Thursday, March 20, 2008

Dumb and Dumber: Mahalo loving the Semantic Web and Social Graph

I haven’t written here in a while, but today’s topic will be one I’ve written about before: Mahalo. I want Mahalo to succeed, and I wouldn’t be heartbroken at all about Calacanis someday owning the NY Knicks. That the Dolans are so rich that they just don’t care how badly they run the team – it makes a good case for atheism. I’d rather believe there is a higher calling and a higher purpose.

I also believe that “what’s really going on” is important, and this week, Jason and Mahalo announced a bunch of stuff that was mostly, but not completely, a bunch of crap.

On his blog, Calacanis describes My Mahalo as “…something that builds semantic relationships between our users and our database (fancy way of saying you and the thing you own, have seen, want to see, want to read, etc.). These features take the social graph (fancy way of saying your friends) and integrates it into search results.”

This isn’t complete bullshit, but it’s a large dose of almost complete bullshit. The social graph thing doesn’t work like that, and can’t be automated like that without putting an onerous amount of work on both the person creating the information and the person absorbing the information. Things that require onerous amounts of work – they don’t do very well.

Here’s what I mean. I love a bunch of TV shows, and I can list them all for you, rank my favorites in order, etc. But this isn’t a particularly good or useful filter for my friends and family. For example, I know my brother Steve pretty well. I can say things to him like:

Get the House DVDs, you’ll like them
Get the Boston Legal DVDs, you’ll like them
Get The Wire DVDs, you’ll love them

I can also hedge with shows on the bubble and say something like, “I really LOVED Dexter, but I’m not sure you would love it or not.”

I can do all of this easily because I've known my brother all my life and know his tastes pretty well. Up to this point, it is fairly possible to automate it. But that’s only one view of the picture. Other shows I like a lot are: Stargate: SG1, Stargate Atlantis, Battlestar Galactica, LOST and Supernatural.

My brother would not like any of those shows.

So how useful is a generic list of shows I love to any one individual? It’s not very useful at all, and there's no easy way to make it more useful! It's certainly possible to make it more useful -- sure, I could do a lot of work to tag the information in such a way that “if you like this, you’ll like this, but if you don’t like this, you won’t like that” but that also requires the person using the information to do something similar. How likely is that? It's not likely at all really.

I have at least 15 years of experience with putting some of the profiling work on both the producer of the content and the consumer of the content and I’m here to tell you, at least based on the current state of technology or anything that will be doable in the next 10 years…it’s not happening. The work involved to make the information useful isn’t a ton of work, but it’s more work than is going to be done, especially by Mahalo's target market (basically your mom and dad and everyone else who never heard of TechCrunch, Twitter or the semantic web) and there’s just no getting around that.

I wonder how happy Mike Moritz and the gang at Sequoia is with Jason focusing and burning cash on this stuff. It generates a lot of buzz, but every dollar spent on buzz is a dollar not spent on expanding great search results for Mahalo sooner. The recession is coming too, and honestly, I believe that will wind up being good for Mahalo.

The core use case for Mahalo is not comprised of twittering, semantic web loving Facebook addicts. There is a use case for Mahalo that is huge, but that’s dependent on good search results for whatever you’re searching on. Spending money on semantic relationships and the social graph? That sure seems like a complete waste of money. Economic recession will be good for Mahalo – it will bring focus. Promotion is important. Buzz is important too. But a better search experience than Google provides for the core use case…way more important!

I know I’m on record as being a Mahalo Daily lover (and I am!), but this fits in very well with the target market for Mahalo and the promotional aspects are a bonus. Semantic web? Social graph? They’re 180 degrees from Mahalo’s target user and beyond buzz with the digerati make no sense at all for Mahalo’s target market.

1 comment:

Jason said...

Well, when you see the actually pages I think you'll see that it's not a bad use of time for us. couple of points:

a) this isn't a huge investment on our part. We've got a solid base of users and allowing them to share their opinions on a search result page is fairly easy, while providing a great experience.

b) you're correct that opinions are personalized per user. You might give your brother advice to watch The Wire and your friend advice to watch Battlestar. However, those individuals also know you and your tastes and can digest what you mean when you review each.

c) we're in the third inning of web search/research and navigation. we believe that combining humans and the social graph with machines is the best model.

d) we don't have to own this social information--we're going to import/sync much of it from other places like NetFlix and GoodReads and Facebook.

anyway, thanks for the feedback... it will give me much to consider over the weekend.

all the best,