Summary

  • Eric W. Weisstein recalled choosing groups of letters each day and blocking their entries in November 1998, after a request to restrict the site to encourage book sales.
  • By mid-1999 he removed the alphabetical restrictions, saying his improved monitoring could target large downloads. Ordinary readers had borne the earlier rule's inconvenience.

A mathematical definition can be perfectly sound and still be unavailable because its name begins with the wrong letter. In late 1998, that was a real access rule on the website that later became MathWorld. The mathematics was dependable enough to print. The alphabet was being put on shifts.

Eric W. Weisstein described the arrangement in his signed account of 6 November 2001, preserved on an ESPCI faculty page. His earlier site, Eric's Treasure Trove of Mathematics, had gone online in 1995 while he was a graduate student at Caltech. Readers supplied corrections, advice and new entries. The reference kept changing as people used it.

A printed version had an obvious use. It would work away from a computer and reach readers who lacked Internet access. Weisstein began looking for a publisher in February 1996, treating the book as a snapshot of an evolving collection. A snapshot makes a good companion to a living reference. Trouble starts when the companion wants the reference to become less useful.

According to Weisstein, a publisher's request in late October or early November 1998 asked him to remove parts of the website to encourage purchases of the book. In November he began selecting groups of letters at random each day and blocking all entries beginning with them. Material would return after a short interval. A reader's immediate task still had to wait.

The rule was wonderfully easy to apply to the wrong people. Someone wanting one definition encountered the same alphabetical obstacle as anyone else looking in that part of the collection. The first letter of a term carried no information about how much material its reader intended to copy. A convenient organising device had become a rather indiscriminate doorman.

Weisstein himself found the arrangement unsatisfactory. He described improving the site's monitoring and access controls to deal with attempts to download large bodies of material. By mid-1999, he believed those controls could do the job and removed the letter-based restrictions. The change moved the proposed target from the name of an entry to the pattern of requests.

The improved site took the name MathWorld in December 1999. Years later, in his February 2004 account of its development, Weisstein was still emphasising reader feedback and inviting contributions. That continuing editorial work explains why the online reference had its own value. A fixed edition could sit on a shelf; the website could incorporate the next correction.

The two access rules also assigned different jobs to the operator. Alphabetical blocking needed a set of letters and made ordinary lookup less reliable. Behaviour-based blocking needed monitoring and a judgment about suspicious use. Removing the earlier rule spared readers a wholly unrelated test: whether today's missing mathematical fact happened to have an unlucky initial.