A French rail passenger in the 1980s could buy a ticket by typing 3615 SNCF into a small terminal the phone company had handed out for nothing. Codes like that were printed in magazines and shown in TV adverts, and according to IEEE Spectrum they were plastered on the sides of buses, which is where web addresses ended up too. Minitel is the first thing I picture when I try to imagine a world without the Internet, because I don't think that world would have been offline at all. It would have had screens and modems and online shopping. What it wouldn't have had is one open network joining everything to everything else.

Eric Raymond sketched the likeliest version in a 2011 essay, World Without Web, and it's grim reading. Without internetworking, he argued, you'd get walled gardens run by CompuServe, AOL and Genie, each with its own software and its own addressing. Email would only reach people on the same provider. Heavy users would juggle half a dozen accounts, and there'd be no hyperlinks between services and certainly no search engines. He allows for gateways between providers, but pictures them buggy, often one-way, and shut off whenever a big provider decided a smaller one was leeching. I think he's right about the shape. Computers would still have got faster and cheaper, since that was happening regardless, and people were already paying to go online long before most of them had heard of the Internet: CompuServe launched its consumer service in September 1979. The machines were never the missing piece. What was missing was a common language between them, and a world that never agreed on one wouldn't feel like the past so much as a present where every door takes a different key.

The people cutting those keys would matter enormously. Newspapers, broadcasters, publishers and a handful of large companies already decided most of what reached the public, and a network of walled gardens would have kept them in charge instead of letting anyone route around them. The Atlantic's history of Minitel has the clearest picture I've found of how that works. It describes 101 Online, an American videotex venture where anyone who wanted to publish had to take a floppy disk to the company's office in downtown San Francisco and hand it to an operator, who converted it to the house format and uploaded it. Its own head of marketing later admitted, "we did not create an ecosystem enabling anyone but us to make money." Minitel itself was the exception, and it's a real problem for my argument. Service providers there ran their own machines while France Télécom oversaw only the network, which is much closer to how the Internet works, and a world full of Minitels would be far less bleak than Raymond's. Getting one took a government willing to give terminals away, though, which UnHerd credits with clearing the cost hurdle that kept Britain's Prestel from ever catching on. Even Minitel never really travelled beyond France, partly because most of what was on it was in French. The likelier result is one network per country and a scattering of companies, none of them properly joined up.

This blog wouldn't exist in that world. There'd be no WordPress, as Raymond points out, and whatever public-diary format a provider offered would sit under its acceptable-use policy, readable only by its own customers. He goes further: post something the provider considers offensive and it shuts you down to protect its reputation. I'd be writing letters to magazine editors and waiting to see whether they ran them, or photocopying a fanzine and posting it to forty people. Publishing without asking permission is the thing I'd miss most, more than streaming or maps, and it's also the thing the people in charge would have been happiest never to hand over. People would still find each other, the way the French piled into Minitel's chat services, but mostly inside one provider at a time. A support group for a rare illness needs every patient it can reach, and a campaign needs the strangers on the rival network as well, so both would stay smaller than they deserve.

Those gatekeepers were perfectly capable of spreading nonsense themselves. The famous panic over Orson Welles's War of the Worlds broadcast in 1938 was far smaller than the newspapers claimed. As a BBC piece on the myth puts it, they made sweeping claims about panic-stricken Americans and offered little to back them up, and National Geographic notes that the print press regarded radio as an upstart rival at the time. Nearly ninety years later the myth still needs debunking. So I don't think a world without the Internet would be free of misinformation. The grassroots kind would slow right down, because a pile-on needs thousands of strangers to arrive in the same place within the hour and nobody has built the road they'd arrive on, and a rumour moving by photocopier or phone tree mostly runs out of energy on the way. The lies with a printing press or a transmitter behind them would move as fast as ever, and far fewer people would be in a position to answer back.

Advertising is where I can't tell which way things would go. Minitel anonymised connections at the gateway and asked for no usernames, so chat services never knew who their customers were. Billing went through the phone bill, with the phone company passing a share to service providers, and IEEE Spectrum credits that arrangement with keeping Minitel relatively free of advertising. Nobody needed to harvest your attention when you were already paying for your time. Prestel put a price on individual pages and, UnHerd says, never got past 90,000 subscribers. Micronet's multiplayer game Shades ran over it at 1.6p a minute, off-peak, a detail from The Escapist's history that goes a long way to explaining that number. Raymond imagines the opposite of all this, provider software shoving whatever it likes at you with no other browser to switch to and no blocker to hide behind. Both seem plausible. Tracking would probably still happen, since a provider knows exactly who you are, but nobody could follow you from one provider to the next, and an online celebrity would be famous on AOL and unknown on CompuServe. The influencer economy needs everyone watching the same few feeds, and that audience would never exist here.

AI would arrive later and look different. A big provider could train something on its own members' messages, and one probably would, but the large models of the last few years learned from text gathered across the whole open web, far more than any single provider holds, and they run in data centres paid for by companies that grew rich on the Internet. In Raymond's version the boom and the big fibre build-out never happen, so the text stays split between rivals and the money arrives much more slowly. Minitel did have a faint version of the dream. IEEE Spectrum mentions services called Claire and Sophie that gave out personalised information through natural-language interfaces, decades before anyone asked Siri anything.

Everything physical would carry more weight: record shops, libraries, the local paper, the catalogue on the kitchen table. Mail-order books like the La Redoute catalogue from 1994, with a model's thanks set in tiny type beside the product codes, would just keep getting thicker. Regional culture would stay stronger too. A band could build a following in its own city for years before anyone in London noticed, and a scene in one place wouldn't become a look on the other side of the world within a week. A country you'd never visited would be whatever the evening news chose to show you, and a friend abroad would be a letter every few weeks unless you both happened to pay the same provider. Local also means stuck with whatever your town happens to have.

Discovery would run through people rather than feeds. Whoever decided what you heard next would be a record-shop clerk, a fanzine editor or a DJ like John Peel on Radio 1, somebody with a name and opinions, instead of a recommendation engine working out what keeps you listening. Something obscure would reach you as a mention, then a mail-order form, then a few weeks of waiting, and by the time it arrived you'd have imagined it so thoroughly that it could hardly disappoint. Scarce culture is also culture that gets thrown away, though. The BBC wiped a huge share of its own programmes because videotape was expensive and nobody expected to want them again, and in a world where anything copied online lived on one company's servers and closed when it did, far more would go the same way without anyone knowing what was missing.

The past would close over faster, too. It's already hard to prove anything about an ordinary afternoon in 1991, and Prestel shows how that happens. UnHerd describes it as all but lost to history, surviving as a few pages, out of tens of thousands, that users happened to save to floppy disk. Without the Internet that would be the normal fate of everything online. People would still scan old magazines and share off-air tapes, but they'd do it on a provider's servers, and when the provider folded or merged the lot would go. The years around 1990 would survive mainly in attics and in what people remember, which would make them more mysterious and probably more moving. The strange part is that 2005 and 2015 would be just as far out of reach.

Sources: