What the press actually changed
Before Gutenberg, a book was a luxury object, copied out by hand over months. A study of manuscript-book prices in England between 1300 and 1483 found that roughly three quarters of a book's cost was sheer labor. Print broke that. Once the type was set, the thousandth copy cost almost nothing, Europe produced some twenty million copies of printed books before 1501, and literacy climbed for centuries afterward.
Volume, though, was the byproduct. The revolution was what the historian Elizabeth Eisenstein called typographical fixity: print made identical copies that did not drift each time someone reproduced them. Before print, a wrong digit in an astronomical table, recopied by hand a dozen times, became a dozen different wrong tables. After print, a scholar in Antwerp and a scholar in Krakow could argue about the exact numbers on the exact page. Fixity is what lets knowledge accumulate. A result can be built on only if the result being read is the result that was written.
The historian Adrian Johns supplied the necessary correction in The Nature of the Book: fixity was not a gift of the machine. Early print was full of errors and piracy; even the first printed edition of Copernicus's On the Revolutions departed from his manuscript in hundreds of places. Trustworthiness had to be built by printers, proofreaders, and institutions who made getting the copy right their business. The press made reliable knowledge possible; people did the work of making it real.
Half the write-up is still hand-copying
For many scholars writing is the thinking, and the methods section rewritten three times got better because the rewriting sharpened the method. Any tool that claims to take the writing off a researcher's desk is claiming to take the thinking, and a researcher should distrust it on sight.
But look at what the months of write-up actually contain, and the work splits in two. One half is thought: the argument, the interpretation, the framing, the sentence that finally says what the result means. The other half is reproduction: the citations chased down and formatted, the reference list rebuilt by hand, the manuscript reassembled to a second journal's template after the first rejection. That half produces no new knowledge. It is the work of getting finished knowledge into a trusted, conventional form. It is the scribe's job, and it never went away. A well-funded lab absorbs it with staff. A solo researcher pays for it in sleepless nights, and a researcher writing in a second language pays again, in hours of language conformance a native speaker never sees. Real results sit in drawers because their authors ran out of months before they ran out of ideas.
The press's three effects, of cheap knowledge acquisition, quick distribution, and standardized copies, land on exactly that second half. It lowered the barrier to having a book at all; JigSpec's Paper Writer lowers the barrier to having a defensible draft. It collapsed the time an idea took to cross a border; this collapses the months between a finished result and a submittable manuscript. It killed the transcription error by fixing the copy; this runs a citation cross-check against the actual sources, so the references in the document match the references that exist. The argument never leaves the researcher's hands, because the argument is the research.
The peer review addition
One place this analogy falls short is that the machine is neutral. The same press that carried the scientific record also printed quack medicine and the Malleus Maleficarum, the witch-hunting manual, and the models behind today's writing tools still invent references at alarming rates, with fabricated papers already surfacing on Google Scholar. Cheap copies guarantee nothing about whether the copies are true.
But print did not become trustworthy because the technology improved. It became trustworthy because people built a layer of checking around the machine. Printers hired in-house proofreaders whose entire job was catching errors before a page became public; the historian Anthony Grafton treats these early proofreaders as scholars in their own right. When mistakes slipped through anyway, printers bound an errata sheet into the book, a printed list of known errors. The machine made the copies, but the checking layer made them worth trusting, and that is what turned print from a rumor mill into the engine of the Scientific Revolution.
AI writing tools face the same choice today. A tool that generates text unchecked is a faster way to publish mistakes: a large-scale audit of 111 million references found a conservative estimate of 146,932 hallucinated citations entering the record in 2025 alone. In a recent benchmark, adding an automated checking step cut the rate of dead or fabricated citation links to under one percent, so JigSpec's Paper Writer runs a second check against the text of each source.
Johns's lesson applies here too: trust was never built by machines vouching for themselves, and it will not be now. The check exists to be inspected, not believed, and the name on the paper stays the researcher's. The model is the press. The verification is the proofreader. And a proofreader could guarantee the copy was faithful, never that the argument was good. Whether an idea deserves a paper at all has always been the researcher's question, and it stays that way.
The press didn't come up with the idea
The press never wrote anyone's book. It made the copy cheap, fast, and trustworthy, then got out of the way, and what followed was the Reformation, the Scientific Revolution, and the modern scientific record.
The reproduction half of the write-up is the last place in research where the copy is still made by hand, and for five centuries its cost has helped decide whose results enter the record. It still does. A result today faces two tests: whether it is true, and whether its author can afford the months it takes to put it in publishable form. Only one of those tests belongs in science. The technology to retire the other now exists, with the checking layer built in from the start instead of grown over a century. History has run this experiment once. When the cost of reliable copies collapses, the record does not drown in the flood of new work. It compounds. What stays in the drawer after that is a choice.
Common questions
Doesn't lowering the barrier just mean more low-quality papers?
The barrier it lowers is the mechanical half of the write-up, not the bar for a real contribution. Print lowered the cost of the book and the result was the Scientific Revolution, because a layer of checking grew up alongside it. The citation cross-check is that layer here: it runs as its own separate check against real sources, built to catch fabricated references before the version that gets read. The bar for a real contribution still belongs to the researcher and the reviewers; the check just keeps bad citations from hiding inside good formatting.