And during her frantic searching, his girlfriend won't try something as simple as "python programming book"? I find that hard to believe, especially if her default association for "python" is the snake.
I'm almost certain she would use a more sensible search term. In fact: she'd most likely have the title already, having picked the brains of a friend or colleague. You seem to have missed the point of my (admittedly flimsy) example though.
What I am trying to illustrate is that the alleged relevance of a set of results biased on the basis of an individual's profile is incredibly insular (and harmful) if it's there at all.
Boiling it down even further: I believe that search results should reflect what is on the Internet, not just that portion of it that Google deems 'relevant.'
...anything else, in my opinion, skates dangerously close to what most people railed against in the great Net Neutrality debate of 09.
> What I am trying to illustrate is that the alleged relevance of a set of results biased on the basis of an individual's profile is incredibly insular (and harmful) if it's there at all.
> Boiling it down even further: I believe that search results should reflect what is on the Internet, not just that portion of it that Google deems 'relevant.'
And what I'm trying to illustrate is that you're taking an overly simplistic approach to the question. Google takes in a lot of signals (links, country and language, past behavior with respect to previous queries, sites you've asked to be blocked, ...) and now social signals. Where do you draw the line and why? And why isn't a user responding with more detailed guidance ("I didn't mean X [even though I often do], I mean Y right now") a reasonable exceptional flow some of the time?
>And what I'm trying to illustrate is that you're taking an overly simplistic approach to the question.
I'd agree with that. In fact, I'd say that's a lot to do with my point... I believe that search results should be literal. A search for 'python', in my opinion, ought to return links to information about the animal, the language, the character from The Jungle Book, etc.
>And why isn't a user responding with more detailed guidance ("I didn't mean X [even though I often do], I mean Y right now") a reasonable exceptional flow some of the time?
This is interesting. I guess what I'd like to see is something along these lines. Turning it over in my head earlier: the thought of making two distinct types of searches did cross my mind, ie: searching from your G+ homepage returns focused results from the web at large with social results intermingling whereas searching from www.google.com returns literal results with no fiddling. The user is enabled to select which one is used by default from the search bar in their browser.
I'd draw the line somewhere in the past, before all that stuff you mention.
Links in Hebrew (I'm in Israel) are unreadable and not helping.
I constantly end up with German results polluting my search (maybe because I signed up when I lived in Germany? No idea, it's out of my control). It is a hassle to fight them.
It got never better. I gradually noticed a trend of decreasing usefulness of Google search, without any visible sign of getting better. Only more dumbed down. And now 'social'?
I hope you aren't alluding to anything as ridiculous as "search neutrality". There is no such thing. There are certainly less optimal results, in several dimensions with a poorly defined fitness landscape, but that complexity is exactly why its so ridiculous to talk about any search results "reflecting what is on the Internet".
You have provided no metric for that statement, so how could it even be measured?
There is too much on the internet for you to read it all. Finding what you want at some particular time is going to require a curator, and they are going to be biased.
(besides, the overfitting you describe isn't exactly the hardest problem to figure out and fix (just mix in some other python results and that related results thing that's already there). The hard part is quickly identifying that you actually are affecting users negatively and not just the guy that thinks everyone should have to look at nextag results)
>...but that complexity is exactly why its so ridiculous to talk about any search results "reflecting what is on the Internet".
>You have provided no metric for that statement, so how could it even be measured?
That's fair. My comment is massively vague.
Maybe I could reframe that comment as a question: are you of the opinion that moving results that herd users into Google's own products and content above results that are perhaps less biased, more objective and do not serve to enrich Google directly is better for users than the way that Google search has worked up until this point?
I know that the subject of search is hugely complex and that Googles algorithms already define what is returned to the user but I am questioning what I perceive to be a change in motivation and a change in quality as result.
