3 ms·

I'd love to know why Google has started ignoring quoted terms and treating them like normal words in a query. I put it in quotes because I wanted that exact phr

by 63 5y ago

I'd love to know why Google has started ignoring quoted terms and treating them like normal words in a query. I put it in quotes because I wanted that exact phrase, not something that sounds similar. If that exact phrase doesn't exist, tell me.

DoingIsLearning 5y ago
I have repeated this comment before and it's wild speculation but I try to answer. For me google's results were good up to 2016 then the results started to become worse and power user features like bolean operators and keywords ("site:", "domain:", etc.) started to get ignored. I map this decline to the exit (with some sex scandals mixed in) of Ranking VP Amit Singhal and the start of John Giannandrea leading R&D. From what I understand from the grapevine Singhal was uber conservative about how anything AI would scale in the ranking algos whereas Giannandrea was pretty much a 'let's AI all the things' kind of guy. This aligns with the start of forced 'natural language' queries. After this roll out you now have to second guess how AI expects a query in order to get half decent results out of google. As an example, in the early 2000's you could just type: "phacoemulsification+resonant" Nowadays the expectation from googles AI is to query along the lines of: "What is the frequency of the ultrasonic driver in phacoemulsification?" (BTW the first query still gives me a correct answer in ddg.) In addition to the AI galore, google dropped all the power-user feature keywords (as in they are still there but we will politely ignore them when convenient), Also from forum discussion the GOOG employees feedback was always that the bolean opearators were underused (probably only a few millions of users?!) so it got removed "but you can use quotes", because it is so much more convenient. Google search has been slowly adding just too much friction. Now I just use ddg and get decent results 85% of time with zero friction/frustration. For my use cases the only scenario where google's AI shines is in 'Google Shopping' product/price search, for everything else it truly feels like the right tech for the wrong problem.

by 63 5y ago

DoingIsLearning 5y ago

I have repeated this comment before and it's wild speculation but I try to answer. For me google's results were good up to 2016 then the results started to become worse and power user features like bolean operators and keywords ("site:", "domain:", etc.) started to get ignored. I map this decline to the exit (with some sex scandals mixed in) of Ranking VP Amit Singhal and the start of John Giannandrea leading R&D. From what I understand from the grapevine Singhal was uber conservative about how anything AI would scale in the ranking algos whereas Giannandrea was pretty much a 'let's AI all the things' kind of guy. This aligns with the start of forced 'natural language' queries. After this roll out you now have to second guess how AI expects a query in order to get half decent results out of google. As an example, in the early 2000's you could just type: "phacoemulsification+resonant" Nowadays the expectation from googles AI is to query along the lines of: "What is the frequency of the ultrasonic driver in phacoemulsification?" (BTW the first query still gives me a correct answer in ddg.) In addition to the AI galore, google dropped all the power-user feature keywords (as in they are still there but we will politely ignore them when convenient), Also from forum discussion the GOOG employees feedback was always that the bolean opearators were underused (probably only a few millions of users?!) so it got removed "but you can use quotes", because it is so much more convenient. Google search has been slowly adding just too much friction. Now I just use ddg and get decent results 85% of time with zero friction/frustration. For my use cases the only scenario where google's AI shines is in 'Google Shopping' product/price search, for everything else it truly feels like the right tech for the wrong problem.