This goes down to how Google indexed the pages and do they consider '[something' as a word on its own.
Normally during indexing, texts goes through a process called tokenisation which breaks texts into individual words. Symbols are normally removed in this process. In some rare cases, however, the symbol is part of the word, so special cases must be added to keep those symbols during tokenisation. Obviously they can't really maintain all the special cases manually, so they probably did some data mining in search queries and webpages to see which words when accompanied by certain symbols seem to mean something totally different.
For example, if you search
C# the result will be different from
C, but the results are the same for
ke ha and
ke$ha. Since people would rarely search for
ke ha alone, not treating them as special cases will probably speed up indexing.
In your case, very few people has probably searched for
[something before and even if they did, their behaviour probably indicated that the results for
something were also relevant, so
[something would have never made to the special cases list and will be indexed as
something in their databases.
Thus it's probably not possible.