This paper reports on recent research into the usability of editing software for structured documents. The overall objective is to identify why markup-based editing is regarded as ‘difficult’, and to investigate the potential for changes to the prevailing interface paradigms which might address the problem. The principal systems investigated were SGML-based (XML) and TEX-based (LATEX and ConTEXt).

The demand for markup-based editors was categorised and quantified by an analysis of posts to the Usenet newsgroups comp.text.xml and comp.text.tex and the XML-L mailing list. This highlighted some perceived deficiencies with existing systems as well as some underlying misapprehensions by potential users.

Baseline data on attitudes to XML editing software was gathered from a pilot group of expert users who were surveyed to identify their attitudes and expectations in respect of editing software, both for customer implementations and for their own work. This revealed further dissatisfaction with existing systems, especially for customer use (Sect2 2.1).

Twenty-five editing systems were examined to compare the facilities provided, and to derive a comparative measure of their usability, accessibility, and usefulness. This showed that there is no significant difference between editors in the facilities offered, only in the accessibility (placement of those facilities in the user interface).

Investigation of the requests and comments from new users identified some conflicting perceptions, specifically that a) XML is perceived as hard to use because the editors don't look or work like wordprocessors; b) some users are convinced that an interface to editing XML must exist that can be used without configuration, by novices with no training, to produce completely fault-free documents. Work is ongoing to identify the extent to which the [mis]perceptions may be resolvable through changes to the interface.