Support NeoGAF

[Squirrel Killer]

Are you talking about the honeypots?

The 100 honeypots that Google manually generated, then fed to Bing using a Microsoft tool that specifically disclaims that it will use internet behavior to improve Bing searches, resulting in 7-9 pickups? Yes, those honeypots.

 

[dLMN8R]

To clarify, none of this needs to be specific to Google search results.

The result Google is seeing in their "research" can easily be replicated by simple, generic pattern recognition applicable to the entire internet is all that is required. A pattern/algorithm something like this is all the Bing Toolbar needs to recognize:

1) User enters term <X> into web form
2) Entering term <X> --> URL #1
3) Click URL #2 on page displayed by URL #1
4) Navigates to URL #2

If <X>, URL #1, and URL #2 match an overwhelming number of times, then it's quite likely that <X> is a popular search term, and URL #2 is the desired search result. So the relevance of URL #2 is increased when searching for term <X>.



Once again, the fact that Google only saw this happening in roughly 7 of the 100 terms they tried is a blatant admission by Google that Bing is in fact not copying Google's results.

 

[dLMN8R]

The 100 honeypots that Google manually generated, then fed to Bing using a Microsoft tool that specifically disclaims that it will use internet behavior to improve Bing searches, resulting in 7-9 pickups? Yes, those honeypots.

This - the key point that most are missing was that people at Google were actually clicking results, while using the Bing Toolbar, and deliberately opted-in to anonymous collection of user behavior metrics.

This was not some passive collection of search results, nor does it have to be a special case oriented directly at Google.

 

[loosus]

Google is one to talk about "cheating." Sounds like YouTube and their book-archiving service. They can go fuck themselves.

 

[ThatObviousUser]

The 100 honeypots that Google manually generated, then fed to Bing using a Microsoft tool that specifically disclaims that it will use internet behavior to improve Bing searches, resulting in 7-9 pickups? Yes, those honeypots.

The honeypots that Google specifically said would de-activate if the search for them ever got even remotely popular? The honeypots no normal user would ever search for? And the same honeypots that are at this point de-activated?

 

[notworksafe]

The honeypots that Google specifically said would de-activate if the search for them ever got even remotely popular? The honeypots no normal user would ever search for? And the same honeypots that are at this point de-activated?

Not sure how your points here are relevant. Google employees purposely fed bad info to a tool used by MS to grab search terms and results to put on Bing then complained when the tool grabbed search terms and results.

Were they surprised that this happened? They must have been to put up a blog post that portrays the "copying" as passive.

 

[ThatObviousUser]

Not sure how your points here are relevant. Google employees purposely fed bad info to a tool used by MS to grab search terms and results to put on Bing then complained when the tool grabbed search terms and results.

Were they surprised that this happened? They must have been to put up a blog post that portrays the "copying" as passive.

They were testing a theory. And lo and behold, Bing is scraping Google.

 

[notworksafe]

I think you are misreading. The Google employees clicked all the links that ended up ranked in Bing, which is what the Bing toolbar is designed to do (and is disclosed when installing it). All the results that appeared on Bing were clicked by the employees.

That isn't scraping. That's using what is meant to be a tool to rank certain search results higher based on number of clicks by query for it's intended purpose and having it work well. Scraping would be using Google to search for a term and then taking every result no matter if it was clicked or not.

EDIT: Plus this "copying" thing is crap. Google doesn't have a problem when they "copy" other sites, yet seem greatly injured when 7 out of 100 search queries have "copied" results when their own engineers actively clicked links to get them ranked higher in Bing.

 

[venne]

Google is an advertising company first and foremost. One of the best ways to promote your product is to attack the competition.

Must. Protect. Search. Revenue.

 

[dLMN8R]

They were testing a theory. And lo and behold, Bing is scraping Google.

As I and others have already pointed out, the results of their experiment do not logically imply only their stated conclusions. They are saying that Bing is specifically paying attention to Google, and specifically reacting to what Google (and only Google) does. On the contrary, very generic pattern recognition totally unrelated to Google could easily yield the exact same results to their experiment.

The performed bad science, plain and simple. They went into an experiment hoping they'd reach a desired outcome, they get their desired outcome, even though they're deliberately ignoring everything else along the way.


And as I had also already pointed out, the results of their experiment in fact work against their status conclusion because it only "worked" 7 out of 100 times, even with all other things considered.

 

[notworksafe]

Must. Protect. Ad. Revenue.

There we go.

Google doesn't give a shit what search engine you use. They just want you using the internet, period. Google ads and ads from Google-owned ad agencies are everywhere. That's where their money comes from.

 

[dLMN8R]

Google isn't one to talk about copying anyway.

http://www.businessinsider.com/google-redesign-actually-mimicks-bing-2010-5
http://techcrunch.com/2010/06/09/go...ts-homepage-by-default-at-least-for-24-hours/
http://strongbeer.blogspot.com/2009/03/create-flowcharts-diagrams-in-google.html

 

[Squirrel Killer]

The honeypots that Google specifically said would de-activate if the search for them ever got even remotely popular? The honeypots no normal user would ever search for? And the same honeypots that are at this point de-activated?

Your point? Yes, the honeypots have been de-activated, but Google still used the honeypots to feed Bing incorrect search signals. In fact, these honeypots most likely (can't say for sure since Bing, like Google, doesn't release algo details) exaggerated whatever influence Bing gives the toolbar's behavior data since there are no other signals for the honeypot terms.

They were testing a theory. And lo and behold, Bing is scraping Google.

That's not scraping. Scraping is going to a site and downloading everything. Microsoft is using user supplied internet behavior to discover new content.

 

[Mudkips]

They were testing a theory. And lo and behold, Bing is scraping Google.

No. The experiment does not show that at all.

The experiment shows that IE and the Bing toolbar feed information to Bing. This is blatantly stated in the EULA for both. This is extremely common behavior. You can turn the behavior off with IE, and probably with the Bing toolbar. You can always uninstall the Bing toolbar.

I don't see how you keep dancing around the simple facts of the issue. When Google uses the term "honeypot" you seem to think that Bing is going out and being a baaaaad widdle boy by poking around where they shouldn't be. What is actually happening is Google is actively poisoning Bing search results. When a search term has nearly zero hits, any data you pick up for it is going to be highly ranked. Google poisoned Bing results for a period of weeks, and still only managed to get a 7-9% "success" rate.

This is equivalent to someone putting 100 magazines in your mailbox every day for 2 weeks, and then accusing you of stealing 7-9 of them, with the evidence being "He knew what these magazines were about!!!".

 

[ThatObviousUser]

I also want to say it's very likely we're not getting the whole picture here, this was only made known today after all. I am going to wait a week or two before really deciding what I think of the whole situation, but right now my thoughts boil down to this:

- Google shouldn't have used such a sensational title for their blog post
- Microsoft should stop using Google's results for Bing, no matter how many hoops they're currently jumping through to procure that data
- Both companies should stop acting like babies on Twitter

 

[Jamesfrom818]

Google isn't one to talk about copying anyway.

http://www.businessinsider.com/google-redesign-actually-mimicks-bing-2010-5
http://techcrunch.com/2010/06/09/go...ts-homepage-by-default-at-least-for-24-hours/
http://strongbeer.blogspot.com/2009/03/create-flowcharts-diagrams-in-google.html

You can copy arrows and parallelograms?

 

[dLMN8R]

I also want to say it's very likely we're not getting the whole picture here, this was only made known today after all. I am going to wait a week or two before really deciding what I think of the whole situation, but right now my thoughts boil down to this:

- Google shouldn't have used such a sensational title for their blog post
- Microsoft should stop using Google's results for Bing, no matter how many hoops they're currently jumping through to procure that data
- Both companies should stop acting like babies on Twitter

Right now they are not doing anything specific to Google. They are not special-casing Google in any way.

At least, absolutely none of the evidence that Google has provided remotely implies anything of the sort.


Instead, are you suggesting that Microsoft should special-case their tools to actively avoid google while continuing to collect anonymous data from everywhere else as they do today?

 

[ThatObviousUser]

Right now they are not doing anything specific to Google. They are not special-casing Google in any way.

At least, absolutely none of the evidence that Google has provided remotely implies anything of the sort.


Instead, are you suggesting that Microsoft should special-case their tools to actively avoid google while continuing to collect anonymous data from everywhere else as they do today?

Yes. Leave out clickthrough data on all search engines. Or just not apply the data at all, come up with a competent search algorithm instead.

 

[Axion22]

Because it has better search results? Fantastic reasoning.

Oh look, sarcasm. I honestly didn't expect to see that here.

The reasoning is if it's not any different, in fact it's the same on purpose, there's no point in switching from what I already use, not that you were interested in any actual discussion.

Have a nice day.

 

[dLMN8R]

Yes. Leave out clickthrough data on all search engines. Or just not apply the data at all, come up with a competent search algorithm instead.

Looking at results relevancy lately, sure seems like Microsoft's practice of anonymously tracking user behavior is giving them the edge in returning results that people like.

 

[Medalion]

I could swear Yahoo was like this too when Google was starting to dominate

 

[ThatObviousUser]

Looking at results relevancy lately, sure seems like Microsoft's practice of anonymously tracking user behavior is giving them the edge in returning results that people like.

Cheating gets you great results in life too sometimes.

 

[notworksafe]

Cheating gets you great results in life too sometimes.

Lord. Step out of the fanboy zone please. Google tracks user behavior through Chrome, their toolbar, and Android as well. Not sure how checking to see what links people click when searching counts as cheating anyway.

 

[JBuccCP]

So how many people who wouldn't have bothered before are now going to try Bing, if only to see how it compares to Google's results? Along with their accusation apparently not being true, Google are looking dumb here.

 

[BakedPigeon]

I don't see anything wrong with this, although I still us Google instead of Bing. I'm sure Google has borrowed ideas from other companies at some point in their company's lifespan.

 

[antipode]

Right now they are not doing anything specific to Google. They are not special-casing Google in any way.

At least, absolutely none of the evidence that Google has provided remotely implies anything of the sort.


Instead, are you suggesting that Microsoft should special-case their tools to actively avoid google while continuing to collect anonymous data from everywhere else as they do today?

Isn't it common sense that they have to be special-casing Google?

I mean, if they just used any terms anyone typed into any webform Bing would expose all sorts of personal information. Like if you typed your name into a Biology class form and 2 clicks later were on an article about Chlamydia, and Bing then associated your name with Chlamydia, you would be pretty pissed.

 

[BlazingDarkness]

I will never use a search engine called 'Bing'
Terrible name

 

[dLMN8R]

Cheating gets you great results in life too sometimes.

If it's "cheating" to look at completely anonymous data that probably has jack shit to do with Google, rather just simply generically observing generic usage patterns, then fuck why the hell aren't more people cheating?

It just seems like Google is jealous that Microsoft was more creative in exploiting their opt-in anonymous data.

Isn't it common sense that they have to be special-casing Google?

I mean, if they just used any terms anyone typed into any webform Bing would expose all sorts of personal information. Like if you typed your name into a Biology class form and 2 clicks later were on an article about Chlamydia, and Bing then associated your name with Chlamydia, you would be pretty pissed.

They're doubtlessly special-casing *something* to account for problems like what you describe, but the experiment they performed still doesn't imply anything along the lines of Microsoft specially paying attention to Google any more than they would any other search engine or other form of user input on web sites.

 

[dLMN8R]

(dp)

 

[Jin34]

Bing Video is still better for Porn videos though, someone is putting some great work in there.

Well done person in dark room! Well done

(Tries it out) Wow I had no idea... It is great.

 

[06nbarnhill]

Only use google.....never used Bing. Never will.

(some days when I am bored....I go through all the computers in the computer lab I work at and make sure they are all defaulted to firefox/google XD )

 

[notworksafe]

(some days when I am bored....I go through all the computers in the computer lab I work at and make sure they are all defaulted to firefox/google XD )

I feel bad for your users.

 

[Chichikov]

.
(mispress)

 

[mugwhump]

Bing Video is still better for Porn videos though, someone is putting some great work in there.

Well done person in dark room! Well done

whoa damn, you're right

I now have a reason to use Bing

 

[Squirrel Killer]

Yes. Leave out clickthrough data on all search engines. Or just not apply the data at all, come up with a competent search algorithm instead.

You do realize that Google freely admits that they use clickthrough data for UX engineering, it's a huge part of AdWords, and is widely believed to use clickthrough data as one signal out of many for it's own search algo, don't you? Clickthrough data is a part of a competent search algo.

 

[sprsk]

Google's just mad.

I find myself using Bing more and more these days. Now I guess I know why

Because you are using Google?

 

[Sydle]

Right, but I think each engine makes its own title/description when it indexes a page.

No, the title and meta-description tags, among many others, are standard tags used for indexing that information by pretty much every search engine for years and years (i.e. it was never unique to Google). If those tags are not available then the search engine will use other information on the page to populate the title and description, which typically consists of a headline for the title and the first 150 or so characters of the text following the headline.

You do realize that Google freely admits that they use clickthrough data for UX engineering, it's a huge part of AdWords, and is widely believed to use clickthrough data as one signal out of many for it's own search algo, don't you? Clickthrough data is a part of a competent search algo.

Click through data is a huge part of AdWords since it's the most influential factor in quality score of the ads (it's a signal of relevancy) which helps decrease minimum bids and increase ad rank relative to the competitive landscape. An advertiser wouldn't want to game this aspect of it because they would have to pay per click to help themselves out, which would ultimately hurt them financially and they would spend themselves into oblivion (especially in highly competitive keyword markets; some of my keywords are $30 per click!).

Organic results use other signals for ranking including keyword density on page, cross links, and inbound links. Inbound links are the most helpful. Clicks may be a small part if they're anything at all, because that could be easily gamed. There are supposedly hundreds of factors that go into the search algo, but there are well documented methods to gain ranks, and they're true from my own experience of doing this for three years now, if you optimize the on-page elements, use anchor text well when cross linking, and you have a healthy amount and variety of inbound links.

 

[notworksafe]

Because you are using Google?

Glad to see someone is pulling the good results out of Google (and other search engines) and not placing content farm garbage on the first page. Google should figure out how to do that instead of whining about Bing.

 

[WrathOfOtaibah]

I really dislike Bing except for their fantastic maps. Which is sometimes essential for my job rather than wading in 2+ feet in snow.

 

[Zhengi]

Because you are using Google?

 

[Squirrel Killer]

Click through data is a huge part of AdWords since it's the most influential factor in quality score of the ads (it's a signal of relevancy) which helps decrease minimum bids and increase ad rank relative to the competitive landscape. An advertiser wouldn't want to game this aspect of it because they would have to pay per click to help themselves out, which would ultimately hurt them financially and they would spend themselves into oblivion (especially in highly competitive keyword markets; some of my keywords are $30 per click!).

Organic results use other signals for ranking including keyword density on page, cross links, and inbound links. Inbound links are the most helpful. Clicks may be a small part if they're anything at all, because that could be easily gamed. There are supposedly hundreds of factors that go into the search algo, but there are well documented methods to gain ranks, and they're true from my own experience of doing this for three years now, if you optimize the on-page elements, use anchor text well when cross linking, and you have a healthy amount and variety of inbound links.

My point was that Andrex's proposed blanket ban on clickthrough data is pretty laughable considering how much Google freely admits they use it in UX and advertising and how widely they are thought to be using it as one of many relevancy indicators for organic results. You're right that it's probably not a significant factor in the algo, but I'd be really surprised if Google left that kind of behavioral feedback on the table.

 

[felipeko]

I think people are missing the point.

The problem with this is not that Microsoft is using "signals" that come from people using Google, it's the weight they are giving it.

When Google, with 20 engineers, can fool Bing's engine, there's something really wrong with it. Doesn't matter how many clicks Microsoft got on the pages, they had NO RELEVANCE to the query, and they were there just because Bing's engine was actively just gathering data from Google users without even checking it's relevance.

That may not be unlawful or evil, but it is plain sad to se Microsoft doing that.

 

[turnbuckle]

I'm not really sure that's from tracking cookies so much as just referencing extremely common searches from the aggregate of users. I just tried typing "what do I need..." as well, and got the same result. I'm guessing that P90X is just an extremely popular topic right now, as is the 3DS.

Nah,

I'm on a different computer now and just typed "what do I need" and got:

"what do I need to vote"
"what do I need to vote in Michigan"
"what do I need to take the CPA exam"
"what do I need during a snowstorm"
"what do I need to survive an economic collapse"

 

[jvalioli]

I see lots of non-spellingchecking Google engineers talking about this on Facebook right now.

 

[MidgarBlowedUp]

Yeah cause we all know Google's image search is like an Oracle.

 

[BloodySinner]

I hope Bing continues to grow, so we can see more whining from Google...

Doesn't bother me. I use Google almost exclusively. They can complain all they want.

 

[Sydle]

My point was that Andrex's proposed blanket ban on clickthrough data is pretty laughable considering how much Google freely admits they use it in UX and advertising and how widely they are thought to be using it as one of many relevancy indicators for organic results. You're right that it's probably not a significant factor in the algo, but I'd be really surprised if Google left that kind of behavioral feedback on the table.

Got it. I'm sure it's used to some extent, but it would be soooo easy to game that I just can't see it being a significant factor. As for leaving behavioral data on the table, I'm sure they're analyzing it to figure out how to leverage it for advertising. There is a current internal debate on what is crossing the line when it comes to behavioral targeting, where many of the higher ups believe targeting someone based on their search and click history is going too far. However, there is an upcoming AdWords product that is essentially the same effect (got a preview last week, not sure how much I can talk about it). Point is Google is not leaving that data on the table, just might not be used in the ranking algo.

Doesn't bother me. I use Google almost exclusively. They can complain all they want.

I don't give a fuck about you or the fanboys. I just think it's funny to see big companies act like children.

 

[Uncle]

Yep! It's usually in 100-200 point increments, but it's nice and they add up over time. Just sign up with the email you use for your XBL account.

http://www.bing.com/rewards/signup/SEGetStarted#fbid=Sn3vOH4QE8T&wom=true

It says they don't ship outside the US, but do they actually send a points card or just email the code?

 

[dLMN8R]

I think people are missing the point.

The problem with this is not that Microsoft is using "signals" that come from people using Google, it's the weight they are giving it.

When Google, with 20 engineers, can fool Bing's engine, there's something really wrong with it. Doesn't matter how many clicks Microsoft got on the pages, they had NO RELEVANCE to the query, and they were there just because Bing's engine was actively just gathering data from Google users without even checking it's relevance.

That may not be unlawful or evil, but it is plain sad to se Microsoft doing that.

The fact that they're using complete gibberish words that have literally never been seen before by either search engine means that there is literally no other possible search result that could potentially be returned.

Bing isn't requiring a certain level of relevancy and then returning the result because it meets that level of relevancy. Bing is assigning it a certain relevancy, and because it is literally the only result with any relevancy whatsoever, the result is returned.


The amount that a small group of 20 engineers could influence a real search term is completely negligible. But a previously non-existent term? Of course it works.

That's also, of course, why it only worked for 7 out of the 100 potential terms.

 

[TheOMan]

They ran Bing Toolbar while doing the searches, and they explicitly agreed to Bing's terms of services which said they can look at the webpages you look at to improve their search engine.

You're missing the fact that they click certain results.

Google does this too. Obviously, if you have a ton of people performing a specific search, and then clicking specific results in that search, that inherently says "this is a good search result that people like", and it should therefore be weighted more heavily as a result.


Far different than what Google is accusing them of doing - simply looking at search term / result sets and copying them exactly.

Oh ho! Thanks - that makes more sense now - sorry Nerevar!