Careless use of the ReferenceExpander script has led to references being contracted instead. For example, the script sometimes follows a link that now redirects to a new, uninformative place, but since the link technically "works" the auto-generated citation omits the archive-URL and creates a footnote that is nicely templated but completely useless. It also removes all sorts of ancillary information included in manually-formatted citations, like quotations. If multiple citations were gathered into the same footnote, it creates a replacement based on only the first of them. It can see a citation to a chapter in an edited collection and replace the authors' names with the editors of the volume. It can see a URL for a news story and create a {{cite web}} footnote that omits the byline which had been manually included. Even the edits that made articles bigger, in terms of raw character count, could lose information. For example, an article might have cited multiple different webpages on the same site, all of which were later redirected to the site's home page. ReferenceExpander would then turn all of those citations into identical {{cite web}} footnotes, which another tool would then merge in a later edit.
We ran a database query and sorted the results by change in article size, to find instances where the script likely had the wrong effect. It's a... rather long list. Some of the entries on it may be fine; some of those that were problematic may have already been fixed by routine work. But help turning those s to s would be greatly appreciated. Due to the variety of the failure modes and the possibility of intervening edits on the affected articles, there isn't a push-button fix.
Progress as of 19:40, 2 November 2024 (UTC): checked: 2507; remaining: 14.
ReferenceExpander has been the subject of the following discussions, in chronological order of when they were initiated: