OSNews

I've written a lot of documents using LaTeX and I'm afraid the author has overlooked one terrible problem; indeed he even acts like it doesn't exist! In particular, he writes: You add one sentence, which then pushes an image on to the next page, leaving a massive gap at the bottom of that page where your image once was. This then daisy-chains down, knocking other tables and images out of place all the way to the end of your document! It's a real laugh. Fortunately, Latex is much more clever in this respect and positions your images and tables with a lot of common sense. So, if you want your image to appear at the bottom of a given page, it'll stay there!

Heh, heh. Not really.

LaTeX is infamous for doing exactly the same thing; in fact it's worse, because its algorithms insist on doing all the thinking for you: so, if you want an image (or a math formula) in a certain spot, good luck getting it there! The choice of vspace is often determined by what goes on two or three pages before (or after) the current page you're on, so changing two pages before can really get you confused. I struggled with this very problem in my dissertation: white space would automagically appear all over the place. It took a long time to get things acceptable (I'm still not happy).

For horizontal spacing issues, however, you have the reverse problem: LaTeX is that it would rather overrun a right margin than leave too much space between words. (The infamous "overfull hbox", whose black slug indicating an error certain styles remove incidentally, even in draft mode... grrrr) I'm not sure why Knuth thought an overrun was such a better idea than extra whitespace in an hbox, while extra white space was preferable to an overrun in a vbox, but the result in many published papers, and even some books, has been ugly. It certainly does not look professional, but the only way to fix it is to do some really obscure TeXing, or else completely rephrase your wording (the universal fallback I've seen in all LaTeX manuals). This, I think, is the worst problem: your choice of words is necessarily less important than LaTeX's obscure rules. It's the one frustration that unites both beginners and experts: when you're proofing a 260-page document, you have to remember that hyphenated words like "S-polynomial" (for example) won't break across lines, so they need manual hyphenation, and you often won't notice the overfull 2-pt hbox unless you're VERY CAREFUL (or you pay separate the wheat from the chaff in TeX's output).

Those complaints aside -- for scientific publishing, LaTeX far outclasses anything else. I've never used Word to write a math document; given what I've heard from people (and what I read here) I never will. It was immensely classy of Knuth to make his program freeware.