BING BUSTED! Copying Google Results.

[mbigna]

I'd always wondered if something like this was going on with other search engines as well. It could explain a lot of that extraneous network traffic being transmitted that nobody can ever explain in its entirety.

http://www.pcmag.com/article2/0,2817,2379120,00.asp

http://www.zdnet.com/photos/does-bing-copy-google-search-the-evidence-screenshots/6190648

http://www.computerworld.com/s/article/9207630/Google_accuses_Microsoft_of_copying_search_results

I wonder if this extra traffic could be partially responsible for this:
http://www.overclockers.com/forums/showthread.php?t=667917

Once again, yet another example of NON-innovation at Microsoft. Why bother with Bing at all now--they have pretty much ceded the search market with this revelation.

 

[gregmacknass]

I guess that Microsoft figured that if so many people were stealing their software that they might as well jump on the bandwagon and steal someone's programming. Then they get caught red handed and try to play it off like it's not their fault. Maybe next time they try to sue someone for stealing their software, the accused pulls this out and smacks them over the head with it.
-Greg

 

[Saaby900t]

I fail to see how this is true considering that bing returns results faster than el google.
Or maybe Bing was coded to be just as good as google thats why they get the same results. And I mean Its not totally wrong to see check up on your competition and see what results they are getting and change your code to help get the same results.

 

[gregmacknass]

Faster results just mean that their servers are faster (or they are just stealing searches and not searching themselves). But when you are directly stealing their search results and trying to claim them as your own, that is a joke. When Bing returns a gibberish search the exact same as google's gotcha test , I think that says it all. Obviously you didn't check that link out thoroughly enough.
-Greg

 

[Saaby900t]

Faster results just mean that their servers are faster (or they are just stealing searches and not searching themselves). But when you are directly stealing their search results and trying to claim them as your own, that is a joke. When Bing returns a gibberish search the exact same as google's gotcha test , I think that says it all. Obviously you didn't check that link out thoroughly enough.
-Greg

12 times in 100 yea it happens but the other 88 times it doesn't. So what Thats like 12% of the time.

 

[DaveHCYJ]

I think its awesome Bing is providing competition for google. Monopolies in any market are a bad thing.

I find the Bing commercials (that poke fun of all the useless junk google returns) quite accurate and humorous.

That being said I don't have an allegience to any search engine -- I use whatever works best. Which usually means using Bing and Google about 50/50. Tech searches seem to work better in google. "Life" searches seem to work better in Bing. The later category on google has the problem that marketers have figured out their algorithm so unlike 5 years ago when googled used to give good results all you get now is garbage -- sorry I don't want the first page to be nothing but eHow and sketchy retailers/fake product reviews that are really just links to buy whatever I searched for.

 

[Automata]

Stealing the work of another company is suddenly acceptable? That isn't competition.

 

[DaveHCYJ]

Stealing the work of another company is suddenly acceptable? That isn't competition.

Well I just trying to counter some of the anti-MS retoric. Bing isn't a terrible product, and statements like "yet another example of NON-innovation at Microsoft. Why bother with Bing at all now" are a bit over the top reactionary.

Now that I've read at least one of the articles, it doesn't even sound like "stealing" as you put it. Stealing would be actually copying google's results, which clearly they are not or otherwise 100% of the bogus terms would show up not 9%. What Bing is doing is just taking in more data points.

So for people that opted in to the Bing toolbar and allow Bing to collect data on what they search for and browse to, this is the data Bing is using to generate these results. These people could be searching with anything: bing, yahoo, google, etc. It wasn't a targeted "hey lets go steal google data" operation; it was a bit of innovation on Bing's part to statistically rank links that people click on higher than the links people do not click on. Sounds smart to me, and a way to filter clutter and junk results no one clicks on. Maybe google is just angry they didn't think of it, or angry MS is using the bing toolbar to collect data, but to Accuse MS of stealing is a bit misplaced.

Are people Angry at Amazon for taking a similar approach to tracking and correlating your purchases at 3rd party websites so they can make better suggestions when you visit their website? No, and if they are then they should opt not to use A9 and amazon toolbars and whatver else amazon uses to collect such data.

Keep in mind the bias built into these articles also -- the headline is "Google accuses MS of...." not "Police catch MS breaking and entering at Google headquarters"

 

[Automata]

Copying a portion of the result is just as bad as copying the whole thing. Stealing intellectual property is just as bad as stealing actual property. Stealing is still stealing.

The RIAA would still send cease and desist letters if you downloaded half the movie or the whole thing. There isn't a gray area.

 

[DaveHCYJ]

Right stealing is bad and there is no grey area about it.

My point was Bing isn't actually stealing at all -- they are using their own different processes to arrive at some of the same conclusions.


If some cooks set out to create the best cookie recipe and Cook#1 decided to use some really obscure ingrediants and passed them out to a bunch of consumers. Then Cook#2 polled those consumers and ask them, so what do you like about cookies (not specifically google's cookies, cookies in general, any and all cookies)? Cook#2 then compiles these results and trys to make the best recipe he can and ends up with a recipe with 9% overlap of obscure ingrediants with Cook#1 and maybe 5% overlap with Cook#3 (yahoo).
It it reasonable to conclude that Cook#2 broke into Cook#1's house and stole his recipe? No. That isn't a reasonable conclusion; Cook#2 polled the consumers (bing toolbar), figured out what they like and incorperated the results into his own recipe. Yeah sure Cook#1 might be pissed Cook#2 is polling the consumers to find out what they like and accuse Cook#2 of stealing. But just because Cook#1 makes that accusation doesn't make it true.


This story isn't really a black and white clear cut hahaha we caught you red handed event. To quote the MS employee:

"It was a creative tactic by a competitor, and we'll take it as a back-handed compliment," Shum wrote. "But it doesn't accurately portray how we use opt-in customer data as one of many inputs to help improve our user experience."

If I made a webpage wish some gibberish on it and also made my own search engine which pointed to that gibberish and somehow it ended up in google's search results should the headlines instantly read "Google caught stealing DaveHCYJ's search results" No. Maybe they could read "Dave accuses google of blah blah blah". But in the end its more likely that google's crawler just came to the same conclusion as my "search engine" all on its own without needing to steal my stuff.

 

[Automata]

They didn't arrive at the same conclusions. Google took a search that had absolutely no results and paired it to RIM's website. A user normally searching for this randomly generated sequence would come up with nothing, because it doesn't mean anything. After they paired the two, it was added to Bing. You are saying that Bing somehow magically got paired to the same website, with the same random sequence of words at the same time? Probability says otherwise.

They are taking the information/results from Google, however they do it, and are integrating it into their search.

 

[bigben6]

I would agree with DaveHCYJ that this is NOT aimed at google, but i also think it is shady, because in effect you are stealing results form yahoo, bing, google, or any other search engine instead of developing your own system. and yes 12% seems small, but 12% correlation on a completely gibberish query is REALLY weird...

 

[mbigna]

Well I just trying to counter some of the anti-MS [sic]retoric. Bing isn't a terrible product, and statements like "yet another example of NON-innovation at Microsoft. Why bother with Bing at all now" are a bit over the top reactionary.

This could be characterized as rhetoric only if it weren't true. Google created their own nonsensical 'honey pot' ranking/result that any independent search engine would never return. The fact that Bing now returns the exact same nonsensical results for the same search proves that Bing ranks its results based on the rankings of Google's--and possibly other--search engines--not it's [Bing's] own.

The succinct, and most excellent point, that I made is why should anyone bother using Bing at all if it is going to return the same results as the other search engines that are actually doing the work.

 

[DaveHCYJ]

They didn't arrive at the same conclusions. Google took a search that had absolutely no results and paired it to RIM's website. A user normally searching for this randomly generated sequence would come up with nothing, because it doesn't mean anything. After they paired the two, it was added to Bing. You are saying that Bing somehow magically got paired to the same website, with the same random sequence of words at the same time? Probability says otherwise.

They are taking the information/results from Google, however they do it, and are integrating it into their search.

Ok, lets talk about plausability.

There are what? 2billion people connected to the internet. IE has what? 50% market share right or roughly? Let say 10% of the people using IE have "opted in" which is probably conservative considering most know-nothing users have done it. Anyway, that gives us 100million people providing opt in data to MS. If each of those users does 1 search per day (probably also conservative) we're talking 1.4 billion searches during googles 2 week study. Google has a roughly 2/3rd's market share so we're talking 933million data points from google. Out of those 933,000,000 searches what are the chances someone accidentally mashed the keyboard and came up with one of googles "obscure" words. Its starting to look more plausible to me.
What if MS's opt in rate is more like 50% and people search more like 5 times per day. Now you're talking about ~24 Billion searches. What are the chances that out of 24 Billion searches, not one single person mashed the keyboard and came up with one of Google's honeypot terms? Or rather 9 of their honeypot terms. Eitherway, Statistically its looking pretty plausible to me.

As MS put it:

Google's accusations and the Search Engine Land story was "a spy-novelesque stunt to generate extreme outliers in tail query ranking."

When all it takes is 9 out of 24 Billion (or whatever the accurate exremely large number is) then yeah, we're talking about freak chances actually occurring.
Its kind of like "oh what are the chances someone will win the lottery". Despite how rare and extreme that event might seem, people win the lottery all the time because of the mass numbers of people that play.

 

[DaveHCYJ]

I would agree with DaveHCYJ that this is NOT aimed at google, but i also think it is shady, because in effect you are stealing results form yahoo, bing, google, or any other search engine instead of developing your own system. and yes 12% seems small, but 12% correlation on a completely gibberish query is REALLY weird...

If MS were methodically trying to steal their results it would be closer to 100%. They would simply start with a then aa then aaa then aaaa all the way up to zzzzzzzzzzzzzzzz. They would have hit all the rare terms not just 12% or 9% (as in the one article I read).

The fact that is was only ~10% to me strongly suggests it was just freak chances like mentioned in my post above, and not a concious evil plan by Bing.

 

[Automata]

Again, there is no other possible reason a completely random phrase, that normally turns up no results in both search engines, would magically show up in both within a few days. That literally does not happen without someone taking information from the other. It doesn't matter if they only got a few of them.

 

[DaveHCYJ]

In post 14 I detailed how it is completely possible. Some random person accidently mashed the keyboard and the said "what? what the heck does [blah blah blah] have to do with Research in Motion" and clicked on the link. MS harvested that click (via the user's choice to opt in and provide such data) and incorperated it into their statistical ranking scheme.

Granted I could see how this article could be interpreted as "omg data harvisting is evil" but not "MS is stealing from google".

I mean its a freak occurance that a user clicked on [jibberish]. It all makes alot more sense if you consider MS using things like when a user searches for "plane tickets" and via the opt in data notices no one ever clicks on that "how to get out of a parking ticket" link.

The fact of the matter is (according to MS) that Bing is making these decisions based on user's clicks, not the google search rank/algorithm.

 

[mbigna]

Ok, lets talk about plausability.

There are what? 2billion people connected to the internet. IE has what? 50% market share right or roughly? Let say 10% of the people using IE have "opted in" which is probably conservative considering most know-nothing users have done it. Anyway, that gives us 100million people providing opt in data to MS. If each of those users does 1 search per day (probably also conservative) we're talking 1.4 billion searches during googles 2 week study. Google has a roughly 2/3rd's market share so we're talking 933million data points from google. Out of those 933,000,000 searches what are the chances someone accidentally mashed the keyboard and came up with one of googles "obscure" words. Its starting to look more plausible to me.
What if MS's opt in rate is more like 50% and people search more like 5 times per day. Now you're talking about ~24 Billion searches. What are the chances that out of 24 Billion searches, not one single person mashed the keyboard and came up with one of Google's honeypot terms? Or rather 9 of their honeypot terms. Eitherway, Statistically its looking pretty plausible to me.

As MS put it:


When all it takes is 9 out of 24 Billion (or whatever the accurate exremely large number is) then yeah, we're talking about freak chances actually occurring.
Its kind of like "oh what are the chances someone will win the lottery". Despite how rare and extreme that event might seem, people win the lottery all the time because of the mass numbers of people that play.

Your terrible 'statical analysis' notwithstanding, your premises is still flawed. While it may certainly be that some number of people--no matter how great that number might be--could have accidentally entered in the same search text, no search engine would have returned the results on their own that Google contrived (not even Google--they hard-coded the returned results for the nonsense search text).

 

[Shelnutt2]

Here is another good article:
http://www.seattlepi.com/business/1700ap_us_tec_google_microsoft_search_cheating.html

In the end microsoft tracking your clicks is not different than google adsense and stuff. I don't like it, I feel microsoft is cheating by seeing what links you clicked on compared to what you searched. But I feel like if that is the only way Bing can get results that match google then it'll bite them eventually. I know bing is better at some stuff, and google at other, blah blah, but anyone can see if you give 4 chan a day with bing/google they could make microsoft regret the tracking/changing of search results.

 

[DaveHCYJ]

Your terrible 'statical analysis' notwithstanding, your premises is still flawed. While it may certainly be that some number of people--no matter how great that number might be--could have accidentally entered in the same search text, no search engine would have returned the results on their own that Google contrived (not even Google--they hard-coded the returned results for the nonsense search text).

That is until the user clicked the bogus link google provided to them. The user sent that data to MS and MS included it in their algorithm.

Again its not like MS is crawling the google page and just duplicating their results. MS is coming up with all their own results from probably thousands of different data sources, one of them happening to be User provided opt in data.

Yes it seems like a freak event if the google search for [jibberish] returns just a single link to RIM and then later Bing does the same; it does suggest something strange going on. But I'm not going to go running out and accusing my neighbor of being a witch if she wins the lottery. Freak events happen, especially when there are billions of chances for them to occur.