Google Search Engine Core Called Hummingbird

[Brett_Tabke]

Google changed its underlying search engine technology about a month ago. The new version's called Hummingbird, and the change was made to better answer the complex questions so many of us are asking Google these days. Since the switch, Hummingbird's been put to use in over ninety percent of worldwide searches and positions Google to stay on top of its game for the foreseeable future. It's the culmination of 15 years of work, but Singhal says it's just the next step in Search's evolution, so we can expect more of the same in the next 15.

[engadget.com...]

[MrSavage]

@mrguy, now that's a great point and frankly I'm speechless.

[netmeg]

2 - WebmasterWorld doesn't criticise Google enough. They had broken SERPS, obvious and clear. We should have ALL yelled louder.

Not exactly what WebmasterWorld is here for.

Google doesn't consider itself a search engine anymore, they say they're a knowledge engine. Hence the Hummingbird. If Google *can* scrape your knowledge, they will. So looking forward, your best choices are probably going to be either staying below the radar, or going off the grid.

[EditorialGuy]

OK, let's breathe into a paper bag and look at Hummingbird objectively:

This is not the end of the Web, or a face-saving gesture to save Google embarrassment over past mistakes.

It's simply an evolutionary step forward in search.

Most people aren't librarians, professional researchers, or Boolean Logic gurus. They just want to ask questions and either get answers ("What's 100 meters in feet?") or get sent to sites that can provide the information they need ("What's the difference between a bridge and a crown?" or "How is chocolate made?"). Hummingbird is a step toward making search more useful for the public at large, and toward making search more about concepts than about keywords ("things, not strings," as Matt Cutts said in one of his videos).

Yes, Hummingbird may be bad news if your business strategy is built around exploiting keywords and keyphrases, but that doesn't mean it's evil. Evolution happens, and you need to decide if you're going to crawl up on land or stay in the water with the fishes.

[Dymero]

I wonder if this now means that pages don't need to have finely targeted keywords in order to rank well, assuming you have other positive signals like good links. Obviously, your title tag and other content has to still have to be about the query in question, but now Google knows what the content is about even if the exact words in the query are not on the page.


Google's been heading this way since it started being able to determine synonyms, but now it's moved into matching concepts.

It'd be a worthy test to do, to see if a page can rank well for queries that are less precise. The true metric of success will be how well it matches results to what the searcher intended, rather than just what the query is.

[aakk9999]

It is interesting that with this new development Google still gets the language on the page wrong.

I have a page in German (and on-page content is NOT mixed with English) but Google thinks the page is in English and offers "Translate from English to German" on Google.de SERPs. If I copy the text from the page into Google Translate, Google Translate figures it out correctly it is German.

This started a few weeks ago, so perhaps it is connected with this new release.

[SevenCubed]

What's the reason for Matt Cutts total absence?

I've always had a sense that MC was about the only ray of hope within google, right or wrong, that's how I've perceived him. I have no doubt that many times he had to tow the company line, maybe even openly among his co-workers.

But I had a sense that he was becoming increasingly uncomfortable towing the company line. So maybe he reached a point where his conscious couldn't allow him to do it so openly anymore.

Remove him from the beast's spotlight and I think he'd be someone I could enjoy having a conversation with.

[reseller]

But I had a sense that he was becoming increasingly uncomfortable towing the company line.

I have the same feeling. Matt Cutts has a big heart and generous spirit and maybe he doesn't feel comfortable in Google anymore. Now he has been with Google since 2000, maybe its time for Matt to leave Google and see the world.

[EditorialGuy]

What's the reason for Matt Cutts total absence?

It could be as simple as "He's on vacation," "He's busy organizing an antispam raid," or "He's at home with the flu."

[netmeg]

Or even - it's someone else's responsibility.

[reseller]

Matt Cutts has just tweeted for 3 hours ago about "cool Chrome extensions& Android apps for Google Tasks". No mention of any Hummingbird! Don't you find it strange?

Keep in mind that Matt Cutts use to be the Googler who bring us news and updates about matters related to changes in search infrastructure and algorithm changes.

[lucy24]

If you noticed, there is a link there where you can "report" incorrect information. Of course that's assuming that you actually know the answer yourself and are 100% right. If you get 50% of people who are wrong, but report the right answer as being wrong, I think you see where it's headed.

Oh, ###, it's Google In Your Language all over again. If three people* click past a translation, they stick it with the "we are very confident" boilerplate-- even if it's out-and-out typographic garbage-- and you can forget about trying to correct it. Machine translations of course have the same problem. "I've got a better translation. It's what I was taught in school in 1938 so it must be right."

If you provide a simple unconditional answer to a question whose real answer is "it depends", have you provided a service? What if someone asks about the causes of {historical event} or the reasons for {cultural institution} and the tool gives back the #1 current most popular answer without mentioning that there exist others?


* I don't know the actual number, but based on personal experience it's got to be extremely low.

[mcskoufis]

I am wondering whether what Singhal implies is that now long-tails are 90% of queries performed @ Google?

Since it is related to complex queries...

So high volume keywords now make up less than 10% of the total?

[EDIT:] Removed a word [/EDIT]

[jvmills]

Hummingbird has hit my data/informational sites. Lost around ~20% of traffic. Started on 20/21st August steady drip away of traffic.

Own around 20 data driven sites, still waiting for a settled pattern. Assume more semantic markup is the key?

[jvmills]

Or option b) open a beer and catch up with Breaking Bad on Netflix...

And why am I marked as a new user? I joined this forum in 2005!

[jmccormac]

This is not the end of the Web, or a face-saving gesture to save Google embarrassment over past mistakes.

It does provide a very useful opportunity for Google to roll back problematic tweaks to its algorithm and the way that some people have, apparently, noticed an uptick in traffic may not be solely attributed to query parsing and handling. The problem with this "upgrade" is that the Google fanboys and fangirls will continue to drink the Google koolaid whereas the analysis will be left to those who are adversely affected and to those who know about Search. But as with many of Google's algorithmic tweaks and changes, some people may lose their businesses or revenue and others will gain.

I still think that Google's shift away from a link orientated algorithm is due, in part, to its failure to deal effectively with link spam typically found in hacked sites and link spam due to backlink building schemes. It simply may not be able tell what is a good outbound link and what is a spam outbound link. Some compromised websites are still in Google's index. (This is because the owners of the compromised websites often treat their sites like brochureware and only update them annually if at all.) But then that's based on working on algorithms to detect compromised sites and this kind of linking rather than on reading the latest cargo-cult SEO news. It is still too early in the rollout of this latest tweak to be certain of anything.

Regards...jmcc

[EditorialGuy]

I still think that Google's shift away from a link orientated algorithm is due, in part, to its failure to deal effectively with link spam typically found in hacked sites and link spam due to backlink building schemes.

If you depend on SEO for a living, it's easy to think that everything Google does is a reaction to SEO, but the odds favor a common-sense explanation:

Hummingbird is simply an evolutionary step to make search friendlier, more appealing, and useful to people who aren't information-retrieval experts or librarians--i.e., the kind of people who go to forums, Twitter, or Facebook to ask questions like "Why is my car leaving oil stains on the garage floor?" or "What does ringworm look like?" or "Can anyone recommend an all-inclusive naturist resort in Kazakhstan?"

Why shouldn't a search engine be useful for those people, and not just for the search-savvy users who frame questions as "keyword AND keyword OR keyword"?

Just as important, it is realistic to expect the average user's one- or two-keyword search to yield useful results from a Google index that has grown exponentially, from about one trillion pages to more than 30 trillion pages, over the past five years alone?

[diberry]

Why shouldn't a search engine be useful for those people, and not just for the search-savvy users who frame questions as "keyword AND keyword OR keyword"?

Maybe everyone I know is a genius (I don't think so), but more and more people are moving away from using Google (and Bing for that matter) for simple queries that could better be answered by directly visiting Wikipedia, Amazon, Yelp etc. Or even Pinterest, which is great for "scarf tying" or "pie recipes" type searches where you're open to anything in that topic.

We only use search engines for research into information or stuff where we don't know of any site that will have our answer.

My young Millennial friends are ESPECIALLY good at collecting various websites they can go straight to for certain types of stuff. They have their time-wasting sites, their fact-checking sites, their info sites, their "what does someone else think about this" sites, their shopping sites, etc. They ONLY use search when they haven't got a clue or all those sites have failed them, and by its nature, that's going to be a more complex query than "When did Elvis die?" or "Where pray tell can I purchase DVDs for I only recently arrived from Alpha Centauri and don't know about Amazon?" And I don't see ANY of the engines rocking on these kinds of queries anymore.

In short, Hummingbird would have been great in 2002 when Baby Boomers and older didn't understand how to phrase queries. Those people have adapted or moved on. The Millennials see "search" in general AS a library function. Google's missed that boat, IMO.

Don't get me wrong. I don't care what Google does because for about 7 years I've been building my businesses as if Google doesn't exist. I just think these instant answers are not what anyone thinks search is about. I'd never have seen them if y'all hadn't pointed them out to me.

Increasingly, people see search as a replacement for libraries, and other sites as having the answers to more simple queries.

[jmccormac]

If you depend on SEO for a living, it's easy to think that everything Google does is a reaction to SEO

As I pointed out above, the link spam explanation is based on analysing the content of hundreds of thousands of websites and developing algorithms to deal with the issue. My viewpoint is that of a search engine algorithm developer rather than just some end user dabbling in a bit of SEO and reading a few SEO blogs.

Google's original approach of using the numbers and authority of inbound links to measure the authority of a site put it ahead of every other search engine. The problem was that it was going to be gamed. Any human designed system is going to be gamed when there is a financial advantage to doing so. SEO is a part of this. Merely denying that there is an issue because you can't see it (most compromised sites with link spam do not display these links to human users) doesn't work. But the primary target of the link spam is Google's algorithm. And the highly contested keywords like payday loans etc show the problem that Google is facing. Playing 'whack a mole' with these spam links just doesn't work. Because of the nature of most of the web (brochureware and rarely updated), many of the compromised sites stay compromised and stay in Google's index.

Hummingbird could also be a reaction to the rise in the trust in social networks. Google's G+ has not been anywhere as successful as Facebook and Twitter. Google's plundering of Wikipedia for its "knowledge" graph is also an indication of how crowdsourced knowledge is hurting Google's search product. The trust people had for Google being able to answer questions may be diminishing while the trust in social networks grows.

As for the trillion page index stuff, Search fragmented years ago. Google used to be a great general search engine but people search differently now and there are general, social and local search elements. The multi-trillion page index makes for great press releases for the technology churnalists to recycle like they have a clue but the reality is that a lot of those pages will be noise and duplicate content.

A better query parser will buy Google some advantages over its challengers but it still has the link spam problem. Moving away from a linked based algorithm towards a "concepts" type algorithm would devalue a lot of the link spam type links and would provide some kind of immunity to this kind of gaming. Google tends to tell people (media and webmasters) what it wants them to know rather than everything. It relies on its FUD buddies in the cargo-cult SEO business and its fanboys and fangirls to accept every one of its "explanations" without question. However the "trust but verify" aspect amongst more seasoned webmasters has become more apparent of late. And if one's business depends on it, that attitude is essential.

Regards...jmcc

[EditorialGuy]

We only use search engines for research into information or stuff where we don't know of any site that will have our answer.

Nothing new about that. My generation used search in the same way back in the '90s. It was only later, when the Web audience became more mainstream, that people started using search engines as entry points to sites they already knew (e.g., typing "Yahoo" into a search box to reach yahoo.com).

In any case, as the Web grows larger and the search audience continues to become more diverse, it makes sense for Google and other search engines to help mainstream users do "research into information or stuff" more easily (and, ideally, with results that are driven by concepts and not just by keywords).

To use an analogy, search engines have been a lot like the HTML editors of the early and mid-1990s. Now that we're in the WYSIWYG era, it's time for search engines to be less like HomeSite or HotDog and more like Blogger or WordPress.com.

[jmccormac]

t was only later, when the Web audience became more mainstream, that people started using search engines as entry points to sites they already knew (e.g., typing "Yahoo" into a search box to reach yahoo.com).

That was mainly due to the fact that various browsers had some search engine pages as their default homepage.

Regards...jmcc

[reseller]

So it seems Google has been running Hummingbird for a few months not just last month!

Update: After contacting Google’s press department I was informed that Google has been running Hummingbird “for a few months,” not just since last month. The press department also reiterated their guidance for those doing SEO, “Our guidance to webmasters is the same as always — we encourage original, high-quality content, since that’s what’s best for web users.”
[forbes.com...]

[diberry]

jmccormac, that was a fantastic post for those of us who don't have your depth of knowledge and expertise about building engines. Thanks for your insights!

I very much hope Google moves to a more "concept" based ranking system that can't be gamed. That would be awesome for everyone except those who game Google for a living, and well, they have been warned the day was coming. ;)

I don't mean to be overly critical of Google - I hope that under the hood they've come up with something great. I just think this instant answer doodad means Google believes dumbing search down will make them more competitive against other engines, but they might be underestimating how many "simple answer" sites they will then be competing against. I don't think this is their best strategy, but I could be wrong.

[mrguy]

I very much hope Google moves to a more "concept" based ranking system that can't be gamed. That would be awesome for everyone except those who game Google for a living, and well, they have been warned the day was coming.

I think all of us "game Google" in one way or another in trying to get traffic to our sites.

If you're just following their guidelines, then your doing it to gain traffic and that's gaming Google is it not?

I think as time goes on, we are going to see a lot less Google dominance in providing traffic to websites. In fact, it's already being seen by many.

That's bad news for Google because without Adwords revenue, they are toast.

[diberry]

I think all of us "game Google" in one way or another in trying to get traffic to our sites.

If you're just following their guidelines, then your doing it to gain traffic and that's gaming Google is it not?

No, I follow their rules to avoid losing traffic by having them penalize me - same as I follow Pinterest's rules, Facebook's rules, etc.

[lucy24]

doing it to gain traffic

No, I follow their rules to avoid losing traffic

"gain" vs. "not lose" is a pretty hair-splitting distinction.

[EditorialGuy]

If you're just following their guidelines, then your doing it to gain traffic and that's gaming Google is it not?

Nope.

I think as time goes on, we are going to see a lot less Google dominance in providing traffic to websites. In fact, it's already being seen by many.

Yes, I keep reading here that everyone anybody knows is switching to Bing. Now, if only someone would tell comScore Media Metrix and Hitwise!

[diberry]

"gain" vs. "not lose" is a pretty hair-splitting distinction.

From an intent standpoint, it's the same as the distinction between simply not speeding in order to avoid speeding tickets vs. speeding but doing everything you can to avoid getting caught.

However, it clearly a hair Google doesn't always split correctly, or we wouldn't have had so many people hit by Penguin and having no idea what they'd done that constituted "spam."

EditorialGuy, if you added together everybody known to everyone here at the forums, I'd still not expect more than a 1% shift in the percentages toward Bing. ;)

[viral]

EditorialGuy, if you added together everybody known to everyone here at the forums, I'd still not expect more than a 1% shift in the percentages toward Bing. ;)

Diberry what we need is for Kevin Bacon to switch to bing!

[atlrus]

I personally welcome all these "knowledge" changes at Google, as it will ultimately spell their demise with the users.

One of the main points of early Google was that it was really simple. Search for something and get a clean looking page with 10-14 links. Today G has added (and keeps on adding) so much bull - it's a nightmare for the user.

It's information overload in its best - just search for something like "hotels Palm Beach" and see the mess Google has turned into. I was literally forced to go to TripAdvisor because Google left me just starring at that mess, not knowing where to even begin :)

While I admit that something like "100 ft to meters" is not as messy, the direction has been set. They don't make money from providing answers, so new ways to clutter the "knowledge" graph with ads will be constructed. Google is making it more and more complicated for the user who simply wishes to search for something (anything), especially with the constant push towards G+, something that people do not want nor need.

Now if we can only get one good competitor to emerge...

[jkm0449]

Google Hummingbird is yet another disaster for me. It has:

a) encrypted all search traffic queries so that none of the tools - including Google Analytics, Webmaster and Adwords (where it depends on Webmaster) are work.

b) I have had no data in Webmaster since 9/23/2013

c) Analytics is essentially worthless now from an analysis standpoint. It no longer knows what keyword a searcher used to find my web site.

Google is destroying my business. Our search traffic is down 50% since September 2012. We have never used any 'black hat' SEO, but we are a business to business site. Most of our customers think Tweating is what birds do. Google's insistence that we be a popular site on the social media networks is untenable. (In fact, most businesses discourage social media interaction while at work)