The worth of a scientific paper is judged by its citations in other papers. Daedalus's electronic publishing database of last week ranked the citations themselves by the citation score of the cited authors. He now has more consequences of pure Internet-based publishing.

For a start, it will relax all size constraints. Many scientific papers are jargon-packed and atrociously written because the authors have to squeeze their message into limited journal space. And thousands of interesting small observations are never published at all, being too brief for any paper. In particular, many negative results languish in file drawers. But an electronic system, with the total indexing it makes possible, could publish papers of any size. All would have the chance to be retrieved, studied and cited, bringing prestige to their authors. And with number of citations, not number of papers, as the criterion of prestige, size doesn't matter.

How would the system be paid for? A scientific paper is an advertisement for its authors (indeed, some are little else). So page charges levied on those authors should work well. The tricky problem is refereeing, to keep out (or at least point out) the nutters and incompetents. A democratic database should accept any paper, however mad, provided the author could pay for it. But he would be wise to offer it to a referee, who would ‘mark’ it, and append comments in the form of a separate note cited in the paper. This citation would neatly reward the referee for his work. The system would multiply the referee's mark, positive or negative, by his own citation score. An author who approached a high-ranking referee would thus be taking a big risk — condemnation would give the paper a very negative mark. Users could scan the database through a filter cutting out papers below a certain mark, thus excluding the low-ranked stuff.

The whole database would probably divide itself into self-contained mutually citing regions. Some would be inhabited by cliques of nutters all citing each other. Some would form rival schools of mutually negative refereeing. But science, alone among philosophies, converges on consensus. The biggest mutually citing region would be the domain of respectable, authoritative scientific opinion.

The further Inventions of Daedalus (Oxford University Press), 148 past Daedalus columns expanded and illustrated, is now on sale. Special Nature offer: m.curtis@nature.com