Deliberate Ambiguity

Finding Powerset Factz

Obviously, I don't need any more Powerset t-shirts, but I thought I'd show off some of the fun queries that I found and talk a little bit about the process I went through to find good queries.

Some knowledge about how Factz show up are helpful in finding them.  First, Factz are currently very simple "triples" of subject:verb:object. So, if you start asking questions with prepositional phrases (Who starred in Six Degrees of Separation?), a when (when did Mount St. Helens erupt?), or any other construction, you will get search results, but no Factz results.  Also, Factz most often return about popular subjects, because they are written about in many places in Wikipedia, not just their own Wikipedia article.

With that out of the way, let's look at an example.  I like chess, so a good starting point is a chess query: bobby fischer.

The Factz are grouped by relation and, even though by default Powerset only shows three results, there over 100 relations featuring Bobby Fischer as the subject.  Just click "more" to see all of the relations. 

The third relation is about Bobby Fischer defeating, which in turn suggests a simple question: who did bobby fischer defeat? (yes, I know that it should be "whom")

Though this isn't a complete list, it's a list of all of the sentences in Wikipedia that have Bobby Fischer as the subject for the verb defeat.  So, one way to get a good Factz query is to type in the name of a topic, look at the top relations, and then turn that relationship into a question.

However, sometimes Factz can also be used to get a list of subjects.  In this case, you can flip the question around to see who defeated Bobby Fischer.

Another good general principle about great Factz queries is thinking about a question that will generate a long list.  I watched The Incredibles for the first time last night.  This got me thinking about the query, who saves the world?

If you click "more" you'll see that this is a really long list with some that are expected (lots of heros saved the world), logical (Christ), funny (Godzilla), and wrong (Nova Scotia).  Note that Powerset always shows the sentence from which the Factz were derived, allowing you to double check whether the assertion is true.

Sometimes Factz can reveal things that are either not true or "almost" true.  For the query, who invaded Mexico, I was curious about Japan, since that bit of history evaded my memory.  Sure enough, it wasn't my memory that was deficient, but the sentence was: "On June 3, 1942, Patton believed the Japanese were on a course to invade Mexico."  Powermouse ignores these types of modalities and negations, even though Powerset indexes them properly.  We do this because we err on the side of giving users too much information, rather than too little.

Hopefully this will help you find some good Factz for Powerset's t-shirt contest.  If you have any questions, you can comment on my blog or drop us a line at feedback@powerset.com

Powerset crosses the Uncanny Valley

In case you haven't heard the news, Powerset launched today.  Since I can't sleep due to launch adrenaline, I suppose it's time to weigh in on Powerset unofficially.  It's fun to write a press release, but it's more fun to philosophize on my own time!

Last year, I came across the concept of the Uncanny Valley though an employee of PARC.  The basic concept runs something like this: as you make a robot more like a human being, people tend to respond with more empathy.  However, there's a dip in the empathy curve when the robot looks enough like a human, but has behavior incongruent with what we'd expect from a real human being.  We tend to forgive weird mannerisms (jerkiness, oddities in speech, etc.) when a robot looks like a robot.  When a robot looks like a human, we expect it to act like a human.

As an example, cartoon characters are usually more believable than computer generated characters, unless the CG is so perfect as to be indistinguishable from reality.

Keyword search engines sit somewhere on the left of the curve.  We're willing to accept the choppy "keywordese" language of a search engine because we appreciate the results that come back and we can make excuses when it goes awry: "It's just a computer.  It doesn't really understand me."  As with most technology products, the utility curve and the expectation curve are aligned exactly because we treat keyword search engines as an unfeeling technology.

Oddly enough, when a system is endowed with an understanding of meaning (call it "semantics"), we suddenly demand that the system act, think, infer, and comprehend just like another human.  Instead of just being a cold of technology, we begin to treat the system with agency and want it to act like other agents (i.e., humans). 

The question shouldn't (and can't) be about releasing a system that has a perfect understanding of language, because that perfect understanding is many years down the road.  How is it that companies like Radar, TrueKnowledge, and Powerset believe that we can create products based on imperfect technology?

There are two ways to stay out of the Uncanny Valley.  First, create unique features that are so compelling that users are willing to deal with imperfection.  Keyword search hasn't survived because we always get the right answer - plenty of searches yield nothing.  But, though we may get wrong answers, we often enough get answers that would never have been possible before keyword search.  Second, access to semantic features shouldn't require a significant change in user behavior.  No matter how good (or "natural") a feature is, if a user has trouble finding it, it won't get used.

People often think that NL search is doomed to failure both because computers don't understand enough and, even if they did, users won't start entering natural language queries.

I think Powerset has scored on both counts.  As an example, for topical queries like moses or machiavelli, Powerset assembles a summary of Factz based on our deep linguistic analysis of each sentence.  This kind of automatic aggregation of content just wasn't possible before Powerset.  But, users don't have to do anything special.  They just get Factz "for free" by using their keyword usage paradigm.

I'm really proud of Powerset's first product.  The debate on whether or not we've crossed the valley will continue, but one thing is certain: a better, more natural way of interacting with technology is coming. (Of course, whether we sit to the left or the right of the Uncanny Valley is the subject for another post.)

Bed time, finally!

My Untidy Bush

The LA Times today reported why I have an "untidy bush growing on my chin and upper lip area." (Hat's off to Frank Zappa for the phrase "upper lip area.") A number of employees here at Powerset have banded together for an underground project: Powerstache.  I, for one, am ready to go under the razor, but not until launch. . .

 

Powerlabs Screenshot Demystified (+ pr0n)

Earlier today, Steve Newcomb posted a screenshot of Powerlabs on his blog.  Though we're not going to be opening the doors of Powerlabs until September, I'm stoked at the response we've had so far.  In light of some of the blog posts that have popped up, I wanted to clear up two points and then make one of my own.  The quick points are: 1) The screenshot is of Powerlabs, not the Powerset search application 2) Yes, the P-word is a possibility, but not something we're actively exploring 3) Expect much more from us in the near future.

What in Lord's name is that screenshot?

It's not a screenshot of Powerset's search application, it's a screenshot of Powerlabs.  I made a video to explain what Powerlabs is, but I'll go into a little bit more detail here.  Powerlabs is a grand experiment to get feedback from users as early in the development cycle as possible.  We're going to open up everything: screenshots, demo applications, relevance tests, technology, UI mockups, and whatever else we can find.  Most companies hide their great ideas under a bushel until they launch, for fear that someone will steal their thunder or that the community will see things that aren't perfectly polished.  Because of that, a lot of sites that are in "Beta" miss the mark.  I firmly believe that Powerset will benefit from all of the encouragement, ideas, and criticism we'll get from Powerlabs.  Our search engine might not be a big surprise when we launch, but I'd trade "surprise" for "great product" any day.

So, what you see in this particular screenshot is a category within Powerlabs exploring how natural language technologies can benefit travel.  We'll ask questions through surveys, gather ideas from community, allow users to comment and rate those ideas, run design competitions for the ultra-motivated, and, of course, show off our concept products to get feedback.  Travel is just one of the many categories that we'll explore, which brings me to my next point. . .

You didn't just say the P-word, did you?

Well, yes, we did.  Except for the fun glossy effect that Siqi created, we didn't alter this image at all from when it was created several weeks ago.  Porn was originally stuck in as a quasi-joke category: we doubted we'd launch a porn category in Powerlabs, but everyone knows that search engines are often used to find various kinds of prurient content, even if it's not polite to mention it in conversation.

However, I apologize to those who were getting excited by the possibilities opened up by natural language porn searches.  Powerset is too busy right now working on getting our natural language engine to understand "politician" and "know" (not in the Biblical sense) to be perfecting more ribald vocabulary.  Sorry to disappoint =)

When we decided to release this image to the outside world, the question was, of course: do we take out "Porn."  The product team stood by our desire to "open up the kimono," as Steve would say.  If we're serious about showing the inner workings of Powerset, then y'all are going to see things that are good, things that are bad, and the in between.  But, we gotta remain authentic.  It's your feedback that we're interested in, which brings me to my final point. . .

Just a screenshot?  Show me the beef!

Powerlabs is all about being open and engaging the community in the process of creating a mind-blowing next generation search engine.  It's only proper that we apply this same philosophy to the building of Powerlabs itself.  This screenshot will be the first of many that we release.  The final version of Powerlabs will likely look much different than what you see today (and that's a good thing!).  So, hit us with your best shot and give us feedback.  'Cause I'm listening.

Disclaimer: Oh yeah, I'm sorta speaking in my official capacity as Powerlabs Product Manager, but this is my own opinion and doesn't necessarily reflect the opinion of my employer.  [In fact, I hope they're OK with this when they read it tomorrow!]

Powerlabs Video

Making videos is fun!  I showed the world that there are real people behind Powerlabs, I played with iMovie and my built-in iSight, and hopefully I'll score a few dates!  Here it is:

Though we're currently working on Powerlabs, there's still a chance to enter into the dialog and give suggestions.  As we get further along in the development process, we'll definitely be soliciting feedback from the users who have signed up.  This communication will be the first of many about what's going on inside Powerset.

In sending this video to our Powerlabs , I was reminded that one needs to be very careful and precise when sending out mass e-mails.  The link in the e-mail was broken, because the YouTube video was set to "private" instead of "public."  A coworker proffered a conspiracy theory in which Google's Anti-Powerset Team (GAPT) changed the setting.  Though tempting, the blame is certainly mine.  Mea culpa!

Launching Powerlabs is the first step in my goal for friendly competition with usability giant Marissa Mayer.  First Powerlabs, then Powerset, then maybe a penthouse at One Rincon Hill (MM live in the penthouse of the much-shorter Four Seasons).

Powerlabs is indeed coming soon

Last week, Powerset made a simple change to its homepage by adding a sign up for Powerlabs.  The world seems to be watching, because there's already a blog post about Powerlabs and we have over a thousand users signed up.  WOW!

In Powerlabs, you'll not only be able to see ideas that Powerset has about next generation search, but you'll be able to join in the discussion and contribute your own ideas.  Though Powerset is still at the early stages of our product development, we want to show the community what we're thinking.  A product that's conceived, created, and refined behind closed doors will never meet the expectations of the market when it comes out. Instead, testing lots of ideas (good and bad) with lots of users and soliciting ideas from the community will yield a mind-blowing user experience.

Sounds interesting?  Leave a comment and/or sign up at Powerset.com!

tags technorati : powerlabs powerset product design user experience

Google Downgraded (thanks, in part, to Powerset)

As reported by Tech Trader Daily yesterday, GOOG was downgraded by Trip Chowdry, of Global Equities Research.  Reasons cited included a number of problems at YouTube, including wildly inflated usage statistics, potential legal liability, and no good revenue model.  However, the wildest claim was surely this: "Early indications show that [the] YouTube distraction is now affecting Google’s Core Web Search business, with search momentum shifting toward competition including Microsoft, Yahoo and PowerSet in [the] consumer market. . ."

Though the market didn't seem to listen to this report (GOOG gained 14 points yesterday), I'm happy to be part of a company that is causing pain to Google.  Ahh, the sweet taste of schadenfreude, even if Trip sounds like the evil step-twin of Mary Meeker.