I wonder how much of the awfulness of their search results this was responsible for.
Conversation
Honesty their results would be so much better if they used lynx to crawl.
2
Too many sites are using tons of JavaScript, so they would be entirely opaque to it. There's a lot of blog and forum software that requires JavaScript to display any results at all. Earlier versions of Discourse couldn't show anything without JavaScript and it had wide adoption.
2
Exactly. Penalize sites that aren't accessible with -100% pagerank.
2
I just want to be able to search that content. I also hope they start making better use of structured data. In theory github.com/GrapheneOS/gra is supposed to help search engines understand the content earlier on the page, but I think the only part Google understands is the logo.
1
It's also frustrating how much effort and research it takes to deal with stuff like getting proper link previews on social media like grapheneos.org/build. I need to figure out the different image sizes they want so I can provide different images and they'll choose correctly.
1
1
One example: github.com/GrapheneOS/gra. Why doesn't Twitter use the normal description? It makes no sense to require repeating this on every page. I have no clue what the documented twitter:site / twitter:creator are meant to do. Can't wait to debug structured data for blog posts.
1
Look at that ugly GitHub link preview because they set <meta name="twitter:card" content="summary_large_image" /> and turn the user's avatar into a cropped rectangular banner. In my opinion, Twitter should be able to show link previews based on the standard title and description.
1
1
Should also use a standard site icon as the image if you don't override it for a page. I can't understand why I need to deal with all this nonsense. Nearly no sites / web developers do any of it properly. Open Graph metadata makes sense but it should be optional to override.
1
1
Structured data could have helped a lot with search but it's poorly supported and there are 4 different ways of marking it up: RDFa (w3.org/TR/html-rdfa/), HTML microdata (w3.org/TR/microdata/), JSON-LD (w3.org/2018/jsonld-cg) and microformats (microformats.org/wiki/Main_Page).
1
1
How does this kind of insanity even happen? I like JSON-LD the most because I can stick it as a separate thing at the bottom of a page without completely messing up the content trying to embed contrived inline structured data markup into it, but then it's less directly marked up.
Also, HTML offers a standard content outline algorithm now, which you can use via an extension like chrome.google.com/webstore/detai, but too few sites support it. It's really nice on pages that are marked up sanely and could be a really nice thing for accessibility, but most sites suck.
