The description box on a book listing takes markup, not prose, and the character allowance is spent on both. That second half catches people out: the four thousand characters are measured against the finished HTML, so every tag you add is subtracted from the same budget as the words, and a description that fits as plain text can overflow once it is formatted.
This formatter converts shorthand into the markup the listing accepts. Double asterisks become bold, single asterisks or underscores become italic, hash marks become headings, and lines beginning with a dash or bullet are gathered into a list. A blank line becomes a break rather than being silently collapsed the way raw pasted text usually is.
The counter beside the output measures the converted markup, which is the number that actually has to fit. Bold costs seven characters each time it is used, a heading nine, a list nine for the wrapper plus nine for every item, and a paragraph break eight. Twenty bold phrases, ten bullets and thirty paragraph breaks come to roughly four hundred and eighty characters of markup before a single word is counted.
Because the limit is applied to the markup, and the tags count. A heavily formatted description can carry several hundred characters of tags that are invisible in the rendered result — exactly the difference between what you wrote and what the counter reports.
Bold, italic and underline, line breaks, headings in the lower part of the range, ordered and unordered lists, and paragraphs. It produces nothing outside that set, because markup the field does not accept is either stripped or shown as literal text.
It rarely survives. Word processors emit large amounts of styling markup the description field does not accept, and the usual result is either a stripped block of undifferentiated text or visible tag fragments.