Hacker News new | comments | show | ask | jobs | submit login
Stack Overflow ranks #2 for Google Search for "Stack Overflow" (google.com)
229 points by momchenr on Mar 2, 2013 | hide | past | web | favorite | 121 comments

Looks like it's a cut/paste error. If you do wget www.doioig.gov, this is the page you get. Notice the meta refresh that points to stackoverflow.com.

  <html lang="en-US">
      <meta charset="UTF-8">
      <meta http-equiv="refresh" content="1;url=http://stackoverflow.com">
      <script language="javascript">
          window.location.href = "http://www.doi.gov/oig/index.cfm"
      <title>Page Redirection</title>
  If you are not redirected automatically, please click the link to continue to the <a href='http://www.doi.gov/oig/index.cfm'>U.S. Department of the Interior Office of Inspector General.</a>

And here's why: http://stackoverflow.com/a/5411601/620239. Forgot to change that first occurrence after pasting it in. "Oh, it works. I'm done."

It has been edited.

Here are the past revisions: http://stackoverflow.com/posts/5411601/revisions

I'm actually glad it's been edited. There's an RFC for this, people!!

That's brilliant, surprised that Google managed to rank them so highly because of it though!

.gov domains have a ton of link juice.

Another good reason why you should always use "example.com".

I edited the answer so it says example.com now.

I'm surprised http://example.com is reserved by IANA.

It's reserved per RFC 2606: http://tools.ietf.org/html/rfc2606

...why is that surprising?

Because it makes sense. These sorts of things normal are allowed to go through so that horrible unintended consequences can be wrought on the unsuspecting.

Unless, you know, you want to be in the #1 spot on a Google search.

I bet there's some sort of logic on google's side that states, if a government homepage url redirects to another non .gov URL, that the .gov address should be the canonical page of the site thats being redirected too. Which would explain why stackoveflow's homepage isn't listed in the results. Just a guess.

I could imagine the ranking favours .gov domains, which should be reliable. But surely for anything Google sees as a redirect, the redirect target should be the canonical URL?

>But surely for anything Google sees as a redirect, the redirect target should be the canonical URL?

That would only be the case for a 301 redirect.

Why would this rank them so high though? Does the googlebot actually follow the http-equiv="refresh". If so, would changing http://stackoverflow.com to say http://dell.com make them rank #1 for searching for dell as well?

That sounds too easy. Also, if doioig.gov redirects to stackoverflow.com that just makes doioig.gov an alias for Stack Overflow. An alias shouldn't rank higher than the real thing.

There is no rational reason. I said as much and got downvoted by someone.

Glad to know someone who works for the government at least knows where to get their information.

Given that the Department of Interior has been forced to take the whole department (expect for vital services) offline multiple times, I would not be surprised if it were hacked. I am hoping this doesn't get in front of a judge anytime soon as it can have some consequence for people caught in the way.

One such consequence, at one point a judge (curse his or her soul) decided since the DOI needed to be off the internet then all "affiliates" needed to be off the internet. This includes the BIA (Bureau of Indian Affairs). Which included both the .gov and .edu domain. At the time many Tribally chartered Community Colleges[1] were told to disconnect from the internet mid-semester. Even those colleges who paid for their own internet connection and had a .edu domain of their own.

Imagine having two weeks with no internet (most of our students don't have home internet) with classes going on. Finally, someone got the order rescinded for the schools.

I am not very fond of how the DOI handles its internet[2][3].

1) accredited just like state or private colleges with transferable classes.

2) don't even get me started about sending mail from a subdomain with no DNS entry for the sending mail server or subdomain and expecting us to not reject it.

3) http://www.doi.gov/archive/news/08_News_Releases/080523a.htm...

Wait what? Why does the DOI disconnect from the internet on a regular basis? I feel like there's a backstory here.

So, apparently in 2001 a class action lawsuit accused the Department of the Interior of mismanaging Indian trust accounts. As part of that lawsuit court appointed hackers broke into DOI computers. The judge ordered all DOI computers to be disconnected from the internet. Some parts of the DOI came back online quickly, but others remained disconnected for over 6 years. Employee desktops couldn't even access the internet, so they could only send emails within their intranet.



Google the Cobell v. Norton lawsuit. It basically is a case study in how not to do internet anything along with some interesting physical site security problems.

Google's algorithm has gotten so big and so complicated over the years, that there are so many cracks and special cases that can cause sites to disappear from or be poorly ranked in search results, unless you're lucky enough to be huge in the tech scene or post here and get your comment seen by a Googler (as I have on occasion).

<plea>Any Googlers reading this, I'm looking into rebuttals of false DMCA requests being ignored by Google for months...</plea>

I don't think that a simple algorithm would lead to less poorly ranked search results.

Also, the internet has gotten bigger, so the possibilities for cracks or errors have widened greatly.

That being said, I do think this specific failure has no reasonable explanation. No matter how I think about it, or look at it, the only explanation I can come up with is essentially: "What percentage of people would need to see a wrong version of a search result before it is reported" or "How long would it take us to fix a knowingly wrong result".

"Google's algorithm has gotten so big and so complicated over the years". Agreed! Just like the US tax code...

Agreed. A hacker put some malicious code on one of my sites over a year ago and it was quickly buried in search results after ranking 1-3 on page one for years. I fixed it, changed hosts, notified Google and have tried everything else I can think of and it's still nowhere to be seen - even though it ranks near the top of both Yahoo and Bing.

If a Googler sees this, the site is http://starbuckslocations.com - please look into it!

Conversation from three days ago at http://meta.stackoverflow.com/questions/169405/google-indexi... suggests that the best hypothesis is that at some point in the past, www.doioig.gov was compromised and maybe redirected to StackOverflow.

(This hypothesis is supported by a Google cache of doioig.gov showing the message "Due to security concerns, our website will be unavailable until transition to the Department of the Interior web domain occurs. We apologize for any inconvenience this may cause, and are working to speed up the transition. The following contact information is provided to assist you.")

(Actually the explanation at http://news.ycombinator.com/item?id=5311430 looks a lot better. I do wonder what's up with the DoI's recent "security concerns").

They have a rather long history of security concerns. Many times ending up in front of a judge.

Even if it had been redirected, why would it outrank the actual site?

Google obviously puts high trust in this domain as it has many high trust links, typical for .gov domains. In this case this seems to have been enough for the algo to assume, that doioig.gov is the more important domain and should replace StackOverflow.

Everyone seems to be focusing on the .gov site but if you take a look at the Stack Overflow home page, the page that would normally get indexed highly, there is very little telling Google what it is.

There is no <meta name="description"> tag in the header. The H1 tag, important to Google, says "Top Questions". The content of the first page constantly changes.

Plus, I would bet that most of the links into Stack Overflow are to individual articles, not the home page. Any particular article probably doesn't outrank a popular .gov site.

This is just very poor SEO on Stack Overflow's part.

I disagree it's necessarily poor SEO. They should do everything they can to ensure that pages that answer questions have better pageranks than their largely useless homepage. That's what gives the name stackoverflow power, questions actually getting answered.

Certainly the individual questions and answers should rank highly. They do a somewhat better job at SEO on those pages, although there still is no meta description tag, leaving Google to guess what to put on the search results page.

They could do a lot better on those pages too. The real content is buried below lots of javascript and other code. The higher on the page your real content is, the better. They do have pretty good titles and H1 tags on those pages, and the urls are okay, although I would move the name of the article up at least a level. Here is one: /questions/15181744/twitter-number-of-tweets-not-updating-testing-on-local

But the original topic here was a search for "stack overflow" and for that search there could be no better page than the Home page (well, maybe the About page, although if they did their job better they would get one of those results with lots of sub-section links). There was no real attempt to optimize that page for search engines. Optimizing it for the keyword "stack overflow" isn't going to hurt the rank of those question/answer pages.

Too good, even. Google has a bad habit of ranking a question page lower (for matching keywords) than some random other (more popular) question page that links to the relevant question page.

Not sure I disagree with SEO on "Stack Overflow" who is googling for that term?

Practically any technical question I've searched for has resulted in a stack overflow (or super user) #1 result. I'd say they are doing a pretty good job, even if they don't do all the old school SEO stuff

According to the Google Adwords tool, "stack overflow" is googled 74,000 times each month. Not to mention all the permutations.

If I worked for Stack Overflow, I would also try to optimize for more general terms. The questions and answers are good at getting those long-tail keywords, but not so good at the more generic ones like "coding help" or "javascript help" or a hundred others. Stack Overflow isn't on the first page for either of those examples, or many others.

Brand Searches drive a lot of traffic to major sites people are lazy and use Google instead of bookmarks

isn't that too long just for


I've often wondered what all those extra bits of information mean and whether there's information encoded in there that I may not want to give someone.

If it has to be short, I like http://goo.gl/e#q=stack+overflow

That link sadly does not work for me who uses iGoogle (for now) on google.co.uk.

If it's a short URL you want, then http://bit.ly/YUJZLu. But you'll have to trust me when you click that as you've no idea where it'll take you to.

Right click and copy link, paste into address bar, add a + to the end to see where it goes.


And that doesn't work on mobile, because they have a "forgetful" mobile site :(

Google results can dramatically change based on various parameters (location/time/browser/history/whatever-new-they-are-testing.) They presumably cram all this info into the search bar in case you want to repro the results.

I'm suspiciously seeing, "2,643 people +1'd this."

While I would be excited to see such enthusiasm relating to a government property, this doesn't smell kosher.

For me it shows that one of my friends +1'd it. I'm pretty sure he +1'd the real Stack Overflow though.

That is worth verifying, as it would be an interesting bug in how Google transfers juice between redirects.

My thoughts exactly.

Stack Overflow ranks #1 for Bing Search for "Stack Overflow".

People actually use Bing?

Yes. And DuckDuckGo. And Google.

Why give all your searches to one search engine and one company?

Depending on what you're searching for, Google doesn't always provide the best results.

Google almost universally provides the best results if you are actually searching for something.

Compare the results for queries like "that movie where a computer plays tic tac toe" A very reasonable search if you forget what the name of War Games is, but bing fails it utterly and completely; not a single mention of War Games until the second results page, and the wikipedia page for it doesn't appear until the third. Meanwhile on google the first results page barely has anything that isn't about War Games.

Now, if you just search "War Games" both will do fine. For that matter, so does Wikipedia's builtin search...

You can push it even further and get more vague, something like "that car that james may goes fast in" and while at that point google starts to degrade, it still easily beats out bing.

Suppose I am looking for "that movie with a button". Yup, google get's it, "The Box". Bing thinks I am thinking of Benjamin Button, which was google's second suggestion (I wasn't). Fair enough though, suppose I actually had been thinking about "that movie where the guy get's younger". Both google and bing don't do great, though google still definitively does better.

> Suppose I am looking for "that movie with a button". Yup, google get's it, "The Box". Bing thinks I am thinking of Benjamin Button, which was google's second suggestion

Hm. There is no one google. In the google that I see, for that search, it's the other way around: "Benjamin Button" is #1, and "The Box" is #2 (both on IMDB).

Life before search engines was different and it's easy to be blasé. IMHO, getting the one that you are vaguely thinking of anywhere in the top 5 is a great and amazing technical feat. And also good enough to jog your memory, so #1 or #2 makes no difference.

How do you allocate your searches? Does the better quality of results justify the inconvenience of using several engines or is there a mechanism I'm unaware of?

Well, most of the time I already know the site best suited for my queries (e.g., SO for programming, amazon for shopping, wikipedia for facts, Wolfram Alpha for calculations etc), so DuckDuckGo's bang syntax really comes in handy. Google is astonishingly good at parsing vague queries, but otherwise DDG or Bing are pretty much OK.

I've tried using Bing several times in the past just to give them a fair chance, but every time came back to Google. They are really not at par in the search business. Actually no other company except Google got hold of the nerve of internet searching.

It's hard to know which is going to give you better results ahead of time. That's why I tend to use google. I discovered this whole thing by accident.

Well, here's an anecdote: when searching for "stack overflow", the first result is more relevant on Bing than on google. So yeah, a few people will go with that one.

I interned at Microsoft so I've been using Bing since last May. Only really have issues with super obscure error messages that I encounter, since Google often indexes the source code that generates the error messages, and Bing just has nothing.

I was lured by Bing Rewards. I plan to use it as my default for a month. So far, I'm content.

I earned my $5 Amazon gift card and then switched back to google. I'm using DDG now and I'm very happy.

Some do. I prefer AltaVista myself.

The other day I used Bing to search for "C++14 ...", because Google doesn't recognize it as a search term. Though it does understand C++11.

Granted I didn't get many worthwhile hits, but I did get some. Google simply provided me with the results of "C++ 14", which is pretty useless.

Also Joel Spolsky's FB profile shows up in Social on right.

Not sure if they fixed it since this post went live on HN but I am not seeing that.

For me SO is first with six "breakout" links below it, and then wikipedia entry about it.

But I don't allow javascript or cookies on google search which may get me a less filtered result.

Gotta wonder though: should StackOverflow.com rank high for "stack overflow"? After all, a "stack overflow" isn't necessarily related to programming questions. Yes, you can ask questions on StackOverflow.com about stack overflows, but that's missing the point. So if I have a domain name that's a thing, but my site has very little content related to that thing, should I rank high for queries about that thing?

So are you saying people searching for "stack overflow" want the definition more often than the web site stackoverflow.com?

Because that doesn't make sense to me. Perhaps an upcoming engineer once or twice needs the definition in their life. And they will mostly go on to use the site as well.

Lazy people like me often type the approximate of a web site into Google rather than trying to guess the exact url / bookmarking it. And we do this continuously.

I can't see the definition being more popular than the site.

Would a person struggling with a stack overflow actually google "stack overflow", though? Surely only a very fresh programmer would need to google it, and only if he/she knew what the concept was. But here's what an infinite recursion in a C program prints if I run it:

    Segmentation fault: 11
Well, that's C for you. If I try Ruby:

    test.rb:2: stack level too deep (SystemStackError)
Ok, still not the same nomenclature. Python:

    RuntimeError: maximum recursion depth exceeded
Nope. How about Go? ... Actually, an infinitely recursing function in Go never completes on my machine. I wonder why. Perhaps it's not using the stack the way I expect.

If you do the same thing in Java:

    Exception in thread "main" java.lang.StackOverflowError
Ok, there it is.

But if do you get that, would you not google "StackOverflowError", as opposed to "stack overflow"?

(Then again, my Google searches are perhaps uncommonly precise. If a function "foobar()" in library "libfoo" overflowed when processing HTTPS URLs, I would probably google for "foobar stack overflow libfoo https url".)

The Wikipedia entry for stack overflow (the concept) is the fourth hit on a search for "stack overflow". Should be acceptable to a newbie.

> Actually, an infinitely recursing function in Go never completes on my machine.

Maybe your function is tail-recursive? Try to use the return value from the recursive call in a nontrivial way, so that stack storage is necessary to store some local variable.

Apparently Go allocates the stack as a heap structure, as was pointed out in another comment: http://stackoverflow.com/questions/4226964/how-come-go-doesn...

"only a very fresh programmer would need to google it"

It's why this is a top story on Headquarters of Noobs. Not surprising. You mean Lenny Kravitz didn't write American Woman?

What's HQoN?

I remember having seen "stack overflow" when I naïvely tried to allocate a 5MB array on the stack in Visual C++ 2010.

Trying this with GCC gives me a segfault (although I had to go higher up to 8 MB).

    #include <stdio.h>

    int main(void) {
      char x[8388608];
      printf("%ld\n", sizeof(x));
      return 0;

My point was that Visual C++ 2010 gives a different error, but it was running in debug mode.

And my point was that GCC was not as helpful in that case. :-) With the GNU C library you will have to install your own signal handlers and jump through a lot of hoops to get anything more sensible than a "segmentation fault" error. You would think our tools would be a bit more modern by now.

Oh by the way, I'm not sure if it was exactly 5MB, it might actually have been 2MB. This particular case was 2 years ago, now.

Google couldn't care less about some abstract notion of whether results are "correct". They want to provide the results that people are looking for, and that's it. I'd wager that most people searching for "stack overflow" are after the site, not the concept. By putting stackoverflow.com at the top, they are providing most people with the result they seek.

Your site should rank high for queries about that thing if people routinely search for that term while attempting to find your site.

There are more links on pages with the words "stack overflow" near the link, or with the anchor text itself "stack overflow", pointing to stackoverflow.com than to any other page on the web. Lacking additional context, the web teaches Google's algorithm that when someone is talking about "stack overflow", they are referring to stackoverflow.com more often than anything else.

If you made a website called "Nikon Cameras" today but the site had nothing about Nikon Cameras on it, it would not rank well for the search "Nikon Cameras". Other people writing about Nikon Cameras would not link to your site more often than something actually about Nikon Cameras.

May be someone fixed something, but I'm seeing stackoverflow as the top result now for both with or without inverted commas.

Yep, it's ranked #1 for me too, but this could be linked to regional and/or personal data (I'm not based in US).

The preferred .gov page has thousands of google plus likes. This seems like a fascinating example of google plus' terrible impact on google search and perhaps google corp.

It's just wrong, overall, the actual Stackoverflow page was the one that had that many +1s.

What doesn't make sense is why it's ranked beneath a Government site? www.doioig.gov

Plus, the content of the homepage doesn't even contain anything about Stack Overflow.

Google thinks doioig.gov is StackOverflow.com. See what happens when you do a search for info: http://www.google.com/search?q=info:stackoverflow.com

Note what site shows up as the result and for the links for "similar" and "link to".

The power of .gov backlinks.

Something strange happened here. Google displays `stackoverflow.com` content in `doioig.gov` description. For example I can see `careers 2.0` in the description of `doioig.gov` that really doesn't have.

I don't see the Stack Overflow home page at all in my search results. http://imgur.com/3WHKhGv

Second link is the login page

Well actually, "stack overflow" should ideally refer to the programming error. The site being talked about is "StackOverflow", and if that term is queried in any search engine, it should, and will provide the correct result.

Don't be silly, that's just the way their logo is stylized. Their FAQ states it's indeed "Stack Overflow", and their mobile version has a logo with a space.

The Wikipedia description of a stack overflow should probably be the best link and it ranks 5th.


That's pretty subjective. Why would that be the "best" link?

Because it describes what a stack overflow is, rather than being a site named after the thing?

That would be my thinking. Ok so maybe wikipedia doesn't have to be the top result, but I'd rather have results relating to the thing I'm searching for come above sites just named after it.

I am willing to bet the City of London against a house-brick that the vast majority of people searching for 'ford' are not looking for a definition of the river crossing, but rather the motor vehicle company named after the man named after several other generations of men named after such a river crossing.

There's a qualitative difference somewhere there. I'd still prefer the river crossing to come top, but recognise that's just me.

In both of these cases you can get google to do what you prefer by typing "define" in front.

"define ford" shows a dictionary definition of ford before all results, "define stack overflow" shows a link to wikipedia before all results.

Not really, as all the results are then about the definition. Definitions, then popular sites, then whatever else, IMHO, would be quite useful.

Of course with the imprecise and constantly moving nature of language this is not easy.

Maybe people are actually searching for the site, and not the thing that named it?

Oh maybe, and maybe they search for that more often than they search for the thing it's named after, and maybe more people would find it more useful that way.

I'm really just expressing a preference :)

You'll notice that people search for facebook not "face book" when they go there.

Stack Overflow is officially named "Stack Overflow" (with the space)[1], though it's common to refer to it as stackoverflow (no space), of course.

[1] http://stackoverflow.com/about

Their logo has no space, nor does their web address. People typing in "stack overflow" I bet are not looking for the web site. I'm sure google analytics for stackoverflow.com could tell us if it was true.

It's #3 if you use quotes https://www.google.com/search?q=%22Stack+Overflow%22 (wikipedia is #2)

The power of a .gov domain authority in google's algorithms.

Aaaaaaand, looks like it's back to normal. :)


Victory for Bing!

Unfortunately our plan for disrupting the Inspector General market got leaked a bit early.

This may be well the first time in my life I'm able to see a bug in Google Search.

doioig.gov ranks highly for both "stack" and "overflow" separately. That site is about stacks and overflows afterall. Perhaps that's why combining the two gives it superpowers.

It's 3rd on google.co.uk, after www.doioig.gov and Wikipedia

It ranks 1, 2, 3 and 4 for me - in the same time.

Do you have your preferences set to return more than the default 10 results? When set to the max of 100, I get a lot of suboptimal results. It's getting better than it used to be, but the worst case I found had slots 1-70 all occupied by the same site.

I think it there must not be much testing for the non-default settings. I much preferred the earlier interface where results from the same site were grouped, capped at a small number, and there was a "More results from this site" link.

Not any more...

I love how this post points out a problem and fixes it at the same time.

Applications are open for YC Summer 2018

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