The more I dig into the new search engine startup Powerset the more I am wondering if it is nothing more than a house of cards. They’re an odd company with a bit of a split personality. For example, in some ways they are very secretive - everyone who gets to see “The Demo” (as it’s now being called) have to sign a nondisclosure agreement. It’s rare for startups to request this. On the other hand, though, CEO Barney Pell is hyping the hell out of the product (which is still just The Demo - Powerset may or may not ship in 2007 from what we’re hearing), the fact that they’ve raised $12.5 million in venture capital and that they’ve hired a dozen or so search experts out of Yahoo (where options are vesting and the search experts are itching for the next big thing).
All this hype is generating a lot of press. Normally even handed Matt Marshall continues to gush about them every time he writes (he got the chance to see The Demo). And the New York Times has written about them twice already this year (here and here).
This does make a good story. People are frustrated with the current state of search and want something better. Google was also criticized when it first launched as trying to tackle a problem too difficult to solve. People draw an analogy to Powerset, perhaps thinking that once a decade a new company will launch and kill off the old, entrenched competition.
Still, Powerset is given little chance by the search engine elite who judge such things.
People who’ve seen The Demo say it is a completely controlled environment. The index is limited to just a very small sample of high quality sites like the New York Times website, and the search queries are driven by Powerset employees as well. With that kind of setup, almost anyone could show a stunning demo. The real experts I’ve talked to who’ve seen the demo said it shows little or nothing real. Others leave impressed.
The company had a party in San Francisco this last weekend to celebrate their venture funding. Sarah Meyers, who regularly crashes parties with a cameraman to videotape what’s going on (including ours), was there to talk to Pell and other employees. The video is embedded below. Pell talks at a high level about the product, not really saying much more than we’ve already heard. At around the 2:47 mark, though, some poor soul gets caught on camera and answers Meyer’s question about what exactly Powerset is. His answer: “We have a demo where you can, like, search web pages and get results, like, like, books by children versus books for children…and that’s what we claim we can do…we’ll see in a year.” Not exactly confidence-inducing stuff.
The Yahoo AngleAll of this doesn’t add up. The company is too far away from launch to be hyping its product and throwing parties. But Pell, by all accounts, is a brilliant entrepreneur. So what exactly is he up to?
I think Don Dodge nailed it. Powerset is promising what Yahoo needs, and they’re building it with former Yahoo employees. Powerset may be hyping the product now because they want to sell, and fast. Dodge says:
What should Powerset do? There no easy and obvious answers…they only appear easy 10 years later. Powerset’s strength is in Natural Language Processing, or understanding the meaning and context of words. Lots of words like those found in a magazine, text book, or newspaper article. Rather than focusing all this NLP power on understanding the typical 2 or 3 word search query, why not help advertisers better target their ads on unstructured content?
What is Yahoo’s problem? Untargeted traffic. Yahoo has tons of traffic to its home page and Yahoo Mail, but no way to effectively target ads. AOL, MSN, and every other portal has this same problem. So they all end up selling low cost, low margin, CPM ads rather than high margin Pay Per Click (PPC) ads like Google. If Powerset technology could be used to “understand” the context and meaning in an email message, they could effectively target ads…and triple the portal’s revenues. If Powerset could scan a portal’s dynamic home page, or each users personalized home page, they could better target ads.
If Dodge is right, Pell is playing a very high stakes game of poker. If he wins, they get bought fast and all those Yahoo’ers go right back to Yahoo. If he loses, his product better be damned good once it launches. Because a lot of people will be watching to see if all this hype is justified. And if it isn’t, Powerset will have a very hard time getting any press at all.
Yep, the’re ripe for hype. I bet Powereset is using a very restricted demo because their NLP works from an old fashioned search algorithm. Restricting and controlling content in the demo can make viewers feel like they’re seeing something new. Innovative GUI use in newer search demos these days usually means developers don’t to resort to these tricks, but this sounds like old wine in a new bottle - for Yahoo. Salut
people have to stop at the concept.
If it is possible that an algorithm can understand what someone searches for, it is impossible to give relevant results from the internet. the internet is not semantic.
Unless you apply it to a specific set of websites where they know exactly what section represents what. so either they go for yahoo, or they can sell their search technology to webmasters.
As a matter of fact, i believe they have taken the problem upside down. Since the web is so chaotic, we should ask web users to categorize the pages they are looking at.
i’m not sold - definitely not sold.
the concept is not there.
it can search “books by children” instead of “books for children”(supposedly what other search engines are limited to) - ok
if you want books written by children - isn’t there a section of amazon that contains that category already? wouldn’t it be easier to just go to already well-known amazon for this category - instead of this powerjoke search engine? and the co is raising boat loads of cash to accomplish this?
i’m not convinced about the founders either, at least based on that clip of them - specially the second guy, he talks about the concept so insecurely - who would buy his pitch if even he has such doubts?
and that girl reporter, what a ditz
Three important lessons here:
1. Never, ever put an engineer in front of a camera.
2. Never, ever give an engineer access to a free bar.
3. Never combine 1 & 2.
Disclaimer: I am a VC who has invested $100,000 in PowerSet and am desperately worried.
Whoo - that last guy sounded drunk. I don’t think Andy’s point #1 is completely true - I’ve heard/dealt with plenty of engineers/developer who can come across much better than this chap did. But certainly, yes, don’t put drunk people on camera half-assing your product if you’re looking to raise more money.
Paraphrasing the last section:
“Google would ignore the words ‘by’ or ‘for’?”
“Yeah and we don’t.”
“How does that help anybody?”
“I dunno, I mean, like…”
Holy cow. yes, Andy, I guess I’d be worried too if I’d sunk 100k in to this!
Does’nt google already “scan the users email” and target ads based on that. How hard can it be for yahoo to do that today without powerset, does not sound like you need the almight NLP for that.
classic. just classic.
“how does that help anybody?”
“um..I dunno…*shrugs shoulders*…well, like…”
how can you be working at a startup and be THAT dispassionate about your product - being drunk isn’t an excuse, you can still be slurring and incoherently enthusiastic about whatever it is you’re making; that guy just seemed kind of indifferent.
but who gives a fvck right, I mean they have a year! (seems to be everyone’s excuse in the video…)
I find a little bit of irony in a “web 2.0″ video company ( Veoh ) exposing a heavily invested search engine start-up to be vaporware.
wow, Why would you let someone crash your party? ..
If you do, get some freakin PR person that will not drink to answer all questions.
I really don’t understand the VC world, I mean 12 mil for a market dominated by (3) players
Google (57+%)
Yahoo (35+%)
MSN (8%)
That to say, you would have to beat out one of these (3) to be in the top (3)… and the bottom (2) are a portal with built in users, so basically its
You Vs. Google … I wouldn’t want to fight a 800 pound gorilla.
-Rb
Good story Mike. Search is changing and organic search is becoming less useful. Blog search is actually becoming a better utility for me.
Search is shifting and there needs to be new products to help me save time, get better quality, and more relevance
I’m with John..
I now probably use Technorati + Google Blog Search + Del.icio.us more than Google.
However, for code search/help Google still rules.
Am I the only one who thinks that girl is the world’s worst interviewer and comes off as a total airhead? Seriously, I was embarrassed at my desk to watch the video. (Mike, what was her coverage like of your parties?)
Like everyone else said, that engineer at the 2:47 mark, um, did he graduate from high school? He can’t be a REAL engineer? (he was wearing a sport coat, so that instantly answers that)
To give him the benefit of the doubt: The drunk tired guy might do something like sysadmin and networking, not one of their core linguistic or search science people. Still smart but not as knowledgeable about the secret sauce. Still, he should have a better grasp on the company line.
Interesting. I went to the party at the high-end Frisson on Saturday night and it certainly did feel frothy and hypey — but then again so did TechCrunch 7 a little bit too, so I’m not sure that really means all that much. I’m sure it cost a mint to have the party at Frisson though. Very high end.
I responded with a similar thought on Matt Marshall’s article on the company last week when Matt talked about possible interest by Google. I’d think far more interest would come from someone looking to catch up to Google. Yahoo, Microsoft, IAC would all pop to mind.
Still, the key is *if* the technology works and Mike, like you I have no idea because I’ve never seen it or tried it.
But remember back to another well funded venture backed company, Riya. I was super excited about Riya when they launched. Facial recognition could have been such a great thing, changing the world for image search and all that. Remember Google was about to buy them and all too… And then it turned out that the software didn’t really work (and I know I actually tried that one myself uploading thousands of photos to the site), and now they are going to be some company that sells shoes or something.
The powerset party was kind of over the top. It was weird. Everyone was name dropping like hell and everyone loved Barney. You couldn’t have a conversation without someone mentioning Barney’s name. I don’t know Barney myself and I didn’t really know anyone else at the party except for a few journalists, oh, and Julie from Photobucket.
It will be interesting to see where this one goes. For them and for the rest of us, I hope they do make search better for everyone.
Dan Dodge writes:
Rather than focusing all this NLP power on understanding the typical 2 or 3 word search query
Which is exactly why NLP search won’t be happening for years (decades?) to come. How many people out there use more than 4 search terms on google?
Powerset has stolen every page from Foldera’s book.
-Zaid
Good story
BlogReader: Most searches are for 4 words or less, true, but if we see some advances in speech recognition (it’s easy to rag on Vista, but they’re trying) then NLP search could become MAJOR since the search tool would need to establish meaning from queries as bad as.. “err, gimme that site about the uh naked chicks riding on a bus and getting it on with guys and stuff”
It’s parties like this that further exemplify the inspiration behind http://bub.blicio.us
Peter,
Given the incredible challenge of NLP, is it more likely we’ll see it as part of a solution with voice recognition…or that there’s going to be a constrained vocabulary and grammar set? Just like you can’t sit down at a computer and type “Computer, show me all the files I wrote recently where I talked about my dissertation topic.”
NLP’s been such a sinkhole (I know legal publishers who were badly burned by its promise in the 1980s) that I think people will find better ways to invest their cash in the long run.
That was one of your best writeups - its good to get some indepth analysis from time to time. We need more like this.
this is sad. i am pretty skeptical about this type of thing. the demeanor of those associated with this company only further that feeling. i would think these guys really have to come up with a homerun type of product or they will sink fast.
Oh man, that last guy just scares me!
“That’s what we CLAIM we can do”
“Is that a false claim?”
“Well, we’ll see in a year right?”
On a more serious note…
How is it that they can raise $12.5 million with a simple query builder?
It amazes me how such possibilities become reality.
I don’t think NLP will ever defeat the power of simple, well chosen keywords. Think about it, when you type “I want a red bike” into the search, how is it going to highlight “I want a red bike” in the search results? It won’t. It will highlight “Red Bicycles.” If THAT happens, I’ll interpret it as a failed query. I want to find *exactly* what I type in, not a modification of it. Perhaps if it made a list of *suggested* queries, I’d be more inclined to use it. We’ll see.
I hope it’s a good demo, best of luck.
holy cow. That video is painful to watch. The interviewing girl is more there for the party than she is the product, but the wost is the young gentleman she interviews. If I had a guy on my development team that said those things, we would have some serious issues. Saying that they “claim” to do something, and not back it up by saying they can do it? Wow.
Mike, great analysis…probably spot on. Regarding the usefulness of organic search, I’m of the same mind as Jay: blog and bookmark search is becoming far better, and google intentionally excludes these from search results so it’s less and less useful.
Try a meta-search like http://www.srchr.com …I think that’s more the future of search than natural language…I mean wasn’t NLS what AskJeeves.com was built for?
Here’s something that didn’t make sense to me, if Powerset’s technology is so sweet, why do they need to license tech from PARC?
“People are frustrated with the current state of search”
i don’t believe this is true at all
but otherwise i agree with your analysis of powerset, and yes, they like, all the other web2/search2/whatever2 sprintshops have one and only one exit available - to be acquired.
but the payday proposition is getting too easy. everyone involved has to start asking why they think google/yahoo/microsoft is going to lavish millions on twelve months of brainstorming and alpha code.
“we’re gonna party like its 1999″
my god the interviews in that video are shocking…do those people work for powerset…
“yeah, you know, you can search for something, and like, um, we’ll give you the results”
can’t wait … ack
This is a good perspective. It never surprises me (after living through the last tech boom/bust/whatever) to see a company overhyping itself or media falling for it
I agree though, I don’t hear a lot of people complain about search.
If this thing ever comes out of the oven and does what it’s supposed to do (doubtful) and in fact is something that Yahoo needs, then Yahoo should just acquire this company. I guess that would mean that they would be hiring back all of those ex-Yahoo’s.
Well, the video was edited with fades so the full context of the conversation seems biased and isn’t clear.
I fully support Powerset and the “dude in the hallway” after watching the party interview and two PodTech Scoble Show geek dinner videos explaining their fledgling company:
http://www.podtech.net/scobles.....with-geeks
http://www.podtech.net/scobles.....with-geeks
Powerset’s CEO is smart and it’s refreshing to see unscripted commentary (good or bad) to get a better feel for who these people are.
These are not unskilled Web 2.0 idea folks, they are industry veterans and experts in computational linguistics trying to solve a difficult NLP problem. Licensing the PARC code was a brilliant move, saving years of development work.
Use Flickr to search for Powerset photos. You will find pics of a strong team trying to build a real company. They are providing jobs and opportunity, give ‘em a chance.
Sounds like fascinating technology to me.
Peter W.
wow, all this hype. i met pell at a party a couple months ago and through the foggy memory of the evening, all i seem to remember is that he was a pompus arse… so the hype surprises me not…
“If it is possible that an algorithm can understand what someone searches”
Yes it is. It’s called english query in some databases. Check out http://databases.about.com/od/.....hquery.htm . Not only does english query understand, but it is also adaptive to future queries. I think Jeeves is powered by this type of system.
As for Powerset, ignoring the “for” or “by” is a non issue. Those are noise words in some databases and it’s easy as editing the noise words list if you want those included in your query.
There are other issues than just “understanding” the query. There’s also the algo to rank the results. Currently Google relies on link analysis among other things and has been really effective. What does Powerset have? There’s also the issue of building a great crawler to index all the data out there. A crawler which takes years to perfect given all the different types of data out there.
Man, that video is awesome. I thought the first guy was bad, but the last guy (the drunken engineer or HR guy or whatever) was so incredibly, painfully awful that I almost felt bad watching it. I think Andy C. hit it right on the nose.
Having recently seen my friend lose over 100k in a tech startup (not Andy) that *promised* to have some great technology, i will say this:
Don’t invest in any company, not matter how brilliant you think the founders are until you can see something that works.
Disclaimer: I am a web entrepreneur saying this.
I like Barney and believe in his ability to execute.
Has anyone actually done the children/books tests on Google?
http://www.google.com/search?q=books+by+children
http://www.google.com/search?q.....t+children
http://www.google.com/search?q=books+for+children
Results are far from optimal, *should* be fixable.
This is all very amusing. I will lay down my home to bet Powerset will not deliver! Andy C…geez man, I’m sorry. What can I say?
Very bad feeling though.
After watching the interview:
Powerset seem to be the following terms:
Powerscrew = those investor may be screwed
Powersuck = may be hyped and suck like any search engine.
Powershit = May be another shit thing that claim to work but never work.
Powerscandal = 12.5 millions scandal that probably spend up in partying then year later cancel because money ain’t enough. Well, it never enough if the technology never work.
Powersex = If it doesn’t work, well, let it specialize in searching porn. Afterall sex sells.
Probably many genius can think up of ways that Powerset may fail, and those who invest probably setup for failure. Well, the smart one will take the money at height of hype and just disappear to enjoy deal of lifetime
With power come responsibility. So what can Powerset do ? It be responsible for one the biggest boo-boo in internet history along with Pixelon.
Why am I so sure. First of all, hype is bad and take money for a vapourware is one of the worst thing a investor can do. If a company like Microsoft or Apple, then I will believe since they have track record, but a company like powerset without credential is very much hyped.
A year later, i’waiting for the company to powerFuc#$@ those investors who spend money unwisely.
Imagine Google with one of the smartest engineers in the world could not do it, why do you think a few smart aleck can do it in powerset ? So ppl stop dreaming and wakeup.
there are plenty of people that do work at Powerset that DO know what they are doing and that DO know what their product does.
I am embarrassed for the interviewer and interviewee.
and he is not an HR or network guy….unfortunately…
That video makes it sound like “PowerNap”
Boy those grapes are sour, all because someone other than TechCrunch got the exclusive?
Re: “I’m sure it cost a mint to have the party at Frisson though. Very high end.”
Frisson is owned by Peter Thiel of Clarium Capital / Founders Fund, one of the investors in Powerset. So perhaps he comped them or gave them a discount.
The party was not paid for by Powerset, so rest easy.
Mike, go check out http://www.Nervana.com - no hype, ltd PR push, good set of paying customers and definitive iterative progress in next generation search. Re: Pset, next-gen search technology is non-trivial - taking lab innovation from research to consistently usable product is really really hard. Frankly, licensing technology may or may not give one a high-octane boost…sure, it may give you great IP protection and a nice IP portfolio (for offensive OR defensive purposes) but an IP portfolio does not a product make…and these days, there’s one more wrinkle - any technology that gets rolled out to the general public (like vertical or horizontal search) MUST scale…so one can criticize GOOG all you want but their search technology scales (albeit regressing to the mean which probably works just fine for 80%+ of the search consumer pop). These are early days. Fun times ahead….
I think everyone here is overreacting. You had an airhead interview a bunch of drunk people, and through the magic of the internets everyone gets to overanalyze and make crazy conclusions. The engineer the end is probably a smart guy with a sense of humor trying to level the playing field with the wig-wearing dummy. It seems to me that he was not fully aware that this was a real interview, and was actually just chatting casually with the girl, as if it is any old bar conversation.
I’m much more concerned about the CEO’s words along the lines of “we raise some money and then we raise some more.. that’s how we do it!”. He should have a lot more training in public speaking and is much more aware of the importance of his words than any of the lower level employees whose only agenda there was to party.
I guess it is the bubble.. 20 or 30..
Get a grip, guys. The “unlucky chap” had clearly had a couple of drinks, and I would not draw any conclusions about him or the company from a 2 minute of video of a party with free booze. People have different personalities, and some like to play down what they are doing whereas others like to hype it up: it doesn’t mean that they don’t believe in it or that it is smoke and mirrors.
Drunken do tell the truth most of the time. That will give a good picture of powerset. The name powerset doesn’t sound Bombastic or breathetaking. But it will be once it get bust as the largest con project since bubble bust. Check if the investor is actually the founder of pixelon as well.