The most worthwhile class I ever took was my Economics of Technology course back in the fall of 2003.  The whole course involved reading papers about technology strategy and discussing them with smart, engaging people– it was everything a course is supposed to be.

One of the major themes of the course was technology standards– why certain things (e.g., QWERTY keyboards, x86 microprocessors, Microsoft Windows) become standards and other things don’t. In light of Google’s announcement of the Open Social API for connecting social networks with social application developers, and the widely held belief that this heralds a standards war with Facebook for the hearts and minds of social app developers and (by proxy) social network users.

Seriously.  Check out Techmeme.  People are freaking out.  As for me, I feel like I’m looking over at the dudes from uncov, waiting for them to say ‘FAIL‘.  (I love that picture.  Best fail ever.)  But no fail is forthcoming, so I’m going to step in and take a crack at it.

First of all, let me say that open standards (as Open Social purports to be) are generally a good thing.  The idea of Open Social is pretty nice– you write your application once against this common API, and then it works with any one of these different social networks.  The Open Social API abstracts away the underlying social network, just like HTML and the browser abstract away the operating system, or the Windows API abstracts away the underlying hardware.

The more technically-minded amongst you will have already detected the bullshit lurking in the previous paragraph, since the abstractions I just talked about are, shall we say, leaky.  They’re not perfect by any stretch,  so developers devote alot of time to handling the different special cases and quirks that arise from slightly different implementations of the same standard interfaces.

Josh’s Corollary To the Law of Leaky Abstractions: The number of leaks in an abstraction is proportional to the implementer’s incentives to go beyond the standard.  When the incentives are high enough, the abstraction gets forked.

Simple example of a relatively leak-free abstraction: you sell disk drives for PCs.  Microsoft specifies a set of functions your drive needs to perform in order for it to be usable by Windows.  You have no incentive to develop any functionality for your disk drives that isn’t completely covered by that standard, since Windows (and by extension, everyone else) won’t be able to use it.  It’s a pretty awesome situation for Microsoft, and a pretty crappy situation for you, the disk drive manufacturer.

The opposite case (and a textbook example of the oft-cited Josh’s Corollary to the Law of Leaky Abstractions) is the forking of Unix.  Once upon a time, there was this operating system called Unix that was developed at AT&T, and it was good.  But then AT&T did something bad to someone or something, and it had to sign an agreement not to sell Unix, and to license the Unix source code to all sorts of acronyms– SGI, IBM, DEC, SCO.  Each of these licensees took Unix and made it their own– creating their own version that implemented the original Unix plus some other stuff, which had the net effect of making all of the Unices incompatible with each other.  You see, they were all competing against each other for the same customers, and so they had very strong incentives to take the “open” Unix standard and make it very much their own.  And while they were fighting with each other and driving Unix application developers nuts (arrgh!), Microsoft swept in with their fixed API that ran on cheap hardware and took all of the developers and (by proxy) all of the users.

I don’t think it’s difficult to tell which of these two scenarios more closely resembles the dynamics in Open Social API vs. Facebook.  In my humble and undeniably correct opinion, the amount of time it will take for one of the original implementers to fork the Open Social API can be clocked with an egg timer.

This has been a longish post, so I’m going to wrap it up now.  In Part Two, I’m going to talk about how Open Social is more about Google trying to get companies to sign up to implement the API than it is about getting developers to create applications for it– in my mind, Google is even more interested in consuming this data than they are in providing it.

Labels: social networks, facebook, withdrawl

[ 0 comments ]

After all, what’s Facebook fatigue without Facebook withdrawl?

I’m taking the week off from Facebook, primarily to prove that I can still live without it. Those of you who aren’t on Facebook may not understand why this is difficult to do, and I can empathize with that– I held out on the whole social networking thing for a really long time, and back before I plugged in, I found the whole phenomenon pretty ridiculous. Ever since I realized I was addicted, I’ve been trying to figure out the why and the how, so that I might find a cure.

I was up in DC for Keith and Natalia’s wedding a couple weeks ago, and on the plane ride up I read Daniel Goleman’s Social Intelligence. Goleman has an odd writing style; the vast majority of his books involve him citing the research of people he went to grad school with at Harvard who are all really, really brilliant. I feel like he tends to short-change all of the mediocre psychologists and neuroscientists out there who are doing difficult, complicated research that isn’t remotely interesting to anyone.

All kidding aside, I really like Goleman’s stuff– he provides the neuroscience behind my gut feelings, which is immensely satisfying to my left brain, which loves to feel like it knows what’s going on in the world. There was one section in the book where Goleman talks about how our brains are wired for social interaction, so much so that our thoughts tend to turn to our social relationships when we don’t have anything else going on.

David Brooks wrote an article the other day on our outsourced brains that made me think about the ways in which I’ve come to depend on a handful of web services to act as an extension of my brain. Google is an obvious (and nearly universal) example of this– the module that our brains turn to whenever a question pops up from the ether (though I wonder when everyone else is going to notice that Google is rapidly turning into a front-end for the Wikipedia).

As Google is the 50 year old bordeaux for the prefrontal cortex, Facebook is the crack pipe for the orbitofrontal cortex. Our brains literally cannot get enough of what our friends are up to, and Facebook provides all of the raw materials via the news feed. Just as Google leverages user feedback to improve spam filtering, I could see a future where Facebook could predict which of your friends were about to get together or about to break up, and tell you how your actions might influence the outcome. Wouldn’t that just be insanely cool creepy?

So that’s 24 hours without Facebook. Let’s hope that I spend the next 24 thinking about it a little bit less.

Labels: cognition, social networks

[ 0 comments ]

Long-time readers (all three of you) know that I have a great deal of respect for Fred Stutzman’s Unit Structures blog when it comes to all things social networking.  His roundup on the Microsoft-Facebook deal is even better than the one I would write if I had the time.

With respect to the 15 billion dollar valuation, I know alot of people feel it’s ridiculous, but as I’ve discussed before, if you look at the potential value of the data Facebook has, relatively small changes in their estimated probability of success add up to many billions of dollars.

The problem right now is that the confidence interval around that probability estimate is pretty wide, and Facebook’s challenge in the next year is to start tightening it up.  I think it’s safe to say that Facebook’s monetization strategies have been quite conservative thus far– their primary focus has been on building up the user base as much as possible and keeping those users happy.  The money they received from Microsoft and a couple of hedge funds gives them opportunity to build out the world’s best data mining operation, and even more importantly, it gives them the time and the freedom to experiment with different ways of using that information to create value.

Sergey was a little snippy about Facebook the other day, indicating that he felt like Google was better off because of the discipline that was required when they were building a company during the bust.  But it sounded to me like there was a tinge of jealousy in that statement (not to mention a general annoyance with Facebook for stealing many of Google’s engineers)– after all, I’m pretty sure Sergey would have taken 750 million dollars for 5% of Google back in 2001.

Labels: tech, social networks

[ 0 comments ]

Dave Winer is one of my pet peeves; his only redeeming quality is that he may be the most aptly named human being on the planet.  He’s whining about the use of the term ’social graph’ as opposed to ’social network’, and claims that the people who use the term ‘graph’ to describe our web of social connections are idiots.

There really is nothing more annoying than being called an idiot by one of your pet peeves, so forgive me if the rest of this blog post comes off as a bit harsh.

Here are a few things you might not want to do if your primary goal is to not sound like an idiot:

• Spell ‘theorems’ as ‘theorums’.
• Claim that all graphs are four colourable. The non-idiotic version of this statement is ‘all planar graphs are four colourable‘. (Don’t like my British spelling? Bite me. My graph theory textbook was written by a dude who spoke the King’s English.)
• Say that combinatorics is related to statistics. This statement is arguably more misleading than the ’social graph’ terminology Dave is railing against– combinatorics is related to the statistics of coin flips, but not to things like bell curves or regression or anything else that a normal person would associate with ’statistics’.

Looking over this list, I wonder if Dave’s idiocy wasn’t a strategic attempt to prove his point– if a nominal “math major” like Dave Winer doesn’t know anything about graph theory, how can we expect a mere mortal to understand it? This sort of thing drives me crazy– it plays off of math phobia to trick people into thinking this stuff is really hard and that they’re not capable of understanding it.

Bad form, Dave. You’re on my list. Well, you were on my list before. But now you’re like, even higher on the list.

Labels: math, social networks, rants

[ 1 comments ]

I want to expand a bit on my ‘what facebook is up to‘ post from a couple weeks back.

Plaxo announced that the launch of Pulse this week, which Michael Arrington dubbed the ‘Open Facebook‘.  One of the relatively few criticisms of Facebook is the blogosphere is that it is a ‘walled garden‘ or ‘roach motel‘ for data– information goes into Facebook, but you can’t get it out again.  The conventional wisdom is that this violates the spirit of the Internet, which was build on open standards so that anyone could share information with anyone else, and that in the end, openness always wins out.

Personally, I don’t think Pulse is going to overtake Facebook anytime soon.  Plaxo started life as a way to keep your address book up to date, and earned a reputation for spamminess as it had a tendency to email you every time one of your contacts changed something on their profiles.  Nonetheless, it raises a very interesting question: is it possible for any open social network to overtake Facebook?

Users and Data Streams

Quoting Wikipedia, Metcalfe’s Law states that “the value of a telecommunications network is proportional to the square of the number of users of the system.”  This same logic has been applied to other domains where network effects are important, like operating systems and (lately) social networking applications.  As you’ll notice from the Wikipedia article, a number of people have made various objections to Metcalfe’s law, saying that it over or underestimates the additional value a network receives from each additional user, though everyone seems to agree that the effect is at least superlinear.  My quibble with the way Metcalfe’s law is applied is that the unit of analysis isn’t right, at least when it comes to platforms like operating systems or social networks: the value doesn’t increase with the number of users, it increases with the number of data streams.

For the telephone network, the number of users and the number of data streams is essentially the same– each person has one voice.  But in the context of software, a single user can correspond to multiple data streams.  On Facebook, I control data streams of messages, photos, links, and games of scrabulous.  The day that Facebook opened up their platform to third party applications, the value of their network increased significantly– and would have increased if they hadn’t ever added a single additional user.  Naturally, the fact that the network increased in value from the multiplicative impact of the additional data streams brought more users (and hence more streams) into the network, which increased the value of the network even more, which brought more app developers, which brought more users, etc.

Open networks do not win out over closed networks by default– the fact that Windows is still very much the dominant operating system indicates that the value of all of those third party applications that have been developed for Windows over the years has not been overcome by the more open Linux platform.  Rather, open platforms often win out because it is generally easier to add data streams to an open platform than it is to a closed one.  For example, the original AOL was a true walled garden: third party developers could not create applications that would run inside of AOL nearly as quickly, cheaply, or easily as they could create new web sites using HTML and free tools like apache and perl.

Social Networks, Identity, and one hundred billion dollars

One of the things that people have been pointing out as a favorable aspect of Pulse is its support for OpenID, an open standard for single sign on across the Internet that has been slowly gaining momentum over the last few years.  The goal of OpenID is to prevent any single company from dominating identity on the Internet, and there have been several calls for Facebook to support OpenID.

I’m going to make a bold prediction: Facebook will never truly support OpenID, because Facebook wants to be the centralized single-sign on source for web applications.  My friend Sean is a firm believer in the idea that ‘social networking is a feature‘– not an application in and of itself.  I believe Facebook feels much the same way– social networking is a feature that lots of applications and sites should have, but there isn’t much value in those sites creating their own, closed social networks, any more so than those sites would derive much value from building their own credit card payment services or even their own supply of electricity.  Rather, those sites should pay Facebook to provide the social networking infrastructure for their sites, with Facebook acting as the one true ’social utility’.

Here’s an example to clarify what I’m talking about.  Let’s say that an online clothing retailer wanted to integrate social shopping features to their site.  When you sign in to their site with your Facebook ID, you can see what other friends are shopping at that site right now, chat with them, see what they recently purchased, etc., etc.  Meanwhile, when you login to Facebook, your News Feed contains information about where your friends are shopping and how they are spending their time online.  Facebook can leverage their network, the platform, and the News Feed to become a traffic director for the Internet (ala Google or Yahoo) and gain control over some incredibly valuable marketing data.  For any kind of community or commerce site, not being a part of the Facebook News Feed would become equivalent to not belonging to Google’s index– it’s as if you don’t exist at all.

I think that Facebook’s current valuation should be roughly equal to the product of two numbers: 1) the cash value of controlling and managing online identity, and 2) the probability that they will succeed.  The challenge is that no one really knows what the first number is (aside from “a lot”), nor do they know what the second number is, although it seems like Facebook has a much higher probability than any other player out there.

Can an Open Network Beat Facebook?

Right now, I don’t believe that an open network, starting from scratch, can catch up with Facebook.  My concern is that an open network is going to have more “friction” associated with adding a data stream than a closed network like Facebook has.  Photos, events, and link sharing are fundamental data streams for people using social networks, but a new user at Plaxo will need to configure access to multiple services to provide those features, whereas they are built-in to Facebook.  It seems like every open network service will need to be tightly integrated with these service providers in order to match Facebook’s ease of use, but if that’s the case, I don’t know how “open” the network really is.

More likely, in my opinion, is that one or more of the other big players– Google, Yahoo, Microsoft, Amazon, and eBay– will figure out what Facebook is up to, although not necessarily before it’s too late to stop them.  The pivotal company, in my opinion, is Amazon– they’re savvy enough to see what’s what and have a large enough user base to be a kingmaker for social shopping and online identity.

I see two possibilities: in case one, Amazon and Facebook form a significant alliance that involves Amazon taking an equity stake in Facebook, Amazon building social search apps on top of Facebook identities, and significant data sharing between the two companies.  In case two, Facebook provides identity and social shopping support for a wide swath of Amazon’s competitors, almost certainly including Barnes and Noble.  I believe Facebook is more likely to go with the second option, since Amazon has a significant investment in their own ID system and platform that they might not be willing to sacrifice.  The concern is that if Facebook’s partneships with Amazon’s competitors works out, it will drive Amazon into an alliance with Google, Yahoo, or Microsoft that could counter Facebook.

That said, If I’m Mark Zuckerberg, I’m willing to bet that the Big Five, each with their own positioning and interests with respect to online identity, won’t figure out how to place nicely with each other until it’s too late (if ever).  It’s not an easy bet to make, but the stakes are so high and the game is so nervewrackingly fun, I don’t see how he could help but go all in.

Labels: social networks

[ 2 comments ]