The quality of search results has drastically fallen off for my use cases lately. It seems like the user categorization they used to do is entirely gone. Prev

The quality of search results has drastically fallen off for my use cases lately. It seems like the user categorization they used to do is entirely gone. Previously my searches would often bring stackoverflow or other programming forum posts to the front, now it's an absolute chore to find them if I don't add a specific "site:" or other "shibboleths" to the query. It feels like they're trying to use language models to add context to the search terms, but because of the weakness of the model, it very often misclassified what my search concerns and shows the least relevant results possible. This also seems to make it much more likely to selectively ignore words from my search queries. The "must include: <term>" on search results is exceptionally frustrating. I didn't add words to my query for fun, I really do expect every single word to be considered when I make my search.. why would I have wasted time typing it? Do they really expect me "to" "quote" "every" "single" "search" "term?" Finally.. their "did you mean" query diversions have gotten much worse lately, again from what feels like total misclassification of my search. This happens quite a bit with queries that I used quite often and always previously worked. They've clearly been making changes, and those changes, for my use case, have essentially ruined what the product used to be for me. I'm now actively trying my queries on other platforms, and as soon as I find one that does even slightly better than google, I'm out.