I'm struggling a bit with using metadata in quarto. I have a project that includes some custom metadata of the form

key1:
key2:
key3a: true
key3b: false

and so on. I would like to be able to overwrite some of these when rendering to modify which parts of the document are included in the output via
{.content-hidden unless-meta="key1.key2.key3a"}

I've tried to do that using something along the lines of

quarto render -M key1.key2.key3a:false, but that appears to have no effect.

Any advice on how to make this work, or how to achieve the same effect in a different way would be much appreciated.

#quartoPub #rstats

@thomasfuchs I agree with that. Although that isn't something all general search engines do, as far as I can tell. It also doesn't really impact usability that much (at least in this case). For me, the second result was a link to an alphabetical list of African countries, which provides the correct answer.

There are plenty of reasons to avoid Google and too be concerned about the nonsense LLMs pump out, I was just pointing out that this didn't seem like a particularly solid argument for either of those issues to me.

@thomasfuchs I'm not saying you should keep using Google but I think it is worth pointing out that the snippet highlighted by Google is actually from a YCombinator post that highlights the problem with Chat GPT.

Not that there aren't issues with the results returned here, but I think the lesson from this isn't so much "LLMs ruin internet search" (although that may also be true) but "don't rely on skimming of search results if you want reliable answers". I'd expect that to be true for any search engine. The problem is that the Google snippets encourage you to do exactly that.