Part of the reason I've been so long in posting is my typical obsessive-compulsive problem with wanting to structure my posts. This is, however, a blog, so a "stream of consciousness" approach is eminently defensible.
As a writer and amateur literateur, I can understand the temptation to dabble in Great Projects of composition or translation in an auxlang. It's a bad idea, though. And curiously, the linguistic maturity of both auxlang and user have the same effect: younger means smaller.
The upshot is that new users and auxlangs should stick with short projects--probably just a paragraph or so. An auxlang designer may want to try something longer as a proof the auxlang works, but basically only a more advanced learner will bother reading anything over a page or so. A new auxlang has no such learners except the designer, so it's pointless to work on anything long at first.
At-sight projects skew this slightly: the less at-sight readable your project, the shorter the early texts should be. On the other hand, an at-sight project can have longer texts: Gode's Dece Contos in Interlingua are fairly reasonable both as writing and reading exercises.
Now, all this goes against the grain for newbies. "It's so easy, I bet I could do a novel!" No. Novels have all kinds of problems, even if you're just translating. Stick with short stories--the shorter the better. Otherwise you'll get a few chapters in, run into more and more difficulties, and give up.
(For readers, short chapters or sections can make even a novel seem less daunting; I've seen this sort of thing done with longer short stories and novellas on certain blogs. But the auxlang has generally been "at sight.")
What are the advantages of shorter pieces?
* Easier for writer and reader.
* Allows more variety. A novel will generally stick with a genre and setting (place and time); short pieces can cover a wider range, giving examples of numerous types of texts. How do you say, "Once upon a time"? Cliché phrases are actually useful for learners, and they only become a problem through overuse, which won't be an issue at first.
* Less likely to contain needless difficulties. I'll explain this next with a critique of Dracula as a translation project.