3 ms·
This seems to be a semantics issue. "To search" has multiple meanings, and only one of them implies loss. In computer science, you don't (only) use search as a
by rbehrends 15y ago
This seems to be a semantics issue. "To search" has multiple meanings, and only one of them implies loss.
In computer science, you don't (only) use search as an information retrieval mechanism. You use it primarily (whether that's Google, find <dir> | xargs egrep <pattern>, Microsoft/OS X help, etc.) as a mechanism to reduce entropy.
Information stored in files or on the internet is generally too vast to be easily absorbed by a brain that, marvellous as it is, has difficulties processing more than 7 +/- 2 symbols at a time [1]. We need a way to reduce the entropy and extract and filter information. And that's what search tools in computer science primarily tend to do (online or offline).
This is also why by default we use "and" to combine search terms and are generally more interested in narrowing search results rather than expanding them.
Some people admittedly also use search as a tool to look up information; I recall that there's some usability research where some people prefer using search, while others prefer a catalog scheme (such as bookmarks) and that forcing one type of person to use the opposite scheme reduces their productivity.
[1] http://en.wikipedia.org/wiki/The_Magical_Number_Seven,_Plus_or_Minus_Two http://en.wikipedia.org/wiki/The_Magical_Number_Seven,_Plus_...