I know, I know. They’ve been around for less than a year, so how could they possibly be worth a billion dollars? It’s basically just a Twitter clone that’s a little bit faster and doesn’t go down as often, right? Despite the surface similarities, FriendFeed has the potential to become one of the more valuable services on the Internet, and here’s why. Google’s primary goal is to index the world’s information. FriendFeed’s primary goal is to index the world’s conversations. We all know the value that Google created out of their index. Here’s a quick overview of how FriendFeed can do the same.
Phase 1: Grow a Rabid User Base
In order for FriendFeed to build up a index of the world’s conversations, they first needed to create a front end with two important components. First, it needed to be as easy as possible for people to pull in their conversations from all over the web, and to do it in the most automated format possible. Second, they needed to build in a viral distribution model that would allow them to build up their user base as quickly as possible. My guess is that somewhere in the FriendFeed offices, they have a graph on the wall that shows two curves. One is an exponential growth curve that represents their predicted growth (based on power laws and estimated viral coefficients) with a clearly defined upper threshold. The other is a line showing the actual user activity. The one thing FriendFeed should focus on in this initial phase is to ensure that their actual growth is showing a viral growth pattern and will reach their target threshold by a certain date. Once they reach this threshold, they will have proven out their viral distribution model. Take an exponential growth pattern and multiply that by an exponential content aggregation engine, and FF has the underpinnings of a massively powerful engine that can quickly build out the data needed to form the basis of their “conversation index”.
It will be pretty obvious when they hit their target threshold. It will be approximately 2-3 weeks before they announce a pretty sizeable VC investment. FF won’t be the only ones who are watching this growth curve very closely. :)
Phase 2: Mine for Context
At this point, FF can use the new investment to build out a pretty sizeable server farm to handle the incredible amounts of data that will start flooding in. You’ll probably also see some job postings on their site for PhD’s with backgrounds in data mining and contextual relevance. Why? Because unlike a traditional search engine that can do a pretty good job of indexing information just based on things like page rank and keywords, extracting value from conversations needs to take context into account before it can truly be valuable.
For example, compare the view of Twitter posts on the Twitter public feed to the view of Twitter posts on Twistori. This is a very crude example of type of analysis needed to pull a signal out of the noise. Once the algorithms have been created to extract the context, FF just has to sit back and wait for their index to grow to a point where they are able to provide statistically significant results for search queries. Which, of course, brings us to the start of phase 3.
Phase 3: Search v2
And here’s the real value of FF gets unlocked – search. Scoble actually touched on this a few weeks back (see #3), but here’s my take on it. Imagine you’re looking to buy a new laptop. You go to FF and ask the question “What laptop should I buy?”. The results will come back in two forms. First, a list of posts from your friends displaying their opinions on their favorite laptops. Note that unlike the search that is on FF now, these contextual results will filter out any noise such as conversations about problems they are having with their laptops, posts that have the word “laptop” but are not specifically about laptops, etc.. Second, a list of posts from all users displaying information about their favorite laptops. Again, this will be a filtered list, and it could also be aggregated in a format like “120 people like the MacBook Air, 75 people like Dell Latitude D820, etc..”. Of course, you could click into any of these lists to see the details of the conversational threads. And finally, if you didn’t get a satisfactory answer from your search, you could simply post your search query as a message post to your followers, and get a special notification back once people start responding. This goes one step beyond anything a traditional search engine can do, and really takes a best-of-both-worlds (mechanical vs personal) approach to finding the information you’re looking for.
In addition, there’s one more killer application of this new way to search. Imagine you’re in downtown Seattle and you’re looking for a place to eat. You pull out your phone, do a search for “best restaurants in Seattle” (or click on the “Restaurant” icon on the FF mobile app), and get the same FF search outlined above for both your friends and everyone – filtered to show only opinions from people from within 1 mile of your current location.
I’m not in charge of the budgets at G, M, or Y – but if I was, I don’t think it’s too crazy to think that a service like that tied into my advertising platform might be worth a pretty nice chunk of change.
What do you guys think? Am I drinking way too much of the Web 2.0 Kool-Aid, or is FF really on to something here? I’d love to hear your thoughts in the comments below.
Update #2: TechCrunch is reporting on FF’s hockey-stick growth. Now it’s time to keep an eye on their “We’re hiring!” pages…