The way Delver sees it, there are two basic problems with search the way it is today. The first problem is that you could be a dog or the President of the United States, and you’d still get the same results from a search engine if you both put in the same keyword. Or, to put it a different way, a sixty-five-year-old history professor searching for “Goth bands” probably won’t be looking for the same kinds of results as his fifteen-year-old granddaughter, but they’ll both get the same results anyway. Delver thinks this approach is a mistake.
The second problem is that, not only do most search engines not know who you are; they don’t know who your friends are either. Since they can’t see your “social graph,” as it’s coming to be called, they can’t tell whether your friends might know something that’s relevant to your search query. Delver thinks that, if you’re searching for “New York,” you might want to see blog posts and YouTube videos and such from friends that are about or mention New York.
If Delver can make the technology work, TechCrunch notes, it could add real value to online social networks, “because, let’s face it, social networking doesn’t offer much functional value beyond allowing people to connect with one another.” Unfortunately, Delver won’t be entering beta until March, and it will be a private beta at that. But we can take a look at what the would-be search engine has revealed about itself so far.
Delver is short on the details of its background. Headquartered in Herzliya, Israel, the 20-strong company plans to open U.S. offices in Silicon Valley a little later this year. It has backing from Carmel Ventures, a VC company also based in Israel. Delver CEO Liad Agmon has previously served as the Director of Product Management for McAfee, and was co-founder and CTO of Onigma, a strong competitor in the field of host-based data loss prevention solutions. A regular jack-of-all-trades, Agmon also worked in Israel’s high tech and movie industries. Incidentally, he’s very well-connected on LinkedIn; one can imagine that he uses Delver’s technology every day.
Here’s a screen shot of Delver’s current home page, cropped and reduced a bit to fit:
Delver’s technology has been in the works since 2005. TechCrunch describes it as combining search technologies, semantics and Natural Language Processing (NLP). Delver says that it organizes a user’s friends’ tags and information found on social networking profiles, blogs, bookmarks, photos and video sharing sites. When a user searches for information, rather than prioritizing Web results based upon the popularity of particular pages or web sites, Delver ranks information it has organized based on the user’s social connections. Delver points out that every person’s social graph is as unique as a fingerprint – thus, no two people will get exactly the same results.
“People want trusted information from their friends, but may not know who in their network is knowledgeable about a given topic,” Agmon explained. “We make Web search more fun and meaningful by prioritizing results based on a user’s network, while enabling the user to discover others in their extended network who share common interests.”
It’s very simple to sign up for Delver; registered users will (one assumes) get more out of it. TechCrunch indicated that users would be able to authenticate sources they wanted to be associated with by providing usernames and passwords – for example, for their Flickr account, their YouTube account, their LinkedIn account, and so forth. Once Delver knows who you are, it’s easy enough to find out who your friends are; for many social networks, this is effectively public information.
Delver takes an interesting approach to making information in one’s online social network more accessible and relevant. I’m looking forward to seeing how that translates into action. But it seems unlikely to work as a primary search engine. What if your friends don’t have the kind of knowledge you’re looking for?
Tac Leung made some insightful observations in a comment on TechCrunch’s coverage of Delver. Leung noted that Eurekster and Friendster partnered about four years ago on web search. “We created a unique search engine for each user, and the results were influenced by the activity of your 1st degree friends, and to a lesser degree by your 2nd and 3rd degree network.” They found that friends rarely search for exactly the same terms, but that broadening the network didn’t necessarily make the search results more relevant.
They also found that relevance was easier to achieve by focusing unique search engines on groups within Friendster that had similar interests rather than a person’s specific social network. There was no need to log in to make this work; one only needed to search on the engine geared to a particular community’s interest.
Unless Delver’s approach is particularly compelling, it risks getting lost in the shuffle. There are other web sites already trying to tackle at least one of the two problems that Delver is trying to conquer. For example, Google and all of the other major search engines offer some form of personalized search, often based on search history but sometimes taking other factors into consideration. This addresses, at least in part, the issue of search engines not knowing who you are. And Searchles, among others, gives you the option of limiting your search to links posted by your friends, or by your friends’ friends.
As one last possible issue worth noting, it seems as if new Web 2.0-based companies are running out of original ideas in the area of logos. I can’t claim the credit for spotting this originally, but Delver’s logo looks a lot like Cozi.com’s logo.
Cozi’s concept is a bit different; they’re building something more along the lines of an “Internet fridge” to keep everyone connected. Still, the logo resemblance is a little discomfiting, and could lead to confusion.
It’s practically impossible to say anything meaningful about a company’s chances for success when you haven’t seen their product in action. Still, I’d like to point out that Delver does have a number of things going for it. First, their experienced CEO has proven he can create a company from scratch and make it valuable. As CTO of Onigma, he helped build the security firm to a position where it was purchased by McAfee less than two years after its creation for $20 million.
Second, as I mentioned earlier, there’s no one out there doing exactly what Delver is doing. You can search within your own online social networks on some sites, but there’s no easy way to check through all the content you and your friends have posted to several networks simultaneously. You also may miss friends who are on several networks.
Let me give you an example. I belong to Searchles, LinkedIn, LiveJournal, and other sites. Say I’m a costumer and have a query about building a Klingon costume. Let’s further assume that one of my LinkedIn contacts also has a Flickr account that I don’t know about, on which he’s uploaded several very detailed pictures of Klingon costumes. I wouldn’t even think to search my LinkedIn network for this kind of information – but if my LinkedIn contact and I have both claimed all of our accounts, if I understand Delver’s technology correctly, it would turn up those pictures for me. And my contact and I will discover we have something else in common.
Even so, I wouldn’t want to use Delver as my primary search engine. As with other technologies I’ve reviewed, and some that I now use regularly, I could see using it as a supplement to a search engine such as Google. I could even see one of the major search engines buying Delver somewhere down the line. I don’t think that it will be the be-all and end-all of search, but I sent them my information for an invitation to their private beta, and I’m looking forward to seeing the technology in action.