New Interview with Google Engineers about Algo

[aristotle]

The interview can be found at

[msn.finance.com.my ]

Here is a quote:

Proposed changes to Google's formula are first tested on a separate set of computers that imitate real-world search.

Those deemed worthy are next sent to evaluators around the world who act as online searchers and rate the relevance of results in various languages and regions.

Google then does live testing, with promising algorithm enhancements carefully blended into results served up by the main search engine.

"At any given time, some percentage of our users is actually seeing experiments," Huffman said.

[tedster]

I couldn't help but chuckle that this is an MSN Finance article, and we're also seeing Google Search coverage from CNN Money columnists. When Google Search was only being covered by technology writers, they habitually wrote very inaccurate articles. Now that the world knows how much money and power are involved, the articles move to finance reporters and they do a better job than tech reporters ever did.

Scott Huffman's team tested "many more than" 6,000 changes to its search engine in 2010, with 500 of them passing the grade to become permanent.

That's an even faster pace than the 400 algo changes reported in 2009. Also interesting to see that only 1 in 12 tests actually make the grade and become "permanent".

[edited by: tedster at 2:00 am (utc) on Apr 16, 2011]

[walkman]

"If you think of the scale of what we are talking about, it is almost absurd to say we could rig results," Huffman said, noting that Google handles more than a billion searches daily.

Shopping comparison sites that actually provide unique reviews and compete with G Shopping disagree with you. Soon the Yelps, and travel sites will do the same...and financial sites, and...

To be on page #1 soon you will have to pay, Google will fill all searches with it's own properties.

It's all a PR effort, Google is choking on cr@p, and is trying to please the big media in a futile attempt to extend it's life as a leader by a few more years.

Their media campaign shows the true state of Google search.

[tedster]

That kind of "rigging" is done by inserting new blocks onto the results page. That's how Google does it. They are not rigging the actual organic results, they are displacing them.

[indyank]

I think these interviews are getting more irritating.They just boast of what they do do and what they plan to do.Too much of boasting...

They keep saying that they improve quality and they are not giving any good examples of what the search looked like before Feb 24 and how it looks improved now.

Matt always says that he loves webmasterworld but people here aren't giving real world examples.But they aren't doing it either.Except for one example which he keeps repeating in interviews there aren't any good ones.Even the one he used as an example doesn't anyway prove the kind of quality they have unearthed now.

[Sgt_Kickaxe]

I can't even read a lot of the finance articles these days because the writer is trying to use SEO speak and is blatantly keyword stuffing the articles.

"People are lookign for blue widgets and red widgets and purple widgets and sometimes pink widgets along with pink-purple widgets and black widgets and ...and...and" bleh!

I miss articles written for people, the worst companies have staff on hand to immediately write about whatever is hot.. those are the ones to avoid.

[walkman]

That kind of "rigging" is done by inserting new blocks onto the results page. That's how Google does it. They are not rigging the actual organic results, they are displacing them.

Sure they are "rigging" them Ted. How come they are more prominent and less prominent during certain updates.

Now whether it's done because other division execs told /"hinted" at them or not is a different matter. Imagine a "man, aren't those InsiderPages and Yelp searches filling the SERPS" comment to the search team, it doesn't have to be a memo or a secret meeting.

Who decides what prominence to give to such pages in the SERPS? It's obvious that someone dials it up and down, we see entire categories move up and down in batches.

[browsee]

"If we care about our users -- don't care about money -- everything else just falls in line," Singhal said. "A healthy Web and happy users are key to our future."

Really? Healthy web and happy users? Just come out your glass doors and search like a normal user and you will know how healthy is your web. No wonder why users are moving to other search engines.

[indyank]

Singhal pictured a day when search engines understand users so well that they predict what people wanted to know and cue them with messages on smartphones.

"That is the ultimate dream," Singhal said. "We are nowhere close to that yet."

it looks like they are dreaming about how they could spam people on the pretext they knew what the user wants.His dream of "cue them with messages on smartphones." is what we really call as SPAM and unnecessary intrusion.

I already see a lot of these irritating spam text messages from all kinds of advertisers in my country here.They don't respect anything and they only spam my smartphones everyday.No law can control these spammers in this part of the world which tops corruption these days. I couldn't believe that Amit seem to be dreaming about doing the same thing.

[walkman]

Yahoo! Search and Bing achieved the highest success rates in March 2011. This means that for both search engines, more than 80 percent of searches executed resulted in a visit to a Website. Google achieved a success rate of 66 percent. The share of unsuccessful searches highlights the opportunity for both the search engines and marketers to evaluate the search engine result pages to ensure that searchers are finding relevant information.

[hitwise.com...]
This is the reality Google ^^^, stop giving interviews, shut up and get to work. This was post-Panda too so we can see what a success it was, Bing cleaned your clock and gained over 3% market share in march alone!

_{(unless Hitwise is not reliable)}

[browsee]

Bing cleaned your clock and gained over 3% market share in march alone!

This is US search only. They clearly gained market share after Panda. BTW, my bing referrals are better than ever.

[crobb305]

BTW, my bing referrals are better than ever.

My bounce rate dropped from 40% to 27% during a large dip in Google traffic last weekend (Google traffic dropped 25%, with most referrals coming from Bing). My bounce rate has only gone up to 40% since Google implemented silly features like preview and auto-complete. For six years, up through mid 2010, my bounce rate was 32% or lower.

[tedster]

Can your analytics isolate bounce rate by entry type - by search keyword in particular?

[semseoanalyst]

Stop dreaming G..because Bing is knocking the door...got a good response from Bing after Panda for my sites..

[rlange]

indyank wrote:
Matt always says that he loves webmasterworld but people here aren't giving real world examples.But they aren't doing it either.Except for one example which he keeps repeating in interviews there aren't any good ones.Even the one he used as an example doesn't anyway prove the kind of quality they have unearthed now.

Has anyone, anywhere been citing real-world examples of improved search results? I haven't been reading a broad range of sites that might be talking about Panda, so my experience is very limited, but... all the "positive" comments I've seen about the update were not much more than people congratulating Google on doing something about the problem. I don't think I've seen any of these people even claim that they've actually seen an improvement.

walkman wrote:
This was post-Panda too so we can see what a success it was [...]

Another way of looking at that is: If Panda is a success, then what was Google's success rate before the update?

--
Ryan

[crobb305]

Can your analytics isolate bounce rate by entry type - by search keyword in particular?

I can when I use my raw log files (I have been saving those data for years so there is a lot of information hiding in those zipped files). Normally, I just use basic analytics for my day to day traffic monitoring; but, I may delve into some of the raw files and see if anything pops out. The bounce-rate trends since mid 2010 (after auto-complete came along or when I lose/gain Google traffic) are clear, even when using just my basic analytics.

[browsee]

Just saw that Google 1Q (Jan-Mar) earnings missed analyst target. Lets see what will happen to their earnings post panda.

[AlyssaS]

Yahoo! Search and Bing achieved the highest success rates in March 2011. This means that for both search engines, more than 80 percent of searches executed resulted in a visit to a Website. Google achieved a success rate of 66 percent.

That's because 20% of searches done on Google are by webmasters (and their ranking bots) desperately checking and rechecking whether they've moved up a place or two.... Or down, ahem.

[Reno]

it looks like they are dreaming about how they could spam people on the pretext they knew what the user wants.

That's exactly what they're dreaming about, and that is one seriously profitable dream. Can you even imagine how many smartphones there will be in the world when Google undertakes this spam campaign? BILLIONS. Each ad that they send will come with a price tag, so I have no doubt that everyone in the Plex is salivating over those $$ numbers.

I already get targeted spam when I use my gmail account. If I write to someone about something specific, I suddenly start seeing those targeted ads. But gmail is free and that's the deal, so I have no room to complain.

But when people start getting unsolicited spams on their smartphone, look for a HUGE backlash, especially if that particular Google project is as cr@ppy as Panda. At that point, it won't just be webmasters who are disgusted.

.....................

[chrisv1963]

Just saw that Google 1Q (Jan-Mar) earnings missed analyst target. Lets see what will happen to their earnings post panda.

I don't know much about shares and stuff like that, but for GOOG I see an "After Hours" drop of 5.35%. What is the meaning of this?

[browsee]

I don't know much about shares and stuff like that, but for GOOG I see an "After Hours" drop of 5.35%. What is the meaning of this?

Panda Cannibalization.

[tedster]

This forum is here to discuss SEO and the algorithm. We have another forum to discuss Google Finances, including stock, lawsuits, etc. In fact, there's a new thread just started to discuss Panda's possible effect on Google Stock [webmasterworld.com]

Returning to this interview with the engineers, I can't help but wonder how many testing cycles Panda went through before it was launched - especially those "small subset of user" live tests. Givem how big it has been (12% of all searches were affected on the first roll-out) I assume it must have been more extensive than usual.

[Reno]

From the interview...

"Alongside changing the engines, the plane has become quieter, the ride got more comfortable, and we even changed your seat while you were sleeping," he continued. "We just do it in small steps that go unnoticed."

By "changing your seat", I assume he meant the position in the SERPs (not the fabric), and believe me, we notice. Otherwise, the person is in the exact same spot, only the surroundings have changed.

The move was part of an ongoing duel between the search titan and low-quality websites that feature only content copied from elsewhere on the Internet or use techniques to trick their way high in results.

OK, so this interviewer apparently got the engineers to tell him that a "quality" site does not use copied content and is white hat. So that begs the question: Why did so many of us get hit so hard when we are not using tricks and have original content?

Which leads to my third point:

"People are not just expecting a search engine to return every document that has most of the words typed in a query box," Huffman said. "They want the context understood; there are a lot of nuances hidden within that."

Otherwise known as the semantic web, which makes me wonder if the Panda algo is favoring pages which have been more strictly designed for semantic web standards? (as opposed to older HTML pages). Would this be one of the reasons why so many long term established & clean sites got unfairly hit?

Read the... Semantic Web Components [en.wikipedia.org]
(Scroll down to #4: "components")

.......................

[Freedom]

I am so sick and tired of GoogleSoft "engineers" coming up with stupid metaphors and similes.

[tedster]

OK, so this interviewer apparently got the engineers to tell him that a "quality" site does not use copied content and is white hat. So that begs the question: Why did so many of us get hit so hard when we are not using tricks and have original content?

I wouldn't put any stock at all in what the reporter wrote - only what he quoted.

[Dan01]

Indyank - your statement above is spot on. They haven't really showed any SERPs before Panda and after. I think it is in their best interest to improve them, but I would love to see what they did.

Walkman:

To be on page #1 soon you will have to pay, Google will fill all searches with it's own properties.

It's all a PR effort, Google is choking on cr@p, and is trying to please the big media in a futile attempt to extend it's life as a leader by a few more years.

I agree with you walkman. Ted also makes a good point; they may not be changing the result, but just displacing them. That could cost them, I don't know. Yahoo tried to make themselves the one-stop shop online.

[TheMadScientist]

All this talk of Panda being worse and their stock price / earnings being down (or below projections) seems to fly in the face of the 'conspiracy theory' relating to them making results worse to increase ad clicks we've been hearing for years, doesn't it? I'm inclined to believe the quote about them not discussing or considering the bottom line when they're trying to rank sites.

I'm sure Panda was well tested before it rolled, but whether the results of those tests were entirely accurate, only Google knows ... The number of 'didn't make the grade' tests is surprising to me ... As far as the update being 'bad' goes, I keep thinking there was quite a bit of noise about Florida too and I think they just moved forward from there, didn't they?

[edited by: TheMadScientist at 2:20 am (utc) on Apr 15, 2011]

[Dan01]

Returning to this interview with the engineers, I can't help but wonder how many testing cycles Panda went through before it was launched - especially those "small subset of user" live tests.

They worked on it for a year. I am still not sure what it did. It sounds like it was a major change.

Usually, when a programmer writes a program, the program will assign values, etc. Panda sounds like subroutine (or function) of a larger algorithm. The program calls on each subroutine. Each subroutine runs and spits out a value.

The values are weighted. In other words, too many unnatural links may be more important than the ratio of ads / content. A final value is computed for each site, depending on the the query.

Finally, the program will spit out a result (SERP).

The first reports said Panda went after content farms. And then someone said it went after scraper sites. Low value sites. Shallow content. Background color. Oh goodness, I have heard it all. LOL

I would love to see a thread that tries to analyze the criteria used in Panda. What factors are considered.

[indyank]

I would love to see a thread that tries to analyze the criteria used in Panda. What factors are considered.

We would know them only if we know of good examples of how the SERPS have improved post panda.We need a few solid examples (queries) that clearly show how post-panda is better than pre-panda.

[indyank]

That's because 20% of searches done on Google are by webmasters

if that accounted for 20% as you say, then what kind of feedback that G had got to tell them this is better? Are they saying it based on this feedback?

I don't see any public forum where they accept feedback. If they say that this is overwhelmingly positive, why don't they open a public forum to accept feedback.

I thought their feedback is based on their perception of user behavior, while searching on Google.If that be the case, neither what you say nor the hitwise data confirms their theory of "overwhelmingly positive feedback".

There is nothing they do to support their claims of "positive", "successful" or whatever.