A critique of TeX?
I've just found this paper, which seems to address some of the issues I want:
25 Years of TeX and Metafont: Looking back and looking forward — TUG 2003 keynote address by Nelson Beebe [TUGboat 25:1, 2004]
Abstract:
TeX has lasted longer than many other computer software technologies.
This article reviews some of the history of TeX and METAFONT, how they have come to be used in practice, and what their impact has been on document markup, the Internet, and publishing.
TeX has several design deficiencies that limit its use and its audience. We look at what TeX did right, and with 25 years of hindsight, what it did wrong.
We close with some observations about the challenges ahead for electronic representation of documents.
Are there any other published references?
E-TeX: Guidelines for Future TeX Extensions by Frank Mittelbach [TUGboat 11:3, 1990]
Abstract:
With the announcement of TeX 3.0, Don Knuth acknowledged the need of the (ever growing) TeX community for an even better system. But at the same time, he made it clear, that he will not get involved in any further enhancements that would change The TeXbook.
TeX started out originally as a system designed to typeset its author's own publications. In the meantime it serves hundreds of thousands of users. Now it is time, after ten years' experience, to step back and consider whether or not TeX 3.0 is an adequate answer to the typesetting requirements of the nineties.
Output produced by TeX has higher standards than output generated automatically by most other typesetting systems. Therefore, in this paper we will focus on the quality standards set by typographers for hand-typeset documents and ask to what extent they are achieved by TeX. Limitations of TeX's algorithms are analyzed; and missing features as well as new concepts are outlined.
LuaTeX: Howling to the moon by Hans Hagen, (Tugboat, 26:2, 2005)
Not really a critique of TeX the program, but explains some reasoning that went behind luaTeX--the future of TeX.
Abstract:
Occasionally we reach the boundaries of TeX and programming then becomes rather cumbersome. This is partly due to the limitations of the typesetting engine, but more important is that a macro language is not always best suited for the task at hand.