You can’t Google “Google+”. Or maybe you can?

Update 2011-07-21:I’m leaving the original post unchanged below, for reference, but I just got an email from Jon Trowbridge saying he gets different search results with “Google” and “Google+”. I do too, but not in the way I would expect, at least for more complex searches. When I first try the search I want, Google suggests just eliminating the term “google+” from the search string entirely:

Google suggests that maybe I don't want to search for 'google+'.

(But that could be because the word “google” is itself treated specially, perhaps because people often needlessly type it in searches? I don’t know; only someone with access to the logs can say for sure.)

Anyway, once I choose the real search…

Tell Google I really mean it.

…this blog post comes up at the top, but no other result on the front page contains the character “+” nor the string “plus”, let alone “google+” or “googleplus”:

These results have a notable paucity of '+' signs.

This screenshot doesn’t show the results below the fold, but I did an in-page search to make sure. I also searched within the second and third hits (i.e., the first two after my own blog post) and they don’t mention “google+” or “googleplus” or anything similar either.

Why is it finding this blog post so accurately and yet nothing else on the same topic? Surely I can’t be the only person mentioning about Debian, Firefox, Iceweasel, and Google+ in the same article. Even if no one else has this browser-compatibility problem (which seems unlikely anyway), I’d expect people writing for other reasons.

But if I do the same thing but just comparing the terms “google” and “google+” as Jon did, without all the other stuff, I get a fair number of results that are clearly about Google+ — well, actually, it’s all about Google+ and the iPhone, but I guess that just shows what’s important on the Internet these days.

That makes me think that it’s not simply the case that when Jon and I add the “+”, we’re getting the old “match this exact string” behavior (which could still lead to different results for the two terms, because Google might treat various misspellings of its name as synonyms). Google really could be indexing “+” signs now, or (more likely?) at least treating “google+” specially when the crawler encounters it.

Color me baffled. Anyone know what’s really going on here?

Original post below.


I’ve got a zillion Google+ invitations. I’m ready to try it out. But I’m getting an error that my browser is “no longer supported” (I’m not sure how a new service in beta testing can say anything is “no longer supported”, but whatever):

My browser coulda been a contender.

(The “Learn More” link above goes to a non-existent page, by the way, so that doesn’t help.)

Note how it lists Firefox as one of the browsers to try instead. The thing is, I’m running Firefox already!

I have a tentative theory. For complicated reasons, the operating system I run, Debian GNU/Linux, repackages Mozilla Firefox under the name “Iceweasel”:

Iceweasel, I mean, Firefox.

When identifying itself to web servers, the browser transmits something like this:

  Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.1.19) \
  Gecko/20110701 Iceweasel/3.5.19 (like Firefox/3.5.19)

The server on the other end could interpret that in a number of ways. It could decide to treat its interlocutor like Firefox 3.5.19… or it could decide the browser is some mysterious beast called Iceweasel, and refuse to serve to it. No doubt somewhere there’s an RFC that spells out what both sides should do, but I have no idea which one if so.

Anyway, I don’t know for sure if this is why Google+ is rejecting my requests. I may test the theory by having my browser impersonate regular Firefox, after I finish this blog post. At the moment, all I know is “It’s Not Working”.

But the worst part is: I can’t Google up an answer, because “+” doesn’t work in Google searches.

You can’t do a Google search based on the presence or absence of a “+” sign. You might get results that contain “+”, of course, but the matching will have been based on the alphanumeric words around it, not on the “+” sign itself. (Google even turns the fact that they don’t index “+” to advantage, offering it as a metacharacter in search strings for suppression of the automated synonym matching that would otherwise happen automatically).

So Googling for debian firefox iceweasel google+ doesn’t work. As far as I know, all the other major search engines are the same way: “+” isn’t indexed, so you can’t search for it. Fine. The reasons for this are technical, having to do with size-versus-completeness tradeoffs in building search indexes.

But then I wish they hadn’t named an important new service in such a way that it can’t be searched for :-).

Update: using “googleplus” as a synonm gets some useful results, I guess because people are sometimes using that spelling when writing for the web. I wonder if that’s because they’re aware of this problem. Of course, people sometimes spell it with a space and sometimes without, which means if you use an actual “+” with your search for “googleplus”, you’ll miss half the results. It’s the irony that keeps on ironing.

5 comments

  1. It could also be that they’ve simply dropped support for FF3.5 because Mozilla no longer supports it either?

  2. Have you tried the latest version of Iceweasel? Firefox is now up to version 5.0.1, so I imagine that there’s a more recent version of Iceweasel as well. If that doesn’t work, you could always try Clearing Google Cookies.

  3. Decided this post was boring, and that the problems may be due to pilot error or to Debian’s ‘testing’ dist Iceweasel not keeping up with latest Firefox. The post is still live, but it’s not on the blog display anymore.

  4. I see neither problem that the author complains about — Google+ works fine in iceweasel, and Googling for “Google+” gives obviously relevant results.

    Perhaps the problem is, as Kennis suggests, not the name of the browser, but the version; the FF one gets from the “download FF” link is version 5.x, and the author was using version 3.5 (I use iceweasel version 5.x from Debian sid).

    As for the search term, I dunno. “C++” used to be annoying for this reason, but it became googleable in the past few years. I don’t know if they made “+” a generally significant search character (except in the prefix position where it actually has meaning as a meta-character), or do it on a case-by-case basis, but if the latter, they’ve clearly moved quickly to add “Google+” to that list!

  5. David, Miles,

    I’m pretty sure at this point it’s my error — thought I’d swear I did this comparison search a couple of weeks ago, and did not see the kind of results I would have expected. However, I was jet lagged and it was a smoky Internet cafe behind the Great Firewall, which was slowing everything down, so it’s quite possible I messed something up.

    Kennis,

    I installed raw Firefox and everything’s fine now :-).

Leave a Reply

Your email address will not be published. Required fields are marked *

Rants.org Comments Policy

two + one =