagreeing with him. LitRPG is a manner of narration. It tells you what genre the prose is in. It does not tell you what genre the world is in, and the world here is a first-contact story about an alien entertainment conglomerate that demolished a planet for resources and spectacle. Strip the goblins out and you have Pohl and Kornbluth: a satire about advertising, conducted at the scale of a species.

On the word “constructed”

A friend of mine put down the first volume of Dungeon Crawler Carl having come to it directly from Titus Groan, and said the thing felt constructed. I have been turning that word over for a fortnight, because it is the right word and the obvious explanation for it is wrong.

The obvious explanation is that the book was planned to death. It was not. Dinniman says the opposite, and cheerfully: “it’s my favourite part of the process, making it up as I go along.” He keeps a spreadsheet of questions he has not yet answered and refuses to outline.

Peake did not outline either. So the difference is not planning. The difference is what the improvisation is answerable to.

Peake improvised inside a sensibility. What governs Gormenghast is atmosphere, ritual and an almost geological patience, and when Peake invented on the day he invented downward, into more castle. Dinniman improvises inside a rules engine, in public, a chapter at a time, in front of an audience that checks his arithmetic in the comments and will tell him within the hour if a power curve has gone soft. That is not a criticism of his character. It is a description of a different job. The work is answerable to a system, and the reader can feel the system being balanced, because balancing it is the plot.

A difficulty curve being tuned is a pleasing thing to watch and an unmistakable thing to read. It is why the book feels made rather than found, and fantasy, when it is any good, usually feels found.

Which explains the converts

The people evangelising this series are very often not fantasy readers. They will tell you so themselves, usually in the same sentence, and generally with the air of someone who has been let out of something. Miller’s piece opens with a sixty-eight-year-old non-gamer who was humouring her son and is now devoted.

The received explanation is that Dinniman is a gateway, and that these readers will proceed onwards to the deeper stuff. I doubt it, and I do not think they need to.

They love it because it is not fantasy. Fantasy asks something of a reader that a great many perfectly intelligent people simply do not wish to give, which is submission: you must consent to a world whose rules are other, trust that they exist, and accept that the book will not be explaining itself to you. Dungeon Crawler Carl asks for nothing of the kind. It shows you the rules, in a box, in bold, and then invites you to enjoy watching a man in a leather jacket, heart-print boxer shorts and a pair of borrowed pink clogs beat them. Nothing is withheld. There is no submission, and consequently there is no barrier.

Fantasy asks you to consent to a world you will never be shown the rules of. This book shows you the rules in a box, in bold. Of course it recruits the people who bounce off Peake.

That is not a lesser kind of reading. It is a different appetite, and it is being extremely well fed. I am told there are people who prefer a set menu with the prices printed beside each course, and having now met the argument in its strongest form, I am prepared to believe such people exist and to wish them well.

The verdict, and what it costs

So: not fantasy. Science fiction, and rather good science fiction, in a costume borrowed from the shelf next door.

It would be easy to read all that as a way of taking something away from the book, and it is meant as the reverse. Filed as fantasy, Dungeon Crawler Carl looks thin. Its world has no depth behind the interface, its magic has an owner, and its mysteries are all administrative. Judged against Peake or against Scott Lynch’s Camorr, a city with a history it never stops to explain to you, it comes off badly, and the reader who came from those books puts it down feeling obscurely swindled.

Filed correctly, it looks like what it is: a sustained, inventive, genuinely savage satire on the entertainment industry, built in the one grammar capable of carrying the joke, and sold in numbers that would embarrass most of the literary novelists of my acquaintance. The misfiling is doing the book far more harm than any critic has.

Put the thing on the right shelf and it stops being a disappointing fantasy novel and becomes an excellent science fiction one. That is the whole of my complaint, and I am confident it will not be heeded for another fourteen million copies at least.