When I am preparing a page for the web, the last stage is to validate the HTML – use software to check that my coding is correct. This is a helpful way to spot small faults, tags opened but not closed, images with no alt text, and so on.
But – particularly if my starting point was a word-processed document, or has foreign words with accented letters – it also queries non-standard characters: not just the és and üs, the ampersands and double quotes (which have their own uses in coding, and can’t therefore be used in the text) but also non-standard punctuation, smart quotes and smart apostrophes.
Sometimes these are numerous, and it all seems like nit-picking: is it really worth the bother? And then I visit a web page (like this one, for a vineyard in France) which has skipped this detail, and where my browser stumbles over the unfamiliar characters, in this case rendering every o circumflex (ô) as TM – and realise again that yes, it is.