Hacker News new | past | comments | ask | show | jobs | submit login

"But aren't you a part of the relevance equation? The ideal results for a search like [bitcoin crash] should be different for a Japanese-speaking searcher in Tokyo vs. a German-speaking searcher in Munich vs. a bitcoin expert vs. a programmer trying to diagnose why compiling bitcoin is crashing vs. my Mom who has never heard of bitcoin before, right?"

Maybe, but that is not the point.

Besides, you are mixing localization (japanese vs german) with personalization (expert vs naive), and what is worst is that you are assuming that Google knows so well each user so as to be right (and that is either impossible, either extremely creepy), at every single instant of his life (a person can change interests).

Furthermore, to grab your example, how can Google know that an expert in bitcoin and expert in bitcoin compilation and crash solver, is not just interested in hearing about the "market" crash of bitcoin?

Google CANNOT read the users' mind. And even if it did, there would not be a need for "personalized search", as the "mind reading" would give enough search criteria to nail the results more easily (albeit, most people do not know exactly what they want, so it will still be an iterative process, which is good, as randomness is the seed for evolution).

So, back to the point, it is that in the quest for "adequate results" for each person, Google is turning web search into a non-deterministic event ().

Imagine the web being a library, and the search being searching for the library's book database, why would the search for a given book return different results to different persons? It should always return the same results, if the person doing the search is not satisfied with the results, then she/he will add more criteria. In other words, let the person do the filtering!

Once that Google accepts that in his quest for "better results" (where 'better' is a concept decided by solely Google and whose ranking parameters and algorithm are unknown) there is a potential (probably demonstrable already) for a "filter bubble" with positive feedback loop on user behaviour, which, as with any positive feedback loop, can go out of control, exacerbating certain ideologies and fueling extremisms.

And there is a fundamental difference between a "self guided" (as in self controlled) filtering, where users would knowingly filter out results in order to find those that they like, and a "google guided" (as in externally controlled), filtering.

() strictly speaking, search will not be deterministic as the web is a dynamic system and it grows, so search results can vary with time, but they should not vary from person to person at a given time.




Of course Google can't read minds, that does not mean they should ignore information that they have when deciding which results to show or what order to show them in. Sufficient data to provide a better filter for a given user is not the same as mind reading.

You are absolutely right that more criteria should be used if a user is interested in better results, but I fail to see why a deterministic base case is superior. If I never click on news links and always click on travel links, it makes perfect sense for Google to assume that my search for "Egypt" is looking for information related to travel to Egypt. If I am not following my usual search patterns, I can look on the second page, or I can disable personalized search, or I can search for "Egypt News" all of which would give me better results.

Why is it preferable to always make everyone clarify their searches when there is sufficient information to narrow the search down somewhat without requiring additional intervention by the user? This is usability 101 right here.


http://www.google.com/ncr to disable. NCR=no country redirect.

Works fine, but how do I return to country redirect? Am in Germany, used it for some global searches, now want my localized (German ) searches back. Must be missing something obvious here, please help, thanks!




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: