Okay, I feel bad for being a prickly, churlish man. Yesterday, I left a snarky comment on Valleywag (yay, I auditioned and got accepted!!! Nerd auditions so cool!). The post was talking about Web 3.0 – gag gag gag – and Spock + Radar Networks. Here is what I said after hearing about Spock’s crashes on Day One of launch:
Who decided that Spock.com is a semantic web play? Any semantic web shop, even as loosely defined as possible by a layperson’s standards should have an RDF-OWL component to it. Radar’s does, indeed. So does that of Metaweb.
And what of Spock? What have they offered above a LinkedIn crawl with tagging and voting capabilities? Not jack, that’s what. Oh wait, they have offered us a site that falls over within the first few hours of launch. Sounds a lot like Web 0.03, dontcha think?
Mea culpa. Jaideep Singh informs us that they had budgeted for 100 page views per second and ended up receiving 400 page views a second! Whoa!!! One Billion page views a month!?? You gotta be f*ing with me. For what? LinkedIn + MySpace with tags? How much of this is hype and how much is a sustainable business?
After my last post (where I totally got busted for being a dirtbag cynic), I wanted to post my thoughts on the SFBeta event today.
PowerSet was the main event; companies like Kosmix and others did smaller demos (no PowerPoint, wobbly table, poor wireless; you get the idea).
Scott Prevost, PowerSet’s director of Product took the stage at 7:30pm after a rather lame speech by the event’s main organizer (no offence, but he definitely looked bombed). After this, Scott proceeded to an extremely slick demo site for PowerSet and showed us how PowerSet blows the rocks off Google for the following queries:
- Who did PeopleSoft acquire? PowerSet gets it right; Google blows it and shows links for the Oracle Corp.
- Acquisitions in 2001. This is a keyword query, not a question. PowerSet presented a nice list that included snippets with words such as “sold” and “bought” instead of just sentences with the word “acquisition”. Google did not do so hot.
- Who mocked Tony Blair? PowerSet returns results with the words “mocked”, “caricatured”, “lampooned”, etc. Google half-asses this one.
- PowerSet only showed results based on a Wikipedia index. We’re talking about 2 million pages here, so this is effectively small potatoes and nowhere near Internet scale. Getting to Internet scale isn’t trivial by any means, especially since doing heavyweight NLP processing at that scale has got to be a dog. But those guys are smart and probably have a handle on the issue.
- PowerSet did not take audience questions. This totally stinks, but it may well have been an issue of “event format”. Additionally, there were plenty of drunk people there, so this may have been a good call after all.
- PowerSet did not accept audience queries. To be fair, though, they *did* set up demo pods for a little product called “PowerQuotes”. I had to cut out and didn’t get a chance to play with it…sigh.
All in all, a good evening. Except for all the pain-in-the-ass drunk Giants fans on the Caltrain.
I’m super excited. After tons of hype, I finally get to see a live demo of PowerSet, the “secretive new search engine that claims to have cracked the natural language search problem”. The event is SFBeta; it’s being held at 111 Minna in the city.
The geek within wishes for a great demo and restoration of a “childlike sense of wonder” (thanks Fake Steve!). But the cynic within wishes for a fall-flat-on-their-face demo. Oh, the sense of schadenfreude!
I’ll update with my thoughts when I get back!
I’ve mentioned earlier how I get a fair amount (ahem, relatively speaking) of traffic from plain old web search. This allows for a little bit of insight into “The Database of Intentions”, to quote John Battelle. In other words: people type funny shit and I get to laugh.
I went to see Bill Maher at Flint Center the other day; see complete review of his performance here. During the show, Maher claimed that Republicans had a newly blossomed man-crush on Reagan and that “they want to put him on a stamp so they can lick his ass”.
Unless you are totally uptight, this is pretty funny. But check out this screen shot of what someone searched for to get to my review (“Reagan stamp ass Bill Maher”):
Is it just me or can that search term be interpreted several different ways? Good luck, PowerSet!!!
My good buddy (who will remain unnamed because he may not want to associate with this post) recently forwarded me an invite to Spock, the “secretive people search engine” whose executives created a nice PR shit storm for showing Victoria’s Secret models during a product demo. Naturally, anytime trashy people are involved, I want a ringside seat.
Here is what I said about Spock a month or so ago:
Spock.com, the “secretive” people search engine claims that 30% of searches on the web are people-related. This is the basis of their entire freaking business model. Sure, but what the hell does that 30% mean? Almost no one out of that 30% is searching for Saumil Mehta (I wish I had a stalker; I’ve wanted one for a long time but no one seems to oblige). 20% of those searches are for some combination of “Lindsay Lohan”, “Britney Spears” and “Sanjaya”. In other words, head terms, for which a Wikipedia page written by a freckled 17 year old will probably suffice. If it doesn’t suffice – well, that’s a sign that the reader has bigger problems than a poor flipping search experience.
Okay, so time to find out if I totally called it or if I totally blew it.
I played around with Spock for about 30 minutes (including time to sign up, get myself a damn password, etc.). Here are the points I want to make:
- As of now, Spock is primarily a profile aggregator across multiple sources such as LinkedIn, MySpace and Wikipedia.
- As of now, Spock’s ONLY differentiator over existing social networks is the ability to tag a profile and the community’s ability to vote upon those tags.
- For head terms such as “Barack Obama”, Spock may provide a set of tags or quick-hit answers for users unfamiliar with the person. Anyone with half a brain and a few minutes to spare would simply head to Wikipedia or any other site (maybe even Mahalo! Actually, no, that would be stupid) with more meat than what a bunch of tags can offer.
- For tail terms such as “Saumil Mehta”, Spock once again provides tags that serve as quick-hit answers for someone too lazy to look at my LinkedIn + MySpace.
- The social network crawl is still incomplete as of first week of July. My LinkedIn profile was not found when I signed up for the service.
- The visual treatments are very nice. The UI is clean, crisp and pleasant.
- Unless these guys have real tricks up their sleeve (or a strategy to drive massive user adoption a la LinkedIn), they are fucking doomed. As far as I can tell, the technology to simply crawl social networks and use a Wikipedia DB dump ain’t that impressive.
I’m not the biggest TechCrunch fan, but every once in a while they’ll put out an insightful post. When I read it, my jaw almost dropped: Technorati’s organic traffic has risen to nine million uniques, up from 3.5 just two months prior. This blockbuster number runs fully counter to the idea that Technorati was going to get a good old fashioned ass-kicking from Google Universal Search. For those of you that missed it, Universal Search throws away content type silos (blogs, news, images, videos) and blends results from different types. Check out the example query for “Darth Vader”.
Naturally, this smells of a massively successful SEO campaign piggybacking off Google search results. Since The GoogMonster supposedly hates “search results within search results”, Arrington claims that Technorati could be headed for the toilet.
Could this, then, be the reason for why Technorati put out a massive UI relaunch that puzzled the blog world??!! Technorati recently moved away from a classic SERP interface. Try a search on Technorati and you’ll see a page that tries to masquerade as content using tabs and rich media elements such as Youtube videos. Nothing particularly innovative here, since the UI simply mashes existing content types outside Technorati’s control.
But in the context of getting hitlisted by GOOG? Well, it makes a hell of a lot of sense!
Is it me or are people routinely making up numbers on search verticals and categories? Over the last few months, I’ve heard search vertical percentages thrown around like money by drunken sailors at a Vegas whorehouse. I’ve never seen a source, neither has anyone who seems to quote these:
- Spock.com, the “secretive” people search engine claims that 30% of searches on the web are people-related. This is the basis of their entire freaking business model. Sure, but what the hell does that 30% mean? Almost no one out of that 30% is searching for Saumil Mehta (I wish I had a stalker; I’ve wanted one for a long time but no one seems to oblige). 20% of those searches are for some combination of “Lindsay Lohan”, “Britney Spears” and “Sanjaya”. In other words, head terms, for which a Wikipedia page written by a freckled 17 year old will probably suffice. If it doesn’t suffice – well, that’s a sign that the reader has bigger problems than a poor flipping search experience.
- While I can’t personally afford to attend the swanky All Things D show, a colleague of mine made it down to San Diego to hear Jason Calacanis brag about how 24% of all searches are for the top 10000 terms. This is the basis behind Mahalo, a Hawaiian word that Bloat! so hilariously derides as translating to “Bandwagoneer”. While the product IMHO is a redundant version of Wikipedia + Google, the number sounds like it came from where the sun don’t shine (don’t quote me on it though)
- Lesser luminaries have quoted me other BS search percentages: 15-30% navigational queries, 20% local queries and 68% pornographic ones (okay, I made that last one up).
Does anyone know what the real numbers are? Do they even exist, or are we all just making shit up as we go along?