I’ve been thinking a lot about what social search is, what it means, does it matter if we all use it to refer to different things? (I highlighted a few of the different ways it’s being used in commercial applications towards the end of this talk I gave at BarCamp in San Diego this weekend.)
In that talk, I reiterated my basic instinct that social search will be most powerful when it exploits a user’s personal social network. I care much less what the whole web feels, prefers, or is doing, than what the friends in my social network care about, prefer, and have done. (For this reason, OneRiot fails to excite me as a “social search” engine.) I mentioned that FriendFeed has an interesting model of social search, where search queries first return results from content your friends have shared into the system. Luckily Ofer Egozi commented on my blog post to remind me that Delver does this too.
I wanted to share a specific example from a search I subsequently performed on Delver that reminded me of its power as a social search engine:
I am (very lucky to be) going to Kauai over the Christmas/New Year break this year. I will stay with my parents in a rental house in Poipu for the most part, but my boyfriend and I thought it’d be fun to get away to the North part of the Island (a whole 1.5 hours away) for a few nights—only, I’ve never stayed up there (Hanalei, Princeville, etc.) so I don’t have any conception of what’s available, what’s good, or what we might like.
The short story is that I spent half-an-hour doing general searches on Google this morning only to come away with a set of B&B websites that were equally ugly (in design) and impossible to decide upon. Then I searched for “hotel kauai” in Delver; here were the results:
This obviously isn’t the most complete set of results (Ofer points out that one problem with social search is its lack of recall/coverage), but the very first result seriously caught my eye! The title of the link (“our hotel in kauai pretty much rules”) was interesting since I was struggling to find awesome-sounding places earlier in the day. Plus, the author of this link is a woman named Dana Robinson, who I don’t know, but who is apparently friends of Tantek Celik’s, who is a friend of mine. I now automatically view this link with some credibility and authority, even though I don’t know the author. Instead I’m trusting Tantek to vouch for her.
Before I searched on Delver, I had already been thinking of friends I knew who spent time on Kauai or who might know of people who had—specifically so that I could get an expert’s and also a friend’s opinion on where to stay. (And in this people-search task, I failed.) So, I am already feeling better about my quest for lodging in Kauai thanks to Dana Robinson, via Tantek, via Delver. Is this the future of search?
2 Comments
Brynn, I couldn’t have phrased it better myself… your experience is exactly what we’re trying to create, of trusted and relevant results from your network. It’s a great promise, and surely a major factor in the future of search, but also an immense technological effort – you need a very large coverage of the social graph and the content attached to it, and the graph-based ranking function is practically different for every user (rather than, say, a single consolidated PageRank). So we still have a lot of work to do before we can deliver this experience in all of your searches, but we’re heading there.
I’d actually be giving a talk on exactly these issues next month, unfortunately not in your area
Thanks for the review!
Oh, is that going to be in Israel? What a shame! (I’d have loved to attend…)
It’s been enlightening to share some experiences with you these past few days. I have never tried to implement a social search engine—I don’t know what the technological challenges are, but they will be important for me to consider in my work. Thank you!
One Trackback
[...] 22, 2008 · No Comments As a followup on Brynn’s review of Delver, I’ve had an interesting exchange with Lachlan Hardy, where Lachlan expressed his [...]