• On TechRepublic: Why Linux will triumph over Windows
August 14, 2009 4:00 AM PDT

Google's Varian: Search scale is 'bogus'

by Tom Krazit
  • Font size
  • Print
  • 33 comments

Google's Hal Varian would likely have raised an eyebrow at a term paper submitted by Microsoft CEO Steve Ballmer on the search market.

Varian, currently on leave from the University of California at Berkeley to serve as Google's chief economist, thinks a lot of the arguments advanced by Microsoft in justifying its 10-year deal for Yahoo search are, in a word, "bogus." Microsoft has said that it needs "scale" to compete in the search market against Google, saying that larger amounts of traffic and data allow it to improve the quality of its search experience.

Hal Varian, chief economist at Google

(Credit: Google)

As might be expected, that's not exactly the way Varian sees it. He's perhaps best known for perfecting the ad auction system that generates the vast majority of Google's huge profits, having worked for Google since 2002. But he also holds forth over the array of statistical data and processes that Google uses to make just about any decision.

Varian shared his thoughts on the Microsoft-Yahoo deal, the state of the economy, and the changing nature of innovation and Silicon Valley geography during a conversation at Google's headquarters in Mountain View, Calif., this week.

Q: One thing we've been talking about over the last two weeks is scale in search and search advertising. Is there a point at which it doesn't matter whether you have more market share in looking to make your product better?
Hal Varian: Absolutely. We're very skeptical about the scale argument, as you might expect. There's a lot of aspects to this subject that are not very well understood.

On this data issue, people keep talking about how more data gives you a bigger advantage. But when you look at data, there's a small statistical point that the accuracy with which you can measure things as they go up is the square root of the sample size. So there's a kind of natural diminishing returns to scale just because of statistics: you have to have four times as big a sample to get twice as good an estimate.

Another point that I think is very important to remember...query traffic is growing at over 40 percent a year. If you have something that is growing at 40 percent a year, that means it doubles in two years.

So the amount of traffic that Yahoo, say, has now is about what Google had two years ago. So where's this scale business? I mean, this is kind of crazy.

The other thing is, when we do improvements at Google, everything we do essentially is tested on a 1 percent or 0.5 percent experiment to see whether it's really offering an improvement. So, if you're half the size, well, you run a 2 percent experiment.

So in all of this stuff, the scale arguments are pretty bogus in our view because it's not the quantity or quality of the ingredients that make a difference, it's the recipes. We think we're where we are today because we've got better recipes and we have better recipes because we spent 10 years working on search improving the performance of the algorithm.

Maybe I'm pushing this metaphor farther than it should go, but I also think we have a better kitchen. We've put a lot of effort into building a really powerful infrastructure at Google, the development environment at Google is very good.

So, how's the economy look?
Varian: The news on the economy I'd say is pretty good. You look at the housing sales, they've leveled out, prices are up slightly. The auto sales and production are up, The financial markets have all stabilized, the initial unemployment numbers are down over 100,000. So everything is looking pretty reasonable, and it's somewhat earlier than expected.

That doesn't mean we're out of the woods, because we've got a long way to go to go back up. But you look at a most of the economic statistics, they have really turned around in the last couple of months, not only here, but in Europe and Asia.

Has anything changed fundamentally?
Varian: Well, the savings rate is up. People had a negative savings rate for several years and now it's more like 7 percent. In some sense, that's a good thing, I know people are complaining about it but you have to restore some reasonable balance. Maybe it's not so good that we had to get to it by going through this recession, but at least we're coming out at a more balanced rate than we were going in.

How do you see things in Silicon Valley? We've been wondering about the growing cost of living in this area and what effects that has on business development.
Varian: Well, there are some outposts: I've been telling Diana (Adair, Google spokeswoman) that she should go buy a house in Pleasanton. That's where PeopleSoft used to be, and Oracle has a big establishment there, it's a nice town....

It's 105 today in Pleasanton.
Varian: Actually, I live up that way, if you get hot, you jump in the swimming pool. Anyway, there are some outposts that are still closely connected to the Valley. I think it is getting awfully expensive to live here, and commuting is getting more and more unpleasant, so I think you will be seeing some expansion.

A year ago, I told Google they should buy in Stockton. But nobody listened to me. The deal is, you have to pay for your food but your house is free.

What did they say?
Varian: They said, nah, we couldn't get anybody to live there.

Is innovation in the Valley as high as it was 10 years ago?
Varian: I'll tell you an angle that I think is different from 10 years ago, and that's what I call the micro-multinational.

One day I bumped into a friend of mine, and asked what she was up to. She said, I've got a company. And I said tell me about it, and she said there are 12 people, three in New Delhi, two in Mountain View, and there's somebody in Spain.

And then two days later I ran into another guy, and he said I've got a company, and there are four people in Italy, two people in the Czech Republic, one in Spain, and three in San Francisco. And I said, whoa, what is this? A trend! It's two of them!

But you talk to them and it's amazing what you see in this area. I think the reason is that communication costs have basically gone to zero. We've got e-mail, Skype, Google Docs, wikis, doing round the clock continual communication and coordination that only the biggest multinational could have 10 years ago. The fact is that you have this essentially free communications and you've got entrepreneurs everywhere else in the world that can sync up.

So the question is, is it in the Valley? Well, in many cases, two or three of the people are in the Valley, but it's not limited to the Valley.

I think it's a crucial part because in a lot of cases you can find expertise, you can find venture capital. You've got the legal people to draw up the contracts, you've got the financing people, you've got the consultants and experts... But maybe it is part of the answer to this cost question, because you've got the expertise but the work can be distributed.

Tom Krazit writes about the ever-expanding world of Internet search, including Google, Yahoo, online advertising, and portals, as well as the evolution of mobile computing. He has written about traditional PC companies, chip manufacturers, and mobile computers, spending the last three years covering Apple. E-mail Tom.
Recent posts from Relevant Results
New Google search UI brings color, search options
Crude Michelle Obama image dumped by site owner
Google to track TiVo viewing habits
Google places ad explaining offensive image
Can News Corp. afford calling Google's bluff?
Brin: Google's OSes likely to converge
Judge sets February hearing for new Google Books deal
Google has its own plan for Netbooks
Add a Comment (Log in or register) (33 Comments)
  • prev
  • 1
  • next
by DCDummer August 14, 2009 5:32 AM PDT
lol@ Google's kitchen..
Thanks for asking a clever person some clever questions Tom!
Reply to this comment
by only_buy_FLAC_4_quality August 14, 2009 8:43 AM PDT
clever? where?


"bogus"

hello bayesian learning??



" there's a kind of natural diminishing returns to scale "

why then does google continue to expand its offering of "free" products (that collect VAST swaths of personal information)??


"this is kind of crazy"

translation: google is worried by bing




"we do essentially is tested on a 1 percent or 0.5 percent experiment to see whether it's really offering an improvement."

When then is the gizmo5 tie to Google Voice breaking down so soon?




"it's not the quantity or quality of the ingredients that make a difference, it's the recipes."

speaking of bogus..




"That doesn't mean we're out of the woods, because we've got a long way to go to go back up"

We need to run screaming from The People's Forest notion! (socialism)



" but at least we're coming out at a more balanced rate than we were going in"

more bogus from google.

more bailouts != more balance


"We've been wondering about the growing cost of living in this area and what effects that has on business development."

We're still wondering apparently.



"still closely connected to the Valley"

What? google doesn't have a "google telecommune" project to rave about?

GooTune?



"But nobody listened to me."

:snicker:




"you have to pay for your food but your house is free."

That sounds much like the democrat's plan that forced banks to give loans to unqualified borrowers.
by FutureGuy August 14, 2009 8:41 PM PDT
He didn't sound too clever. He supposed to be an "economist" but he must have been asleep when taught about economy of scale. He sounds too much like Glenn Beck.
by pentest August 16, 2009 3:13 PM PDT
Why would Google be scared of Bing? It is a pale shadow of what google is capable of, with a little bloat added for good measure.
by pdbrickhouse August 14, 2009 5:51 AM PDT
It strikes me that the "scaling" is more important on the sales side of the equation??????????
Reply to this comment
by eadeguzman August 14, 2009 6:07 AM PDT
Interesting conversation. Seems to be a very nice honest guy (and smart too).

The Stockton thing was funny... But I think Google is big enough and important for people to want to go there wherever they are.

I think "we couldn't get anybody to live there" translates to "I don't want to live there" -- was that Eric Schmidt he was conversing with?

Nice article.
Reply to this comment
by disco-legend-zeke August 14, 2009 6:49 AM PDT
1. Scaling... certain fixed costs can be spread over a larger customer base.

2. Micro-Multinational... Although they are not employees, i use a small company in China for real-time communication services, A server farm in Canada, and a couple One-Man programming shops in Europe for image and multi-player game systems.
Reply to this comment
by Seaspray0 August 14, 2009 7:02 AM PDT
"i use a small company in China for real-time communication services..." 100% of the spam I see now comes from china. Is "real-time communication services" just a fancy way of saying spam?
by only_buy_FLAC_4_quality August 14, 2009 8:52 AM PDT
@Seaspray0

"Is "real-time communication services" just a fancy way of saying spam?"

Not unless it's SPIM (sms), or SIP-spam (voip)

Then there's the twit service but it's realtime in the email sense.




"the spam I see "

What scheme are you using to not-see spam?


Why there is not more focus on spam MTA rejection amuses me.
by YankeePoodle August 14, 2009 8:06 AM PDT
I was expecting him to say, Microsoft - Yahoo deal was a good move. Did you say he gets his Pay Check from Google? Never mind.
Reply to this comment
by knowles2 August 14, 2009 1:44 PM PDT
I would say it look like a good deal.
There plenty of these that have look good and sounded good deals but turn out to be a bit of a dud.

Is this one of them it depends on which party you look at, I am still not convince this was a good deal for Yahoo in the long term,10+ years range, where Microsoft could dumb Yahoo, let there share pummet and then cherry what ever else they want to acquire of them, and by that time staff would of got use to work there and they will loose far lest talent.
Good plan my Microsoft, not so good for Yahoo
by JCPayne August 14, 2009 8:10 AM PDT
Exactly. Bing will have "Bung" just like Live and MSN search and all the rest of the stuff they had over there.
Reply to this comment
by viper396 August 14, 2009 12:41 PM PDT
Your statement is on par with the things little kids on playgrounds would say to each other. If the best you can do is slur the names and titles of products then your opinion is worthless. Just a waste of time.
by t8 August 14, 2009 3:16 PM PDT
@ viper396

He gave historical evidence of his statement.
He is basically saying history will repeat itself.

Playground taunts go something like Bing is Bung.
That would be it. No evidence, just an opinion.
by FutureGuy August 14, 2009 8:39 PM PDT
@t8 you mean like since the first gen Xbox didn't do that well the second gen also bombed, ohh wait it didn't!!! Go back to playing.
by Super2online August 16, 2009 12:30 PM PDT
This guy is doing a clever job of trying to muddy the waters. Ballmer never claimed they needed more data giving them scale. What he has always talked about is having more eye balls looking at ads attracting more advertisers. 20% more eye balls increases advertising dollars which provides more money to pour into research and experimentation efforts and results.

It's interesting that all of a sudden we find Google going on the defensive. Evidently they feel it's now neccessary to defend their practices. It's also funny how Google's perception of having real competion can change their desire to be more open about what they are doing.

Have you ever noticed how kitchen's change over the years and how that has a significant impact on what comes out of it. Maybe it's time they take a good hard look at their kitchen and their receipies. It appears there is a growing number of people moving over to MIcrosofts kitchen table where they seem to like what's on their menu instead.
by pentest August 16, 2009 3:17 PM PDT
In financial terms XBOX is a failure. In technical terms it is a billion dollar failure.

Live = fail
MSN Search = fail
Bing = better, but still lacking. If MS hired some quality functional programmers who understand Map Reduce they might catch up a little.

Since it is still lacking and really hasn't made much in terms of waves, it looks to be following the MS history.
by Kwasiowusu August 16, 2009 4:15 PM PDT
@pentest :" In financial terms XBOX is a failure. In technical terms it is a billion dollar failure"

The E & D Division (XBOX, Zune, Winmo etc), made a profit both in the 2008 fiscal year, and the just ended 2009 fiscal year.
Not to mention, the XBOX 360 continues to clobber the PS3 every month in NPD sales, and is on track to start outselling the Wii from September as well. Plus multi-patform games almost alwatys sell more on the 360 than on any other platform
by David Turner August 17, 2009 6:31 AM PDT
@ Kwasiowusu Guess it all depends on the spin you put whether MS XBox is doing all that well. But according to the xbox team its not doing as well as it once was.

"Xbox 360 platform and PC game revenue decreased $110 million or 12%, primarily as a result of decreased Xbox 360 console sales and decreased revenue per Xbox 360 console due to price reductions during the past 12 months, partially offset by increased Xbox Live revenue. We shipped 1.2 million Xbox 360 consoles during the fourth quarter of fiscal year 2009, compared with 1.3 million Xbox 360 consoles during the fourth quarter of fiscal year 2008."

So sales are down even on top of a price drop.

Given that Nintendo has constantly out sold them for many months now even if Ms XBox did out sell them in Sept your assessment of the situation is highly optimistic

http://news.teamxbox.com/xbox/20345/Microsoft-Reports-Fiscal-Q4-Results-Profit-Falls-29/
by jsmanger August 14, 2009 8:12 AM PDT
Varian says that the scale of the data argument is bogus, but it was Google's engineers who were only recently touting the importance of data over "recipes": http://www.computer.org/portal/cms_docs_intelligent/intelligent/homepage/2009/x2exp.pdf
Reply to this comment
by eagunner August 14, 2009 10:21 AM PDT
Ah, interesting comment! I'm not going to read a PDF, so from where I'm standing what you said might be a lie, but if true, veeeeerry interesting.


Dear God man, why a pdf??
by someguy999 August 14, 2009 11:20 AM PDT
a company's economist is like a politician... they say what they're paid to say.
Reply to this comment
by WillyWiggler August 14, 2009 11:21 AM PDT
You're not going to read a pdf?!?!? Wow. An electronic document format snob.
Reply to this comment
by woodycr August 14, 2009 12:19 PM PDT
Business value = Eyeballs * intent

Scale matters. Specially in display ads where intent is really low.
People who argue/use statistics just want to reinforce their view.

To intent one has to understand what intent is and how it is related to information and context to drive it up. Which requires scale of in data.

If we have a consumer driven economy and the consumer has no increased income, the last income increase where statistical and loaned, then the recovery will be short lived.

I'm not impressed with that guy.
Reply to this comment
by Vegaman_Dan August 14, 2009 2:36 PM PDT
Varian's response is pretty much what you would expect someone to say of a competing product. Of course he would try to dismiss it as irelevant and unimportant. To do anything else would be acknowledgement of the competition however small and therefore give it validity.

You see it all the time between Microsoft and Apple, so I am not terribly surprised by this either.
Reply to this comment
by BrianTheRight August 14, 2009 8:06 PM PDT
It is amazing to me all the Microsoft plants on technical sites. Here we have an interesting conversation with a high up of Google and what do we get? We get the trolls of Microsoft loading the comment section with their usual lines of inanity. What a joke! As for the conversation, it was quite interesting and gave insight into Google's mindset as a company. I think both Microsoft and Google are evil corporations bent on taking over the world, it is just Microsoft is more annoying and their people are a joke!
Reply to this comment
by YankeePoodle August 17, 2009 3:47 PM PDT
You are joke if people who created their own functional programming infrastructure, an Operating system, Robotics research and Enterprise software development deployment infrastructure are joke to you. If you are a google fanboy be that. Do not go after people 10 times smarter than you are and call them a joke.
by sam012004 August 14, 2009 9:19 PM PDT
Hey guys search ranking is based on probability theory and is discrete and coutable, that means larger population size makes your contineous model approximate the real thing more closely. The law of diminishing returns also does not always apply. Mr. Hal Varian is, I am sure, bluffing. What ever algorithm they use testing and varification at large scale is always helped by sample size. The smaller the scale the modeling errors swamp out algorithm performance tracking. Sorry for being to technical but I don't see any other way of explaining it.
Reply to this comment
by sam012004 August 14, 2009 9:20 PM PDT
Hey guys search ranking is based on probability theory and is discrete and coutable, that means larger population size makes your contineous model approximate the real thing more closely. The law of diminishing returns also does not always apply. Mr. Hal Varian is, I am sure, bluffing. What ever algorithm they use testing and varification at large scale is always helped by sample size. The smaller the scale the modeling errors swamp out algorithm performance tracking. Sorry for being to technical but I don't see any other way of explaining it.
Reply to this comment
by tsrini408 August 15, 2009 10:12 AM PDT
As a layman who used to be a techie, I can safely say that any searches with Yahoo or Bing for that matter, do not yet provide the precision of Google search returns, even with the obvious chumping of the system that goes on by flakes who want their name to be on top of the list. When I hear a friend ask me to try Yahoo or Bing instead of Google is when I will switch. Until then, Google is king.
Reply to this comment
by bdennis410 August 15, 2009 11:15 AM PDT
Cloud, baby.
AnyThing,AnyTime,AnyWhere (AAA)
Varian's discovery of distributed opportunity through distributed information, through distributed work makes the point about the coming obsalesence of "commuting;" furthers the impetus services liketrue home-schooling based on technology driven "learning" (Not "teaching"-way outmoded-only the teacher's unions prevent more progress)., and all the ohter potential collaboration-based, interactive services and processes that are coming.
Like Health, Work of all kinds, Entertainment, more.
Reply to this comment
by AppleSuxLeo August 16, 2009 3:14 PM PDT
Yahoo has more page views than any other portal , as well as the most popular Web-Mail by far.
Mr sour grapes is just jealous. And he looks like a wax figure LOL
Reply to this comment
by AppleSuxLeo August 16, 2009 3:16 PM PDT
Is VARIAN actually a MARTIAN ? LOL
Reply to this comment
(33 Comments)
  • prev
  • 1
  • next
advertisement

The browser battles go on and on

roundup From Firefox to IE and from Chrome to Opera and Safari, there's no sitting still for browser makers looking to keep their products fresh and competitive.

3G wireless still holds promise

The next generation of 4G wireless may get all the headlines, but advanced 3G technology will likely dominate services for the next few years.

About Relevant Results

Relevant Results focuses on the big Internet companies of our time, tracking the evolution of search, communication, and business on the Web. Tom Krazit examines how a shift to mobile computing and the growing demand for online content affect our understanding of how to deliver information in the 21st century, in between bemoaning the state of the New York Mets and searching for the perfect IPA.

Add this feed to your online news reader

Relevant Results topics

advertisement
advertisement

Inside CNET News

Scroll Left Scroll Right