"When a measure becomes a target, it ceases to be a good measure" (Goodhart's Law).
Google measures how "good" pages are for a given search query. (It doesn't matter how.) Those who own the pages want those pages listed as high as possible, so the measure of how good their page is becomes a target (for them, not for Google). That's enough to make Goodhart's Law apply to Google rankings.
I have noticed that these sort of results pop up whenever you google something related to a Windows problem, but when you search for something related to Linux problem the results are way better.
Yes! And this is all the more fun when you are a Windows sysadmin and need that deeper level of knowledge to solve a stupid Windows-related problem for work.
I doubt the report is factual (fyi, here's what came out for the same query in my browser The twin-spotted sphinx moth is a pretty moth with beige-brown and white wings and pinkish-red and yellow markings...), and I haven't really experienced such results anywhere. Maybe it has to do with the fact that I regularly delete my browser data, but the point of this critique is lost on me.
FWIW, I just ran that query, and got what looked like materially better results.
That being said, I've seen the same sort of useless results elsewhere (so maybe someone saw this particular case and intervened). I have been told that part of the SEO industry works like "take this sample post and rephrase it with the following keywords and product links". So you get the same warmed-over results on the topic, and it's hard to find the posts that drill deeper.
Similar to this but not (maybe) as nefarious is the links to what looks like the result of the <solemn-and-important-font>Quarterly Conversation with Your Manager to Set Goals and Milestones</solemn-and-important-font>. The outcome of the QCYMSGM is "Goal: understand $(technology); Milestone: write a blog post about it". So you get a lot of half-baked "Intro to OAuth2" links that don't really help that much.
One comment after the Twitter post suggested trying neeva.com. First-page results are useful.
For some kinds of search, DDG's image search ... (faster: append '&iar=images&iax=images&ia=images' after the search term(s) ) ... is the best. For example the search on 'brown spotted moth southern us identify' quickly shows several useful images; click on one to see the source title & url.
This is scarily accurate, and partly why I've switched to Kagi (www.kagi.com) for what Google used to be. While they will introduce their paid service soon, I believe it would be worth every penny.
Being this is HN I'll give you a very "probably not your target audience" response from someone with no prior exposure with either:
Kagi seems to have a strong stance and commitment on monetization being solely through subscriptions and being user centric as a whole while You doesn't even seem to be able to commit to offering a path away from ads in your search results; instead just focusing specifically on being more privacy friendly. Based on the names I was almost expecting the inverse where You heavily promoted itself as the hyper user ("You") centric option in all aspects and Kagi to be the differentiated but still somewhat product focused offering.
I can't compare actual usefulness of each as Kagi is invite only at the moment but I will say You seems to have a ton of reflow and it can be hard to follow results. The reflow gets progressively less bouncy on load the smaller the screen is though and if you go down to the responsive/mobile layout size it works great. On a large desktop screen doing a search results in a bunch of pop in/out all over the page and even after it loads scrolling results in more pop in near the bottom bar. There are placeholders but because there are so many different types of visually different result containers which don't all have appropriately sized placeholders they just create more reflow than they save. Moving the mouse around the page causes everything to instantly start reacting and sometimes things like videos stop when you mouse off while other times they don't. The left bar's abrupt mouseover transition falls into this "suddenly reactive page" category as well as it shifts the entire page to the right the instant you hover it and the entire page goes back the instant you don't.
Overall it feels like getting the kitchen sink thrown at you as it loads results e.g. searching "picture of the mona lisa" gives 3 different inline picture streams and a music list in the first page results. The picture streams... alright, it's what the query was asking for even if it's a bit much to have 3 separate displays for it but I'm not sure why it feels the need to meet a "special display section" quota by displaying songs from Spotify. I'm guessing a lot of this goes away as you customize the site and do things like disable the "AI" but it's a hard sell to convince someone to put time and effort into figuring out if they can customizing something enough to their liking as part of their first impression.
Thanks. Very helpful! We will work on the reflow issues and visual app ranking, etc.
Yea, the mona lisa example shows the complexity of ranking.
All other search engines (outside Google and Bing) just take ranking from those two. We're building our own but that comes with some early bugs.
If you try some coding searches you'll see how much we care about this target audience in the actual product. Not sure any other search engine has this many useful features for developers?
I want to know what happened. What were the individual things that made sense in isolation, but added up to this steaming turd.