6 ms·

I see the same behavior. I think it's time to assemble a list of examples.

by daniel_reetz 7y ago

brlewis 7y ago
I bet googlers on HN who work on ranking would appreciate that. Here's a recent one from me: typical tcp packet size
- CDSlice 7y ago
  What exactly do you mean by that? When I search it up I get a SO answer saying what the max packet size is and another (closed) SuperUser answer that says that their is no average tcp packet size because that depends on the specific application. Both of those seem pretty good to me, but I'm not a low level networking guy.
  - dunkelheit 7y ago
    I think the problem is that google confidently and authoritatively displays (in a big and bold font) "65535 bytes" which is an answer to a different question.
  - brlewis 7y ago
    I definitely wasn't asking for maximum packet size, which a search feature showed in huge print as the answer to my question. I don't expect the search algorithm to know my precise intent, but I'd like it to weigh the word "typical" highly. Between google, bing and duckduckgo, only duckduckgo presented this above the fold: https://etherealmind.com/average-ip-packet-size/ https://etherealmind.com/average-ip-packet-size/
    - Kique 7y ago
      I'm seeing that etherealmind link as #4 on Google, and then #3 on DDG. And the results above it on DDG are also related to the maximum size, and the entire sidebar on DDG is the same stackoverflow link about the maximum size. Not sure if DDG is really that much better than Google here.
    - icebraining 7y ago
      In Google, after that card, I get their "People also ask", and the second question is "What is the average packet size?", which links to https://superuser.com/questions/964682/what-is-the-average-size-of-a-packet-sent-via-tcp-ethernet https://superuser.com/questions/964682/what-is-the-average-s... Seems ok to me. It should also be noted that the etherealmind.com link doesn't actually contain the word "typical", so if you think Google should try to outsmart users less, it's not reasonable to blame it for not showing that link.
bogdanoff_2 7y ago
"thesaurus that organizes words by category" - gives a bunch of online dictionary pages. I was trying to find a specific physical book I remember using in high-school. "techniques for interpreting statically typed languages" - gives a bunch of super basic articles/questions about statically vs dynamically typed and compiled vs interpreted languages. DuckDuckGo actually gave better results for this. I feel what's happening is that google is looking at what people are clicking during searches and then using that data to calculate absolute ranks for pages and websites. Those ranks are then not necessarily appropriate for different, less common queries or searching behaviors. However, this algorithm probably performs better in 99% of cases for some internal metrics, so Google ends up keeping it.
- icebraining 7y ago
  > "techniques for interpreting statically typed languages" - gives a bunch of super basic articles/questions about statically vs dynamically typed and compiled vs interpreted languages. DuckDuckGo actually gave better results for this. But is there any good result that includes all those words and which Google failed to find?
  - bogdanoff_2 7y ago
    I don't remember the exact phrasing I ended up using, but I think I remember finding through DDG a StackOverflow question that was essentially "how to make an interpreter for a statically typed language" and it did not appear on Google. To answer your exact question: I don't know, perhaps not.

I see the same behavior. I think it's time to assemble a list of examples.

brlewis 7y ago

I bet googlers on HN who work on ranking would appreciate that. Here's a recent one from me: typical tcp packet size

CDSlice 7y ago
What exactly do you mean by that? When I search it up I get a SO answer saying what the max packet size is and another (closed) SuperUser answer that says that their is no average tcp packet size because that depends on the specific application. Both of those seem pretty good to me, but I'm not a low level networking guy.
- dunkelheit 7y ago
  I think the problem is that google confidently and authoritatively displays (in a big and bold font) "65535 bytes" which is an answer to a different question.
- brlewis 7y ago
  I definitely wasn't asking for maximum packet size, which a search feature showed in huge print as the answer to my question. I don't expect the search algorithm to know my precise intent, but I'd like it to weigh the word "typical" highly. Between google, bing and duckduckgo, only duckduckgo presented this above the fold: https://etherealmind.com/average-ip-packet-size/ https://etherealmind.com/average-ip-packet-size/
  - Kique 7y ago
    I'm seeing that etherealmind link as #4 on Google, and then #3 on DDG. And the results above it on DDG are also related to the maximum size, and the entire sidebar on DDG is the same stackoverflow link about the maximum size. Not sure if DDG is really that much better than Google here.
  - icebraining 7y ago
    In Google, after that card, I get their "People also ask", and the second question is "What is the average packet size?", which links to https://superuser.com/questions/964682/what-is-the-average-size-of-a-packet-sent-via-tcp-ethernet https://superuser.com/questions/964682/what-is-the-average-s... Seems ok to me. It should also be noted that the etherealmind.com link doesn't actually contain the word "typical", so if you think Google should try to outsmart users less, it's not reasonable to blame it for not showing that link.

bogdanoff_2 7y ago

"thesaurus that organizes words by category" - gives a bunch of online dictionary pages. I was trying to find a specific physical book I remember using in high-school. "techniques for interpreting statically typed languages" - gives a bunch of super basic articles/questions about statically vs dynamically typed and compiled vs interpreted languages. DuckDuckGo actually gave better results for this. I feel what's happening is that google is looking at what people are clicking during searches and then using that data to calculate absolute ranks for pages and websites. Those ranks are then not necessarily appropriate for different, less common queries or searching behaviors. However, this algorithm probably performs better in 99% of cases for some internal metrics, so Google ends up keeping it.

icebraining 7y ago
> "techniques for interpreting statically typed languages" - gives a bunch of super basic articles/questions about statically vs dynamically typed and compiled vs interpreted languages. DuckDuckGo actually gave better results for this. But is there any good result that includes all those words and which Google failed to find?
- bogdanoff_2 7y ago
  I don't remember the exact phrasing I ended up using, but I think I remember finding through DDG a StackOverflow question that was essentially "how to make an interpreter for a statically typed language" and it did not appear on Google. To answer your exact question: I don't know, perhaps not.