I tried to find a related question here, but was unsuccessful.

Someone on SO asked what does "!!" mean... I tried searching SO and was presented with the search options page. Even Google fails at this search.

So my question is, is there some way to search for an exact word (in quotes) that won't cause it to fail?

  • Search on the trilogy sites isn't always that great... I'm not sure you'll have much luck with it. – alex Dec 14 '09 at 14:15
  • 21
    Non-alphanumeric characters are very important in many languages. Not being able to search by these characters makes certain questions impossible to search for. – Annika Backstrom Apr 26 '10 at 18:03
  • 2
    At least they're up-front about it: "Generally, punctuation is ignored, including @#$%^&*()=+[]\ and other special characters." - Google help page – Pops Jun 23 '10 at 19:14
  • 7
    status-bydesign, or status-byimplementation? – Andrew Grimm Jun 24 '10 at 5:33
  • Related question: meta.stackexchange.com/questions/49012/… – Andrew Grimm Oct 21 '11 at 2:39
  • 2
    One of the bigger failures of SO I've seen. And it's status-bydesign. Unbelievable. Maybe now, when Jeff is not part of the team someone could reconsider this problem... – Piotr Dobrogost Feb 9 '12 at 18:32
  • check out stackse – ren Mar 25 '16 at 12:40
  • @Mottie, please remove the "accepted" answer flag, as the described search feature is not working. – Cœur Feb 24 '18 at 6:44
  • Using code:"===" still works. – Mottie Feb 25 '18 at 16:23

10 Answers 10


As part of the latest search engine changes, you can now search for symbols just by using quotes around your term or phrase, like this:

|improve this answer|||||

As a manual solution add a tag with the prefix non-word- from this list:

&  (ampersand)
*  (asterisk)
@  (at-sign)
\  (backslash)
`  (backtick)
!  (bang)
{} (braces)
[] (brackets)
^  (caret)
:  (colon)
,  (comma)
$  (dollar-sign)
.  (dot)
"  (double-quote)
=  (equal-sign)
-  (hyphen)
<  (left-angle-bracket)
() (parentheses)
%  (percent)
+  (plus-sign)
?  (question-mark)
>  (right-angle-bracket)
;  (semi-colon)
#  (sharp)
'  (single-quote)
/  (slash)
~  (tilde)
_  (underscore)
|  (vertical-bar)

The Camel Book Index (Programming Perl) is used for the terminology. For example, tag non-word-tilde-tilde would be used for "~~". This tag would be added to SO question, What does “ ~~ ” mean in Perl?.

Using a tag would allow a person to determine that the dot at the end of this sentence does not get a non-word-dot tag.

Rather than placing tags on an entry simply placing these standard strings in the text would allow the existing search system to match at the author's discretion.

|improve this answer|||||
  • 3
    can't say I'm a fan of this, and in fact, I'm actually kind of .. heavily against it, looking at your example: stackoverflow.com/questions/3094916/what-does-mean-in-perl it's very noisy. The SymbolHound option in another answer, below, seems like a much more rational choice. – Jeff Atwood Feb 2 '12 at 17:42
  • 1
    At first glance this looks so "promising" - but crashes and burns with ferocious aplomb - as clarification to such recommendations as "add a tag with the prefix non-word- " never materialize, despite the cute list of words, or are they non-words... and which is which? The rest of it makes me glad I don't live on a Perl monastery... but please, shed some light on this point if theres a ray left... but this has got like, mad cobwebs, yo. – alex gray Feb 4 '12 at 21:37

It is possible; but not on Google.

A search engine I helped develop, SymbolHound, allows you to search for special characters that are ignored by Google and other engines.

Here's a SymbolHound search for "!!":

|improve this answer|||||
  • This got me thinking. How many times in a row are symbols repeated (for drawing a histogram)? I just had fun doing some analysis with this. Unfortunately, I'm getting still more interesting HTTP 500 errors once I'm searching for strings of certain lengths. 163 + symbols in a row breaks (URL of 515 characters), | fails at 217, * fails at 163, = at 325. & works right through until 2725 where it fails with HTTP 414 Request-URI Too Large (the upper limit I had expected). Would you like to fix it, please? :-) – Chris Morgan Nov 24 '11 at 9:51
  • (I should specify that this was with using Python with urllib2.urlopen and urllib2.quote. I don't seem to run across these problems using it in Firefox. Not sure what's going on there, then.) – Chris Morgan Nov 24 '11 at 11:01
  • @ChrisMorgan I am a co-founder of SymbolHound (tcf is the other). Thanks for pointing this out, I believe I have fixed it. Let me know if it still doesn't work. – dncrane Nov 24 '11 at 19:58
  • @dncrane: correct. All working now. Thanks :-) – Chris Morgan Nov 25 '11 at 2:58
  • 1
    I really thought SO had this essential capability "pulled together"... as they say... but to the site's credit, I have often been Able to find answers here when other engines came up empty-handed or worse, dumbfounded.. but here I am, like it seems many others were as well... Lured here by the creepy calls of those darn !!'s'! I commend you for having at least an attempt at a valid and useful solution. Just a note... I truly know almost zilch about them... but does Wolfram-Alpha do this kind of thing? – alex gray Feb 4 '12 at 21:53

Punctuation marks are normally not indexed. They do not convey information and are used much too often to produce any useful search result.

So in this case Jeff could think about, if he wants to make typical programming constructs searchable (!!, !=, ==, ===, ...). I think a proposal for this already exists, but I can't find it for some weird reasons ;)

|improve this answer|||||
  • 6
    Unfortunately, sometimes they DO convey information. Try searching for the MS-Access #DELETE problem. I'm sure there's a solution out there, but there appears to be no way to find it. – thursdaysgeek Feb 15 '11 at 0:19
  • 1
    I DO think this should be written in the FAQ about how to write titles. It's nearly impossible to find Q/A about & | ^ and similar characters. – xanatos Sep 13 '11 at 4:58

Those with 2k rep and the ability to edit posts should go and verbosify the question text to explain the characters used.

Launch off with the list mentioned by C.W. Holeman II and make it accessible to everyone and anyone.

For example:

What does "!!" operator mean in javascript?

Gets superpowered for reading with an edit such as:

What does the !! operator (double exclamation point) mean in JavaScript?

Another one that would benefit would be something like Difference between || and 'or' in Ruby. Some people call || pipes. Others call it the vertical bars.

Doesn't matter what you drop, so long as you squeeze something out and put it into the bowl. Never hurts to explain those characters you're seeing in English words to help flesh out punctuation marks.

Helps the Google drifters and the users trying out the site search for the first time.

|improve this answer|||||

Nick Craver commented here that this feature is currently in testing as a Advanced Super Ninja Search Option, code. So to search for !! in code, you run the query code:"!!"

|improve this answer|||||

If you can't search for it in Google, the odds of us being able to search for it are vanishingly slim...

Well, I can search for !! on Google code search:

http://www.google.com/codesearch?q=!!&hl=en&btnG=Search+Code (discontinued March 2013)

Ability to search for programming punctuation would be an obvious improvement to this site.

Note also that Google does match SOME punctuation characters - C++ is the example I'm most familiar with.

|improve this answer|||||

Try searching for "double exclamation mark" or something like that. Someone writing about this has the reverse problem: how to be found.

this was the 4th link I found.

|improve this answer|||||

Add another feature to markdown for questions and answers wherein the author identifies character sequences as being indexed and searchable. CSV within braces, {}, would be used to define sequences such that their occurance in the article's code segments would cause the instances to be indexed and searchable.

This markdown would not be displayed but would index the equal-sign, plus-plus and semi-colons in the code fragment:


my i = 0; i++;

Then a later markdown would change the sequences to tilde-tilde and at-sign.


my @results = grep {match_test($_)} @list;

This was provoked by SO answer, @array ~~ $scalar is true when $scalar is in @array.

Then in non-code sections double braces, {{}}, around any text instances would indicate the enclosed sequence is to be indexed and searchable. This sentence is making a made up reference to the {{~~}} in the above Perl fragment. The double braces will cause the tilde-tilde to be indexed and searchable.

|improve this answer|||||
  • 2
    Nice idea. Better yet just fix the search! :) – Piotr Dobrogost Feb 9 '12 at 19:21

If you can't search for it in Google, the odds of us being able to search for it are vanishingly slim...

|improve this answer|||||
  • 31
    Not necessarily - it would be madness for Google to index things like != or !!, but because SO is targeted specially at programming content SO could index a specific list of programming related constructs and actually have it work successfully. Probably not worth the effort, but still - it could work! :-p – Justin Dec 14 '09 at 19:24
  • 2
    However, as maddening itself... as the madness of indexing "such things"… Is the fact that they already DO index them... and yet provide no ways or means, even in some dorky / hideously draconian way... of accessing those toilet-bound "results". it's hysterical "simplicity will save everything" marketing idiots - that think they'll be the next Steve Jobs if they strip enough features off of whatever bullsh*t product they're working on.. instead of implementing solutions that are unobtrusive and adaptable to various tastes and skills of those for which they were designed. – alex gray Feb 4 '12 at 22:05

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .