Google x Google

June 5, 2009 3:13 AM

Google has released an experimental search tool, Google Squared, that presents search results in the form of a table. Each column represents some attribute or dimension of the things returned - for example, searching for US presidents yields a column for date of birth, and rows for Abraham Lincoln, George Washington, etc.

Being experimental, Squared does well on some things but (perhaps amusingly) less well on others.

(The precise results keep on changing due to curation and whatnot, so things may have changed by the time you read this.)

The idea of returning structured search results is hardly new or unique to Google, of course.

posted by Zarkonnen (70 comments total)

This looks like it'll be fun to play around with for a while. Unintended humor will surely be had as "midwest states" returns telephone numbers in the fourth column. I can't wait to dial up Illinois.

This really is cool though.
posted by Phantomx at 3:40 AM on June 5, 2009

On what planet is Kansas in the Midwest?
posted by leotrotsky at 3:49 AM on June 5, 2009 [1 favorite]

This one, as defined by the Census Bureau.
posted by stavrogin at 3:57 AM on June 5, 2009

Oooh, this is fun! I'm enjoying entering in whatever random topics come into my head.

Successful squares: Drew Barrymore films, candy bars, ungulates

Pretty good: neural tube defects, brands of toothpaste, Ben & Jerry's Flavors

Less good: live action Disney films, Schedule II drugs

Not even close: Star Fleet Captains, Homosexual heads of state

Bonus: All game shows are Canadian.
posted by Captain Cardanthian! at 3:58 AM on June 5, 2009 [1 favorite]

A couple of my links didn't work out, apparently. Brands of toothpaste, live action Disney films. Let's see how that does.
posted by Captain Cardanthian! at 4:04 AM on June 5, 2009

On what planet is Kansas in the Midwest?

Midwestern States
posted by rokusan at 4:11 AM on June 5, 2009

Yes... yes, we could get a lot of mileage from this.
posted by rokusan at 4:12 AM on June 5, 2009 [1 favorite]

Actually, "Starfleet Captains" works pretty well if you spell "Starfleet" as one word, and ignore the fact that the results, before getting to James T. Kirk, first lists Jed Clampett.

Who struck it rich when he was out shootin' up some food, and up from the ground came a sizeable dilithium deposit.
posted by bicyclefish at 4:17 AM on June 5, 2009 [2 favorites]

Touché.
posted by Captain Cardanthian! at 4:21 AM on June 5, 2009

It out-wolframs Wolfram Alpha.
posted by HeroZero at 4:38 AM on June 5, 2009 [1 favorite]

It seems to have the internet nailed.
posted by Luddite at 4:47 AM on June 5, 2009 [13 favorites]

HeroZero: "It out-wolframs Wolfram Alpha."

Please stop thinking that everything on the web with a one-line text entry box is a search engine. W|A is a computational knowledge engine.

I saw Squared yesterday, and a brief test with Digital Cameras and Hard Drives gave pleasing, if occasionally screwy results...
posted by benzo8 at 4:50 AM on June 5, 2009

heh, search engines?
posted by thatelsagirl at 4:53 AM on June 5, 2009 [2 favorites]

Luddite winz teh Internetz.
posted by kalessin at 4:54 AM on June 5, 2009

On what planet is Kansas in the Midwest?

Um. Earth?
posted by middleclasstool at 5:00 AM on June 5, 2009 [1 favorite]

It out-wolframs Wolfram Alpha.

Eh, I dunno. Alpha may be incomplete, but it's more authoritative. With Alpha, some real live person decided some bit of information was true and put it into the database. This is more like those sites that scrape the web, fill some crappy web page with stolen content, and sell ad space.

It's funny--this would seem completely antithetical to Google's oft-stated goal of getting you off their search results as quickly as possible. Why bother clicking through to Wikipedia, or about.com, or wherever, if this has conveniently scraped all the info you need from those sites?
posted by Garak at 5:01 AM on June 5, 2009 [1 favorite]

Please stop thinking that everything on the web with a one-line text entry box is a search engine. W|A is a computational knowledge engine.

He very clearly did not do that. But since we're all helpfully correcting each other, please stop thinking that "computational knowledge engine" means something.
posted by Garak at 5:05 AM on June 5, 2009 [4 favorites]

Sins--Now with more Christ!
posted by Captain Cardanthian! at 5:07 AM on June 5, 2009 [2 favorites]

The results for parties are not exactly what I was expecting.
posted by burnmp3s at 5:08 AM on June 5, 2009

I wonder if Neil Gaiman knows he's been dead for 15 years?
posted by Captain Cardanthian! at 5:23 AM on June 5, 2009

Liquid densities

Water is 1.00 and oil is 885. Useful to know!

(Oh wait, if you click the squares the units show up, give alternate values and even assign a confidence. OK, that's pretty cool.)
posted by DU at 5:45 AM on June 5, 2009

Yes, this is obviously an attempt to catch up with Wolfram Alpha, and no, it doesn't come close.

For those who still don't understand what Wolfram Alpha is: go there and type in "next solar eclipse visible from Detroit, Michigan", or "distance from Earth to Jupiter on July 4, 1776" or "temperature in New York City on December 25, 1970" or "age of Abraham Lincoln in 1865". It has nothing to do with Google; it's trying to be the computer from Star Trek.

It's natural language parser needs work, and there are gaps in its database, but hey, it's brand new - and it's still a much more impressive entrant in this arena than Google Squared.
posted by ixohoxi at 5:46 AM on June 5, 2009 [2 favorites]

Ok, now we're talkin'! (all nsfw)
posted by sambosambo at 5:48 AM on June 5, 2009

The square for planets is pretty good, though the Description column gives (along with the correct descriptions for other planets) "While you're shopping for Saturn vehicles, we've given you an easy way to save your information..." and "Only at Venus, find the sexiest women's swimwear and clothing."
posted by Johnny Assay at 5:51 AM on June 5, 2009

I like how under the row for "full name" they have George Washington's full name listed as "Mr. George Carver".

I didn't know that.
I also didn't know that John Adams, Abraham Lincoln and Thomas Jefferson ALL had the same full name - "No Value Found".
posted by dirtdirt at 5:54 AM on June 5, 2009

"Chuck Klosterman - no value found."
Now I know Google is on my side and I trust it completely.
posted by TheGoldenOne at 6:08 AM on June 5, 2009 [1 favorite]

On what planet is Kansas in the Midwest?

I was going to say "Um, Earth" but middleclasstool beat me to it.

The one for sex is pretty entertaining.
posted by delmoi at 6:10 AM on June 5, 2009

Yes, this is obviously an attempt to catch up with Wolfram Alpha, and no, it doesn't come close.

Have you actually tried using Wolfram Alpha?

For those who still don't understand what Wolfram Alpha is: go there and type in "next solar eclipse visible from Detroit, Michigan", or "distance from Earth to Jupiter on July 4, 1776" or "temperature in New York City on December 25, 1970" or "age of Abraham Lincoln in 1865". It has nothing to do with Google; it's trying to be the computer from Star Trek.

Of course, if you actually have a query that's not based on their specific examples you'll just get

Wolfram|Alpha isn't sure what to do with your input

We know what wolfram alpha is trying to be, and it pretty much failed.
posted by delmoi at 6:14 AM on June 5, 2009 [4 favorites]

Can you imagine anyone actually using this?

At least Wolfram Alpha is pretty good at telling you when it doesn't know.
posted by smackfu at 6:15 AM on June 5, 2009

...and Tony Blair was apparently dead throughout his entire tenure as Prime Minister...
posted by Zarkonnen at 6:16 AM on June 5, 2009

I work for KPMG. So I typed in 'KPMG'. No KPMG related results and top of the 'square' is PricewaterhouseCoopers.

An oddly upsetting fail.
posted by jonnyploy at 6:18 AM on June 5, 2009

I am curious as to how we can tell this is attempting to be similar to Wolfram Alpha from the fact that it is entirely unlike Wolfram Alpha.

Perhaps the class of Wolfram Alpha-like things is defined as anything with a text box that is not a plain vanilla search engine.

Can you imagine anyone actually using this?

Yes, once the kinks have been ironed out and people figure out what it's good for. I'm guessing running custom-built, saved squares to scrape information such as retail prices is where it's at.
posted by Dr Dracator at 6:24 AM on June 5, 2009

delmoi: none of those are based on their examples; I made them up myself. Then again, I'm a programmer, so I guess I'm used to constructing syntax that a computer will understand.

I agree that WA isn't exactly useful yet, but it's a damn impressive step. To a programmer, at least.
posted by ixohoxi at 6:35 AM on June 5, 2009

Here is the thread we had when wolfram alpha went live, you'll find a ton of failed queries there. for example:

Here are some queries that didn't work for people

_{"Diamond Bar zip code" (really?)
"slayer albums"
"alpha launch date"
"greek alphabet"
"cat species name"
--
who killed Kennedy?
who wrote leaves of grass?
when do babies learn to walk?
how many americans make more than $100,000?
--
"Number of wild pelicans?" Nope.
"Wild pelican population?" Nothing!
"Pelican population?" The population of Pelican, Wisconsin is 2676.}

Here are the ones I tried first:

<sub>what is the hermitian of the 10 by 10 identity matrix?
Wolfram|Alpha isn't sure what to do with your input.

what is the transpose of the 10 by 10 identity matrix?
10^transpose (10 IdentityMatrix[n])

That one is kind of interesting, but not only is not computed, it's also wrong.

what is the transpose of the 10x10 identity matrix?
10^transpose x^10 IdentityMatrix[n]

Still not parsing the string correctly.

how many co-sets does the set of integers mod 15 have?
Wolfram|Alpha isn't sure what to do with your input.

Okay, I haven't gotten one useful result out of this thing so far.

If all this thing can do is look up basic statistical information and manipulate it, well, you can do that with Google spreadsheets already, actually.

how old is the creator of mathematica
Wolfram|Alpha isn't sure what to do with your input.

Weak.

who created mathematica?
Input interpretation:Mathematica 1.0 Released | people involved
result: Stephen Wolfram

Which means the last one was actually a parsing failure, rather then a lack of knowledge.

how old is stephen wolfram?
49 years 8 months 16 days</sub>

Here's more stuff that I tried:

<sub>*It dosn't know how many pine trees are in new york. (fail)

*It dosn't know who created metafilter. (fail)

*It knows how old Bill Gates is, and it knows how old Stephen Wolfram is, but it dosn't know who's older. (fail)

*I asked it how many sea ports there were in somalia, and if named two of them: Boosaaso and Kismaayo (well that dosn't tell me how many there are, but now we know it's more then two, which is more then we knew before, i guess)

*It dosn't know how many pirates there are in the world

*I got an interesting answer for "how many people in iowa and new york", it gave results for both states and gave ratios and comparisons. But didn't add them for me.

*"how far is the horizon when you are 30,000 feet above the earth?" was something I was actually wondering about the other day, and this thing had no idea.</sub>

I agree that WA isn't exactly useful yet, but it's a damn impressive step. To a programmer, at least.

Well, right, it's "not useful". SHRDLU came out 39 years ago, their natural language parsing isn't really that impressive at all. What WA is doing might be impressive because software doesn't usually try to accept input that way because it usually doesn't work, just like it usually doesn't work with WA.
posted by delmoi at 6:55 AM on June 5, 2009 [3 favorites]

Er, what I meant to say was: What WA is doing might seem impressive because software doesn't usually try to accept input that way because it usually doesn't work, just like it usually doesn't work with WA.
posted by delmoi at 7:00 AM on June 5, 2009

Look up US vice presidents...
posted by bink at 7:12 AM on June 5, 2009

Hey! I never knew the Yankees were a borough.
posted by netdpb at 7:13 AM on June 5, 2009

Twister: Harmless fun or deadly robot war?
posted by No-sword at 7:16 AM on June 5, 2009

GOOGLE! STOP! No seriously, stop. Step away from your 500,000 computers and labs and servers and stop doing the next 500 best things. For like, at least 8 hours. Every night.
posted by iamkimiam at 7:19 AM on June 5, 2009 [1 favorite]

Cheese is pretty good.

American auto industry, not so much.

Also: awesome, hilarious, fantastic square for Sarah Palin!
posted by iamkimiam at 7:26 AM on June 5, 2009

The genres of the Muses!

Calliope: Jazz
Polyhymnia: Folk
Erato: Classical
Thalia: Latin
Euterpe: Rock
Terpsichore: Biography
Melpomene: Orchestral
posted by Iridic at 7:31 AM on June 5, 2009

I especially like how Persuasion was apparently set in the US.

But adding rows and columns is pretty neat even if I couldn't get Captain C's "authors" query to add a column that had a list of major works - I tried "major works", "works", "books", "novels", and "bibliography", with some amusing results.
posted by clerestory at 7:33 AM on June 5, 2009

This is awesome, but I can't help the feeling that this is going to become another Google Sets, that is, an awesome tech demo with no practical application that remains in Labs forever.

And I like Sets.
posted by JHarris at 7:33 AM on June 5, 2009

I remember having a conversation with friends about ~8 years ago about text messaging. We couldn't think of a damn use for it. Pagers were long gone, and that seemed like the closest thing. I remember that we were laughing over this cell phone company commercial where a family was sitting around a dinner table using their cell phones to play "duck, duck, Goose." It was as if they couldn't figure it out either!

Now I can't imagine living without it.
posted by iamkimiam at 7:39 AM on June 5, 2009

I also didn't know that John Adams, Abraham Lincoln and Thomas Jefferson ALL had the same full name - 'No Value Found'.

Nihilists! Fuck me. I mean, say what you like about the tenets of National Socialism, Dude, at least it's an ethos.

Look up US vice presidents...

Vice President George Clinton was pretty amazing. He was born in 1947 and died in 1761, and he's considered one of the Founding Fathers of funk.
posted by kirkaracha at 7:44 AM on June 5, 2009

I wonder how the images are selected for these sets. The pictures used for colors (black in particular) left me scratching my head
posted by oozy rat in a sanitary zoo at 7:52 AM on June 5, 2009

I like how the "colors" square doesn't include any of R, G or B, plus all the colors have prices.
posted by DU at 8:11 AM on June 5, 2009

I wonder how the images are selected for these sets.

The images just seem to be the first Google image search result for that term.
posted by Zarkonnen at 8:13 AM on June 5, 2009

Metafilter prices range from $11.95 to $2,051,063.00.
posted by ardgedee at 8:14 AM on June 5, 2009

I was poking arount with this the other night, and it really looks like this is an extension of the old beta, Google Sets. Enter in something it doesn't know about and you get back a form that I'm sure was in sets.
posted by boo_radley at 8:36 AM on June 5, 2009

Metafilter prices range from $11.95 to $2,051,063.00.

Same as in town.
posted by cairnish at 8:46 AM on June 5, 2009

Dr Dracator: WA and Google Squared are comparable because they provide natural-language access to a large body of structured data. They differ in the kinds of data they catalog (WA is strictly quantitative stuff; GS appears to be a little looser), and the kind of processing they can perform on it (WA = lots; GS = apparently limited to building two-dimensional tables), and where they get the data (WA is curated; GS, apparently, is scraped from the web)—but they're still natural-language query interfaces to large bodies of structured data.

This is a new category of information portal—one whose form and mechanics are still being worked out, and one that will probably see a lot of growth and refinement in the next few years. GS is obviously not the exact same thing as WA, but it's part of a larger effort to take data search to the next level—to make it smarter, quicker, more powerful, more accurate, more immediate, and more everything.

delmoi: I already pointed out the shortcomings of WA and said that it isn't terribly useful yet, so I'm not sure what you're trying to convince me of.

It's the fact that we can even approach a project like this that impresses me. Yes, these are imperfect, experimental attempts, but so were the first airplanes and computers. I'm not evaluating them as practical, market-ready tools, but as steps toward that. To my knowledge (and I might be wrong) WA is the most comprehensive attempt at this sort of thing yet—and imperfect though it may be, it's still the best one yet. If you've written a natural-language parser that improves on WA's, I'm sure they'd love to hear from you.
posted by ixohoxi at 8:46 AM on June 5, 2009

Are we sure Google Squared is doing some kind of language processing ? I thought they're guessing for the column values using the data returned by the query.
posted by Dr Dracator at 9:01 AM on June 5, 2009

...and Tony Blair was apparently dead throughout his entire tenure as Prime Minister...

Same in Canada. While our current prime minister is still alive, his immediate predecessor died sixteen years before taking office, and the two PMs in turn before him both died as infants. Going back to before that, I note that Brian Mulroney, Prime Minister from 1984 to 1993, passed away in 2007, which would at least account for his unhelpful answers last month in front of the commission investigating the Airbus scandal.
posted by ricochet biscuit at 9:09 AM on June 5, 2009

It no work.
posted by MarshallPoe at 9:35 AM on June 5, 2009

Hey! I never knew the Yankees were a borough

Oh, haven't you seen the new stadium?
posted by deliquescent at 10:31 AM on June 5, 2009

Dude this thing is worse than bing.
posted by Mister_A at 10:38 AM on June 5, 2009

The results for parties are not exactly what I was expecting

I will now mark "Christmas Party" as my political affiliation, or maybe the un-religious "Holiday Party" option.
posted by filthy light thief at 10:39 AM on June 5, 2009

The results won't load for me. Not in Chrome, Firefox, IE6, IE8... not on any of my computers. :'( Am I the only one?
posted by brenton at 11:59 AM on June 5, 2009

It has good taste in horror movies.
posted by brundlefly at 1:12 PM on June 5, 2009

Oh man, they're right, every library needs an anthem!
posted by unknowncommand at 3:40 PM on June 5, 2009

If you've written a natural-language parser that improves on WA's, I'm sure they'd love to hear from you.

Well, they don't appear to have published anything so how would one even be able to compare it current academic research, let alone anything newer or more advanced. WA is only impressive if you're not up to date with what's already out there.
posted by delmoi at 4:24 PM on June 5, 2009

The search for programming languages works really well. But then I noticed that all the fields are populated by wikipedia hits. I bet this thing would totally rock if you could just search wikipedia for it.

So, instead of wolfram alpha's attempt to create a database (which they'll have to maintain!) of structured data, you could just work out a way to leverage wikipedia as the structured data backend and build something like google squared off of it. Too bad it doesn't respect "site:"
posted by heathkit at 5:08 PM on June 5, 2009

I dunno, I think it's pretty smart.

Worst Presidents Ever
posted by empath at 5:33 PM on June 5, 2009 [1 favorite]

Memes
posted by empath at 5:36 PM on June 5, 2009

Getting these sorts of things to return correct results will get easier as the web becomes more semantic.

Unfortunately, all I ever hear about the semantic web is talk about how great it'll be.
posted by philomathoholic at 6:43 PM on June 5, 2009

Worst Presidents Ever

I am so tired of fucking Anti-Masonic Whigs and their shit. OH, LOOK AT ME, I HATE MASONS, OH, I SUPPORT THE SUPREMACY OF CONGRESS OVER THE PRESIDENT AND FAVOR MODERNIZATION OVER ECONOMIC PROTECTIONISM I'M SO AWESOME.

Dicks.
posted by middleclasstool at 6:50 PM on June 5, 2009 [1 favorite]

Re: Kansas and the Midwest.

As a native Kansan, I can affirm that people seem to be confused by the term. Many assert the plains states should be Great Plains or Central. Which pretty much leaves Midwest for Ohio and Michigan. And would leave the Midwest east of central.
posted by pwnguin at 9:32 PM on June 5, 2009

A simple example: the Wikipedia talk page for the Midwest.
posted by pwnguin at 9:33 PM on June 5, 2009

This is more like those sites that scrape the web, fill some crappy web page with stolen content, and sell ad space.

Oh my god. Of course. It's fiendishly brilliant.

They already have all the ads. Now Google is trying to replace every page that has ads on it.
posted by rokusan at 2:07 AM on June 6, 2009 [2 favorites]