Live Blogging Google Searchology 2009

Almost an annual event, Google Searchology is a time when Google struts its search stuff in front of assorted reporters. Today’s event gets going at 10am Pacific time. I’ll be live blogging from this page, and you can also watch the event yourself via this webcast. You can also follow @google on Twitter, which will be live tweeting during the event. The official hashtag is apparently #searchology09. Now there’s a mouthful!

Here’s the schedule of speakers:

• Udi Manber, VP, Engineering (Core Search)
• Patrick Riley, Software Engineer
• Scott Huffman, Engineering Director
• Marissa Mayer, VP, Search Products & User Experience
• Nundu Janakiram, Associate Product Manager
• Alex Komoroske, Associate Product Manager
• Kavi Goel, Product Manager
• John Taylor, Software Engineer
• Q&A

And we’re now being welcomed. And here comes Udi.

Is search the new rocket science? 20th century dream was to conquer nature. 21st centry will be about “understanding people,” and he thinks search will be a big part of that, possible leading the way

“Our job is to do rocket science that will be taken for granted,” his slide.

His daughter was searching for “griffing music palo alto” and it found “Gryphone Stringed Instruments,” what she wanted, so they did the right thing seemlessly.

“The real goal should always be to solve the user’s problem,” another slide.

If users can’t spell, it’s our problem. If they don’t know how to form the syntax, it’s our problem. If there’s not enough content, it’s our problem.

tsmu is another search, and it takes a long time to load www.tsmu.edu, so if you search for that, there’s an option called [skip intro] to let you save time.

Points out how sitelinks under IRS site listing can save you time getting to the right place. Or how doing a search for an address with no web page can still get you listing.

Showing how Google FINALLY AND BELATEDLY [my emphasis] added earthquake results that will appear.

Search for “sf giants” brings up latest score last night.

Lots of progress on the technical infrastructure, so now most promising advance is being able to “understand.” So “no ‘rockets’ to highlight yet, but it’s a start.

Patrick Riley now up on “Making ‘Did you mean’ Even Better.” You know, that link that comes up that suggests a correct spelling.

They wanted to improve it. Showing a search for “labor” and how you get “see results for: labor and delivery” and other results off the main topic. Apparently, this feature (which has been around for two years now), is known internally as “chameleon.” Scratch that — it’s apparently called spellmeleon.

The showed these middle of the page “different” results in testing, found users seemed to like it.

Doing these things are processor intensive, they have to effectively double-up on the queries (you did one search, but they processed two on your behalf). This mean lots of good engineering to reduce load. But they still found it took a ton of processing but decided it was better for the users, so did it anyway [yeah, like they’d say let’s not].

Then they testing putting the related searches and spell corrections at the top of the page [you know, where like all normal search engines had been doing for years, but this was a brainflash at Google :) ].

Shows graph that lots of usage and that they thing it helps save time.

Scott Huffman now up on mobile search quality. Devices are a big challenge, there are so many. Literally hundreds of devices that they support.

Mobile is also inherently local. Peopel use mobile because of three key elements:

• Complete, have all of Google on it
• Easy, effortless to search
• Local, know where you are

Shows four different screens on same phone, with how Google auto-suggests, makes phone numbers clickable, image search optimized for touch-screen nature of some device [secret insider tip I picked up in lawsuit over Google Images — mobile search for images is often done by those who are touching things other than the screen]. Or product search drill down.

There’s also searching the actual mobile web, pages that have been sized down for mobile devices.

All that’s covering the “complete” part of mobile. Now onto the “easy” part with a live demo, for “ba 284” SF-London (I’m sure I’ve done that flight like a billion times actually). This he does on regular Google, showing how “track status” appears at top.

Now we’re looking at his phone on screen, and since he’s logged in, it’s recorded that “ba 284” query he did on the web has been saved. Not released yet but coming. Hmm. If you clear your browser history on your desktop, that doesn’t clear it on the phone, I bet. And I’m guessing it all depends on you allowing Google Search History to be enabled.

Now searching for sushi on phone and getting local listings automatically appearing, based on the phones GPS/cell tower location, with the ability to click and call.

And now it’s Marissa to talk about search in all its “shapes and sizes.”

She was at a dinner a year ago and man said to her that Google saved him. He had a new bow tie and no one to help him tie it, so he searched, found instructions with video on doing it, along with image results with descriptions. This is all Universal Search blending lauched two years ago.

When launched for an old vampire movie at the time, had little rich media. Today, if you search for the hot vampire movie Twilight, lots of rich media results. And this helps users understand answers on the web much more deeply.

Challenges in doing display, so they’ve worked on the “bento box,” those Japanese lunch boxes that I’m probably mispelling. Idea is a compact presentation of things like video, local, etc (I’ll try to add links to our past coverage of this later, Marissa talks fast, like a hummingbird flies).

Showing SearchWiki now, and they’re getting hundreds of thousands of annotations.

What are big challenges in search?

• Finding recent information
• Expressing that you want just one type of result
• Assessing which results are best
• Knowing what you’re looking for
• Expressing your searches in keywords

To address these, you need a “Slice + Dice” which means the new “Google Search Options” that’s going live apparently right now.

Go to home page, do a search on “hubble telescope” and get results that looks like normal, and you should see a search options link that’s new [I still don’t see it]. Hey, it’s Wonder Wheel! Link is to my past post about these search options in testing, in much depth. What’s new since then is a “search by relevance” versus “search by date” option.

We’re seeing lots of examples of how this works, but honestly, see the Wonder Wheel link above, and you’ll see the same type of stuff.

Now we’re learing about Google Squared (link to go live later this week). Can unstructured data be pulled from the web and put into spreadsheets. Apparently, Google today makes this a reality.

Does search for “small dogs” and results are listed in rows and columns, things like Jack Russell Terrior, its weight, type of dog group, under it a Maltese. I’ll post a screenshot in a sec.

(Click on photos to see larger images)

What if Google gets the “messy” data from the web wrong? Hover over a cell, and you’ll see the source with more info and the other sources for that value. You can also edit the information and save to your own personal account.

Going beyond squares, Google is unleashing rich snippets. See Google Testing Enhanced Listings, “Pagelinks” & Auto-Spelling Correction for our older article on this, and Google Search Now Supports Microformats and Adds “Rich Snippets” to Search Results for an updated look from today. RDF and microformats will both be supported.

And now we’re reaching for the stars. It’s a Google Android application (sadly, I have to send back my demo unit. Kind of overdue by months on that). It’s a starmap view developed on famed 20% time. You can find it in the Android market. Shows you the brightest starts and planets in the sky.

We’re looking for the constellation Orion, which Marissa suggests as her favorite. The one with the belt, right? My astronomy is really sad. Ah, the hunter, we’re reminded.

So why’s this better than a paper map. You knew there would be a catch. GPS in the phone means you can get a map that’s local to where you are, with a yellow horizon line. “This is an Android phone, so it can do more.” engineer John Taylor tells us. Take that, Apple. OK, he didn’t say that. But this is cool, point the phone at the sky, and you’ll see the stars where you’re looking listed, so you can find out more about this. Hmm, wonder if there’s an iPhone app that does that. Hey, I won’t even need to go outside an look.

Now we’re looking up Marissa’s star sign (Gemini, in case you’re wondering). Now a circle shows up and lets you direct the phone like a compass to find the constellation. IE, want to find where say Jupiter is? Search for Jupiter, then the phone will guide you where to look.

It’s Q&A time — folks watching online were able to submit some, as well.

Question about selling semantic keywords. Marissa says not quite applicable put pushes back on the notion that Google can’t do semantic — IE, understand meaning behind the words.

Question on international availability of features and are search engines limited to the online world. Udi says everything is extremely internationalized, can’t tell every product launches the same day in all languages, but the goal is always to launch everywhere. On offine, Marissa says that’s where they’ve started, not limited.

Back to semantic and why Udi seemed defensive about the word. I almost wrote how I hate it myself, and he explains the same — means different things to different people. Doesn’t go into depth on how Google does semantic analysis. Marissa dives in to talk about how Google Squared tries to find pages mentioning same topics/facts and tries to go with the facts that seem to be most common among them.

Online question on what Google’s doing to get “Deep Web” content. Udi says uses feeds, crawling, some other things.

Steven Levy says he won’t ask how they feel about Wolfram Alpha, but highlights how Wolfram was finding databases they were tapping into often had bad info. So is gathering info off the web getting good enough answers? (see Impressive: The Wolfram Alpha “Fact Engine” for more on this).

Udi says you have to look at all types of corroborating evidence, have to do a lot of work. As for Wolfram, he and Sergey Brin had the demo early on, asked to be confidential about is and “will respect that.” It’s OK, Udi — everyone’s talking publicly about the demonstrations they’ve seen. What do you think?

Is Squared a copyright issue, in that it pulls information but doesn’t send traffic to others. Marissa says they’re trying to do something transformative [see, if you transform things, that helps avoid some copyright issues]. But says she’s still pointing at the sources, which should drive traffic.

Question on if rich snippets can be exported by API. Marissa says that since the data is contained in the original HTML of pages spidered, people can get it from there.

Online question, when will Google Squared be available? Official word is later this month.

Rich snippets, will only sites be included to start with or can any sites be involved? Starting with limited set of sites, wanted to beta test it, but announcement was meant to say any site could start tagging up information. There’s going to be a submission form somewhere to perhaps get Google’s attention, but they might find them anyway. Watch Google Blog for more news later today, they say.

How long to the perfect search engine? Marissa repeats what she’s said before, search is in its infancy. There are many different branches and tacks you can take. Shows how big search is (which is handy for those pesky anti-trust people that Google wants to convince that it’s not a monopoly that needs to be regulated).

How long until the perfect mobile search? Seeing lots of changes quickly and just getting started, Huffman says. Udi says patent office quote in 1860s that said anything that can be invented has been. “We know better. You ain’t seen nothing yet.”

Cheerleaders now coming out, spelling G-O-O-G-L-E on stage. Heh. You wish.

Question on data retention. Marissa replies about being sensitive, keeping search logs for only 18 months (hey, that’s supposed to be 9 months now).

Elise Ackerman with good quesions. How the mobile search linked to the desktop search. Is it opt-in and out. If you do opt-in and delete on the browser, does it come off. Only works if logged in, and if you delete in either place, it gets removed in both. And does anything seem like science fiction to Udi? Putting in flight info seems so easy, but thinking back then, not so much.

And that’s it! For live blogging and coverage from others, see related stories on Techmeme.

Postscript by Barry Schwartz: Here are blog posts from Google on these announcements:

• A planetarium in your pocket, Official Google Blog
• Enabling Rich Snippets in Custom Search, Google Custom Search
• Introducing Rich Snippets, Official Google Webmaster Central Blog
• Leveraging rich snippets with Google Site Search, Official Google Enterprise Blog
• More Search Options and other updates from our Searchology event, Official Google Blog
• Sky Map for Android – A Mobile Planetarium, Official Google Mobile Blog