This is more of a technical question than a Buddhism one.
When searching for a sutta, I happened to notice that this commentary(?) shows up in search results, and CTRL-F on these pages confirms it is found on them (see images below): “Notice how the Buddha phrases it: not “concentrate on the breath” as an object, but rather “breathing” as an activity to which one brings mindfulness. The …”
However, I cannot find any mention of it on the linked pages, or actually anywhere on this website.
The technical question is: what is happening here? Is it commentary that isn’t displaying properly on my end?
Hi, that’s one of the notes in the sutta, more exacty, at the end of this paragraph:
And how is mindfulness of breathing developed and cultivated to be very fruitful and beneficial?
It’s when a mendicant—gone to a wilderness, or to the root of a tree, or to an empty hut—sits down cross-legged, sets their body straight, and brings mindfulness to the presentDN 2:67.3. Just mindful, they breathe in. Mindful, they breathe out
This is a comment on the sutta which is mistakenly picked up by the search engine.
I assume you’re using Google? If you could give the search engine and search term that would be helpful.
Anyway, I just tried searching ānāpānassatisutta and my god, how terrible Google has become.
I don’t get the same result as you. The first result has “This passage refers to reviewing or reflective knowledge, looking back over the course of meditation, seeing how the defilements are abandoned, and”. This is another note, further down the page; I’ll come back to this. Then there’s an AI “summary”, which apart from anything else, adds an image of a brain lit up, which is literally nothing to do with Buddhism.
Then Wikipedia, then AtI with another bad excerpt. Then Bodhi’s translation on SC, finally with a good excerpt. This is from the page metadata, which as far as I can see is encoded identically in his translation and mine, so why use the correct metadata here? Lol nothing matters! Next, after results from myself, wikipedia, Thanissaro, and Bodhi, we have the nearly equally authoritative source of … “Guru Viking”.
Anyway, the real conclusion here is that Google sucks.
As to why it is behaving like this, my guess would be that it’s AI.
The note by me that is highlighted in your response begins with “The most fundamental meditation instruction …”, while the first passage in my search results says “This passage refers to …”. I think AI is looking for things that say what the thing is, i.e. things that are as dead literal as AI itself. Which is dumb because, well, check it out.
FWIW, DuckDuckGo is giving me better results, although it still doesn’t get the metadata right.
I doubt if there’s anything to do about this, but I’ve proposed that we add the newly-define ARIA comment role to notes. Perhaps this might help, but it can’t undo AI’s stupid.
We can also review the page metadata, which we haven’t looked at in a while.
I’ve noticed that Google searches seem to favor Bhante Bodhi’s translations… If I was a suspicious person, I’d have thought some people managed to tamper with the algo to favor his over yours!
It’s across many different devices with different Google accounts. And I almost always check bhante Sujato’s translations, since I’m generally looking for the aligned Pali text.
So if it’s tailored to me… It’s some interesting tailoring!
Yes this is generally correct. Even before LLMs, Google has tried to find an “answer” to the query to put as a featured snippet or to respond to in voice queries. Back in ~2016 when I did some SEO work, this was known as “search result zero”, and was a good target for newer sites because you could outrank sites with higher authority scores if you answered the question more directly. Nowadays, the AI Overview is quite different.
Not too much really.
The basics: Google’s own advice is to use metadata tags as preferred snippets, but it doesn’t rank them as highly these days. SC can’t really “curate” to search engines as the content is a translation, not a free-form blog post or such.
One could try to optimize some lengths or make the suttaplex description more visible, but given that it took a normally invisible note instead, I doubt changes like that would make a huge difference. (and that would change the UI/UX too).
A perhaps more practical approach would be to have a different variation of the suttaplex description for the meta tag, i.e. one that is more obviously an “answer” to a query. Then the UI/UX and most reading at least wouldn’t be affected by this more literal style.
Or potentially could have a “definition” of the sutta title as a “term” somewhere (when that makes sense).
But whether those would make a huge difference either, idk.
Hmm the comment role seems to be more for like, social media comments on a post, rather than commentarial notes. Looking at the current role reference on MDN, I might suggest the literal note, which can perhaps be represented more semantically with <aside>. And it could be within a tooltip when that is the setting.
Bhante Sujato’s are almost always the first for me.
In this case, my results are essentially identical to his and Bion’s results, except that the AI Overview is in English:
Ah, I thought I had captured the query in the screenshot, but I was mistaken. It was:
“Notice how the Buddha phrases it: not “concentrate on the breath” as an object, but rather “breathing” as an activity to which one brings mindfulness. The …” site:suttacentral.net
It doesn’t look like it is worth worrying about. It was momentarily confusing because I had happened to not see that commentary.
.now is a static site so it makes sense that it would get the metadata right.
If you run an OpenGraph test on any of the pages it does not find the per-sutta metadata. This is because as I understand, SC is a PWA. Perhpas Google recognizes that all the tags end up being identical so it just grabs something else.
If we could fix the OpenGraph issue, the rest might fall into place.
I know this has been discussed before but I’m not finding the thread.
.express is static, .now is server-side rendered (SSR).
But both mean that the initial HTML served has per-sutta metadata and doesn’t need to run client-side JS to populate it.
SPA (Single Page App) is the term you’re looking for. PWA (Progressive Web App) is unrelated in this case. “Progressive” meaning offline-capable and “Single Page” meaning all other pages are rendered by JS.
Specifically it’s because the tags are modified by client-side JS after the initial render (the literal “single page”).
For reference and to refresh memories, linking to the explanations I wrote in the GitHub issue
No Google’s crawler actually runs JS and has for many (10?) years. It wouldn’t be able to grab the text of the sutta or notes either if it didn’t. So for Google and most modern search engines, there shouldn’t be a significant difference.
DDG’s search engine apparently still doesn’t run JS, although I couldn’t find an authoritative source from them on it. Closest was from their “Duck Tales” podcast (Ctrl+F “JavaScript”) where they essentially say they have some heuristics about whether or not to run client-side JS.
SEO is a bit different (though has some similarities) to social sharing, where apparently most OpenGraph parsers don’t run JS.
and the excerpt is just default metadata, regardless of Ven Sujato or Ven Bodhi.
