Library of Babel: the website that holds every possible page

The Library of Babel is a website that already contains every page of text a person could ever write, along with a near infinite ocean of gibberish around each one. It is worth a look because it turns an old thought experiment into something you can actually click through, and because it quietly scrambles your sense of what it means for a text to exist.

In brief

• The site is an online version of Jorge Luis Borges' 1941 short story about a library holding every possible book, built in 2015 by the Brooklyn writer and coder Jonathan Basile.
• It can show any page of up to 3,200 characters drawn from a set of 29 symbols, which works out to roughly 10 to the power of 4,677 distinct pages.
• Nothing is stored in advance. An algorithm generates each page the moment it is asked for, yet always returns the same text to the same address.
• You can type any sentence into its search box and it will tell you exactly where in the library that sentence has always lived.

The story that started it

The idea is not new. It belongs to Borges, who published "The Library of Babel" in Spanish in 1941, in the collection El jardín de senderos que se bifurcan, later folded into Ficciones in 1944. The story imagines the universe as an endless library of hexagonal rooms, each identical, each lined with books.

Borges is precise about the architecture. According to the Wikipedia summary of the tale, every room holds four walls of bookshelves, and every book runs to four hundred and ten pages. Each page carries forty lines, and each line about eighty letters. The books use just twenty-five orthographic symbols: twenty-two letters, the period, the comma, and the space.

From those fixed limits one conclusion follows. If the library holds every possible arrangement of those symbols, then it must contain every book that has ever been written and every book that ever could be. It also contains an overwhelming majority of pure nonsense. The story's narrator spends a life wandering the stacks, hunting for a single meaningful line.

Borges does not treat this as a happy abundance. His narrator is one of countless librarians born and dying inside the stacks, and the tale is heavy with their despair. Some believe a single book somewhere explains all the others. Some throw themselves down the ventilation shafts. Some roam as purifiers, destroying volumes they judge to be worthless, which is nearly all of them. The story is a parable about meaning, and about how a library that holds everything can feel indistinguishable from one that holds nothing.

The man who built it

The site is not a corporate product or a studio's demo. It is the work of one person, Jonathan Basile, who approached it as a reader of Borges rather than as a game developer. Basile has pursued a doctorate in comparative literature at Emory University, and his interest in the library is as much philosophical as technical.

He went on to write a whole book about the project and its source, Tar for Mortar: The Library of Babel and the Dream of Totality, published by punctum books in 2018. The subtitle names his real subject. The library, in his reading, is less a gadget than a test of the old dream that everything worth knowing could be gathered in one place, and of what that dream quietly costs.

That framing matters when you use the site, because it explains why it refuses to be merely useful. Basile could have built a novelty text generator and stopped. Instead he built a careful argument in code, one that invites you to feel the gap between a library containing every answer and a reader who still cannot find a single one. The project doubles as literary criticism you operate with a mouse.

What the website actually does

Jonathan Basile wanted to know whether the library could leave the page. As recounted on Wikipedia, his first instinct was that someone must have built it already. When he searched and found nothing, he built it himself, launching the site in 2015.

The web version trims Borges' alphabet to a tidy 29 characters: the 26 English letters, the space, the comma, and the period. Every page in the digital library is exactly 3,200 characters long. That single constraint produces about 10 to the power of 4,677 possible pages, a figure with no useful comparison in the physical world. There are not that many atoms in the observable universe, by a margin too large to picture.

The pages are arranged to echo the story. The site sorts its contents into hexagons, each with four walls, twenty shelves per wall, and 640 volumes in total, every volume holding 410 pages. The naming scheme for the hexagons alone allows more than 10 to the power of 5,229 addresses. You browse by hex, wall, shelf, volume, and page, exactly as a visitor to Borges' universe would.

Nothing is stored, and everything already exists

The obvious question is where all of this lives. No server could hold 10 to the power of 4,677 files. The site's About page answers plainly: "We do not simply generate and store books as they are requested, in fact, the storage demands would make that impossible."

Instead the library runs a reversible algorithm. Give it a location, and it computes the text that belongs there. Give it a text, and it computes the location. Because the function is deterministic, the same address always returns the same page. The About page frames this as a kind of permanence: "any text you find in any location of the library will be in the same place in perpetuity."

This is the part that tends to unsettle visitors on first contact. The pages are not written when you arrive. They are not waiting on a disk. Yet they are not random either, because you can return to any one of them forever. The site prefers to say the texts are simply there, "only awaiting its discovery," a sentence that sounds mystical until you remember it is describing modular arithmetic.

Commenters on Hacker News landed on a cleaner description. The library behaves less like a warehouse and more like a very elaborate numbering system, where every conceivable page has been assigned a permanent coordinate and the website knows how to travel there on request.

The cleanest analogy is a cipher. A cipher turns a message into a scrambled string and, with the right key, turns it back again. Babel does something similar between pages and addresses. Every 3,200-character page corresponds to exactly one location, and the algorithm converts freely in both directions. Because the conversion is fixed and reversible, the library is stable without being stored, in the same way that the billionth digit of pi is a definite value nobody needed to write down in advance.

Searching for sense in the noise

A library of everything is useless without a way in, and the search box is where the site earns its reputation. Type a phrase, and the library finds a page that contains it, then hands you the exact hexagon, wall, shelf, volume, and page where that phrase sits. The result feels like a trick, because the phrase you invented a second ago is reported as having always been filed in slot such and such.

There are two flavours of search. One finds a page containing your text surrounded by random filler. The other finds a page of nothing but your text, padded only with spaces. Either way, the library insists it did not create the page for you. It merely looked up an address that was always occupied.

To help readers spot meaning inside the murk, the site offers a feature it calls Anglishize, described on Wikipedia as a tool that highlights recognizable English words scattered through otherwise senseless pages. It is a small mercy. Most pages in the library read like a cat walked across a keyboard for several minutes, because statistically that is what nearly all of them are.

The search also raises a strange legal puzzle, which visitors enjoy worrying at. If the library contains every text up to 3,200 characters, then it contains passages from copyrighted works, future works, and private messages not yet written. Hacker News threads have circled the question of who, if anyone, could claim to own a page that an algorithm reconstructs on demand and never stores. No lawsuit has settled it, and the site treats the paradox as part of the fun rather than a problem to solve.

Beyond text: the image archive

Having exhausted language, Basile moved on to pictures. A companion project, the Babel Image Archives, applies the same logic to a grid of pixels. It permutes 4,096 colours across an image measuring 416 by 640 pixels, which yields on the order of 10 to the power of 961,755 possible images.

That archive contains, by the same reasoning as the text library, every photograph that has ever been taken and every one that ever will be. Somewhere in it is a perfect portrait of your face, a frame from a film not yet shot, and a staggering quantity of coloured static. As with the books, the images are generated from coordinates rather than stored, so the archive occupies almost no space while claiming to hold almost everything.

Basile has framed the image archive as pushing the same idea past the comfort of words. A library of all texts at least sounds noble, like a grand reference shelf waiting for a diligent reader. A grid holding every possible image, nearly all of it meaningless colour, is far harder to romanticise, which may be exactly the point. It strips the last shred of dignity from the fantasy of owning everything at once.

Why people keep coming back

The library is not a game, and it has no score, yet people return to it the way they return to a view. Part of the appeal is the vertigo. Holding a page you know is unique, while also knowing it was never written by anyone, is a sensation few other websites can offer.

The reception has leaned into that strangeness. Wikipedia records the journalist Jerry Adler calling the site "the most fascinatingly useless invention in history," and the scholar Zac Zimmer describing Basile's version as "perhaps the most absolutely dehumanizing of all Library visualizations." Both are compliments, in the way that the better corners of the odd web collect compliments.

There is also a quieter reason it resonates now. The modern internet floods its readers with more text than anyone can process, most of it forgettable and some of it invented on the spot by machines. The Library of Babel is that condition taken to its logical extreme and then made calm. Every possible sentence is here, which means no single sentence is special, which oddly makes the search for meaning feel more human rather than less.

For a gentler take on the same problem, the Boring Websites network keeps The Bureau of Very Slow News, a site that reports events only once they are old enough to be certain. Where Babel offers infinite noise and dares you to find the signal, the Bureau offers almost no news at all and asks you to wait. Both are arguments, in their way, about how much information a person actually needs.

How to visit without getting lost

The site rewards a short, deliberate wander rather than a long one. Start from the search box, feed it a sentence you know well, and watch the library produce the coordinates. Then click into the surrounding pages and see how quickly sense dissolves back into static. That contrast, between the line you typed and the chaos one page over, is the whole point in miniature.

From there, try the random browse. It drops you into an arbitrary volume, almost guaranteed to be nonsense, which is a useful reminder of the odds the story's narrator faced. The library is complete, but completeness is not the same as helpfulness. The books that matter are hidden in a haystack with no straw to spare.

For the full effect, treat it as an experiment you can repeat. Write a line, search it, copy the hexagon address the library returns, then close the tab and reopen the site tomorrow and navigate straight back to that address. The same line will be sitting there, in the same place, untouched. Nothing was saved, and yet nothing moved.

What it is good for, honestly

The library is not practical, and pretending otherwise would miss the joke. You cannot use it to write for you, because locating a specific meaningful page is as hard as writing that page yourself. The search works only because you already supply the words. The library then performs its trick of telling you where your own sentence was always hiding.

What it is good for is perspective. Spend a few minutes there and the ordinary act of writing a sentence looks different. Every arrangement of letters you might pick already exists in the stacks, waiting, which means the worth of your sentence was never in its novelty as a string of characters. It was in the choosing, the act of pulling one page out of an ocean of near identical pages and deciding it meant something.

That is a surprisingly cheerful conclusion for a site built on futility. The narrator of the story despaired because meaning seemed impossible to find among the noise. The website quietly suggests the reverse. Meaning was never sitting in the library at all. It was in the reader who went looking.

FAQ

Is the Library of Babel free to use?

Yes. The site at libraryofbabel.info is open to anyone with a browser, with no account or payment required. It is a personal project by Jonathan Basile rather than a commercial product.

Does it really store every possible page?

No, and that is the clever part. As the About page explains, storing that many pages is impossible. The library generates each page from its coordinates on demand using a reversible algorithm, so the same address always returns the same text without anything being saved in advance.

How is it related to the Borges story?

The website is a direct adaptation of Borges' 1941 tale "The Library of Babel," which imagined a universe-sized library of every possible book. Basile kept the hexagonal structure and the premise, then swapped the story's 25 symbols for a 29 character set and capped each page at 3,200 characters.

Can I find a specific sentence in it?

Yes. The search box accepts any text within the character limit and returns the exact hexagon, wall, shelf, volume, and page where that text resides. The library treats your sentence as having always been filed there, rather than as something it just wrote.