A foolish consistency is the hobgoblin of small minds.
--Ralph Waldo Emerson
In general, writing advice is a series guidelines rather than Received Word from On High. While there are some things that you never do, most advice is given in the nature of suggestions and must be applied appropriately.
Unfortunately beginning writers all too often take advice as Holy Writ. The result can be pretty damaging to a writer.
The classic example is AE Van Vogt, an immensely popular writer of science fiction's Golden Age. Early in his career Van Vogt read a book titled "The Only Two Ways to Write a Story", by John W. Gallishaw that included a number of rules for fiction writing. One of those rules was that scenes should be about 800 words long and the author should introduce a new plot twist or surprise in every scene.
As advice, that's not bad. It's on par with the admonition to "shoot the sheriff on the first page" when writing adventure fiction. Unfortunately Van Vogt took it not as advice but as a command. He tried to use it in every scene in every book he ever wrote.
The result was an unholy mess. It resulted in bizarrely complicated plots with more loose ends than a poorly knit afghan. To see what I mean, try to read, say, "The World of Null-A". (The problem wasn't helped because Van Vogt was an enthusiast for just about every movement that roiled through science fiction in that era, from General Semantics to Dianetics, and some, such as eye exercises, that never made it into the mainstream.)
In an essay titled The Cosmic Jerrybuilder in his book of criticism "In Search of Wonder" fellow author and SF critic Damon Knight attempted to analyze the plot of The World of Null-A and its sequel The Players of Null-A. The result made painfully obvious what an utter mess the books were.
In an interview late in his life Van Vogt claimed that Knight had misunderstood his use of the 800-word scene. Personally I suspect that Van Vogt misunderstood what Knight had said, since he referred to it from a second-hand source and his paraphrase differs significantly from Knight's actual point.
In a sense it doesn't really matter. The system Van Vogt described in the interview of dividing each scene into five steps and using fictional sentences', each containing a hang-up' is sufficiently rigid to produce pretty bad fiction.
It's important to note that what's wrong here isn't so much the advice as the rigidity with which it is followed.
Believe it or not, it's possible to rise above such handicaps. Van Vogt was such an outstanding natural storyteller that he was able to carry the reader along in spite of his method B at least the relatively unsophisticated reader of the 1940s. It's significant that when Van Vogt returned to SF in the 1970s he was much less successful.
Far too many beginning writers commit the same sin of literalism. Unlike Van Vogt most of them don't have the natural talent to carry it off.
Almost every single rule or aphorism or piece of advice has counter-examples. Even something as important as >show, don't tell' doesn't always apply. Taken literally it produces large sections of prose which don't add to the story and do it in minute detail. Summarization and abstraction are important tools of the writer.
Fundamentally, it comes down to judgment. A writer must have the judgment to select the appropriate tool from her toolbox and apply it to the situation at hand in an appropriate. Eight hundred words is not a bad length to shoot for in a scene, but other considerations are much more important to writing a good scene than length.
How do you develop the judgment? Practice, practice, practice. You write and rewrite. You study authors to like to see what works. Above all, you think about what you're doing.
This isn't nearly as easy as internalizing a rulebook and following it to the letter. But anyone who thinks this racket is easy is severely mistaken.