Looking through the wrong end of the wire

There’s a fascinating — and entertainingly pissy, if sometimes obtuse — argument going on over Metcalfe’s Law (which states that the value of a communications network grows exponentially as its number of users grows). On one side are three authors of an article in IEEE Spectrum, who insist that the law was wrong and even dangerous, for it justified the first internet bubble, and they fear it is being used now to inflate a second social bubble. On the other side is Ethernet inventor Bob Metcalfe himself, arguing at VC partner Mike Hirshland’s blog that his law is not only still valid but, when you tie it with Moore’s Law, it leads to the Law of the Long Tail.

And I will argue with them all — not against Metcalfe’s Law but against the way they value networks. The IEEE authors try to tote up the value of a network in terms of who owns it: the market capitalization of companies controlling networks. In other words, they — like telecoms of old — try to value the network at the center. But that is no longer how networks are valued at all. No, now they are valued at the edge. You and I value the networks we choose to connnect to in ways only we can measure. Metcalfe starts down the right path when, as Hirshland summarizes, he argues that with social networks “we need to consider not just the number of users but also the affinity between the members of the network.” Yes, and each network connection we make carries an intangible, personal quality that has direct impact on how much we value those networks and thus how much they are worth as a whole. We tote up our own value in terms like trust, engagement, joy, relevance, excitement, reputation, need, sex, and money. Human networks must be measured on human terms.

Network providers used to try to measure the value we put on networks negatively, in terms of switching costs. That was what AOL counted on for too damned long as they thought it would cost us too much hassle to switch off our AOL email addresses. Ha! But we each value a network positively on what it brings us. That is different for each of us and each of our connections. For example, I see no value, personally, in LinkedIn; it has never done a thing for me but attract a new form of spam, inconvenience, and embarrassment when I don’t link to someone. But for others I know, LinkedIn provides jobs, business, income, reputation; it is damned near invaluable. Similarly, I see little value in MySpace; for middle-aged me, it is merely a curiosity. For others, of course, it gets them songs or gets them laid. Put a price tag on that, if you dare. I see value in my Treo phone because it keeps me connected to anything, anywhere, anytime. I see great value in having a blog in the ‘sphere, for it brings me learning, ego gratification (I admit it), jobs, and money (but no sex).

An important factor in this is openness. The more open the network, the more valuable it is — but the harder it is to own, and thus the harder it is to value in old terms of ownership and market cap. That’s what really argues against the IEEE authors. They are trying to put a corporate value on networks. You can’t. That’s like trying to value air… or the internet. They defy ownership.

Can a network be too open? Of course. Email, one could argue, is too open because it permits spam. It remains valuable only if I gain control over that spam thanks to enabling tools. Usenet was too open and there was no such control, so its value sank to nil. Some wonder whether MySpace will be too open or at least too big: When everyone is your friend, you have no friends. So here we see the value of niches and of communities: the right people in the right relationships. Small is the new big.

Does this argue against Metcalfe? Does it show, as the IEEE guys want to, that the exponential growth in the value of a network does not continue and, indeed, topples? No, the problem is that all these calculations leave out the most important X factor to which Metcalfe alludes: affinity. And affinity is fed by many of those human intangibles I listed above: relevance, emotional connections, convenience, reputation, and on and on. So, I will argue, the value of a network should be calculated with a multiplier, which is the value a network’s members put on it:

Network value = the sum of the value each member of the network places in it.

But, of course, that is incalculable (what, again, is the dollar value of love?).

Does this further argue against Metcalfe that small networks can have more value than big networks? No, because one should value a network as the sum of its networks. We see the internet that way. We also should see the blogosphere that way. This is why I continue to think it is absurd and wrongheaded to analyze the blogosphere on its supposed A-list. The vast majority of people who read blogs never read any of the blogs on that A-for-alleged list. They read and interact with the ones that are meaningful to them. The blogosphere is not the value of the few on top or even of the total but instead of the unlimited connections enabled within. This is also why I see such power in networked journalism: the network is additive.

This is also about the mass of niches. When television as a medium stopped having to serve everyone at once — when cable, VCRs, and now the internet allowed it to serve smaller interests, tastes, and audiences — television as a whole grew; this explosion is far from over. Fewer people watch HBO than NBC but those who do clearly value it more because they pay for it.

And this is about the value of being the right size: I value the Continental President’s Club because everyone in the airport does not belong; if everyone did, its value would fall to nil. This isn’t about snottiness. It is about control.

So all of this leads to my law — everybody has to have a law — which I think I first stated here and have restated ad nauseam.

Jarvis’ Law: Give the people control, and we will use it. The corollary: Don’t give us control, and you will lose us.

The more we control the network at the edge, the more valuable it is to us, and the more valuable it is as a whole, but the harder it is to own and control. Yet that doesn’t stop you from making money. See: Google. It grew by making connections — establishing unlimited networks of information and now advertising — with content and connectivity it did not own or control. See Skype, which didn’t so much grow its own value as deflate the value of its competitors down to their true worth as communications networks stripped of their monopolistic advantages. If you want to talk a bubble, there was none ever bigger than the artificial bubble of the telecommunications companies and their closed networks; open networks have certainly popped that. (See Isenberg’s Law: “Just deliver the bits, stupid.” See also Yochai Benkler’s The Wealth of Networks.)

I think the next valuable network will rise out of helping us find the good stuff in video anywhere — not just on networks and not just on YouTube, anywhere.

But the truly valuable network, the network of networks, the unbreakable bubble of bubbles, will be the one that manages to bring people together wherever we are, not just on MySpace (read: RupertsSpace), not just in Flickr or Del.icio.us, and not even just in the blogosphere, but everywhere. The internet doesn’t need more social networks. The internet is the social network. We have our identities, interests, reputations, relationships, information, and lives here, and we’re adding more every day. The network enabler that manages to help us tie these together to find not just connections or email addresses or information or songs but people — friends, colleagues, teachers, students, partners, lovers — across this open world, that will be the owner of the biggest network of them all: The Google of people.

I’m no mathematician or scientist, so I have to express this in words, but here’s the way I calculate the value of networks:

The Law of Open Networks: The more open a network is, the more control there is at the edges, the more the edges value the network, the more the network is worth.

The business lessons from this: Any choke point of control, via ownership, decreases the value of the network. Enablers increase the value of the network. The network will abhor and find ways around choke points. The network will value enablers and that is the point at which value may be extracted from the network. The value in networks in the open future is not in ownership and control but in enabling others to control.

: And as if all that’s not enough, see also Tom Evslin and Fred Wilson on Reed’s Law — which holds, in Fred’s words, that “if each node of the network was itself a network (a GFN) then the value of the network scales with the exponential of the number of nodes in the network” — and my clumsy efforts to get my head around it here and here. Fred Wilson begs Metcalfe to also tackle Reed. See as well Umair Haque on Google and Reed and on Metcalfe and the edge.

: Here’s Om Malik on Metcalfe.

: Proving that one Hugh cartoon is worth 1,200 of my words:

  • Very eloquent, as usual.

    If you recall, I scribed some parallel thoughts here… http://gigaom.com/2005/09/08/inherent-truths-and-value-of-community/

  • Of course it’s about the mass of niches.
    We’re using the infrastructure of the network (as big as the Internet or as small as a group on MySpace or a pool on Flickr) to serve our own niches and find the people with common interests.
    In many cases, we’re using their networks to build horizontal bonds, strengthening our social network.
    Now let’s play the game of applying this to newspapers.
    How can a local news organization give its readers/former audience the infrastructure to build their own social networks?

  • Yes. Exactly. Social networks on the Internet follow the pattern of the Internet itself: it’s built to route around missing or suspect nodes, and so are they. The conversation goes on regardless of the particular set of network nodes it’s currently inhabiting; if those nodes attempt to direct its direction in any way, it will simply go Elsewhere.

    Martha Soukup had it right, when GEnie flew to pieces in 1994: “It’s the community, stupid!”

  • Pretend to share and gain nothing. Share and gain a bit. Gain everything to become confused. Gather 200 people to create a village. Connect all human life to live in an illusion. Think binairy and develop a new wire…

  • Old Grouch

    Re: the IEEE article, I just love it when people try to apply hard mathematics to sociology:

    We admit that our n log(n) valuation of a communications network oversimplifies the complicated question of what creates value in a network… [it] cannot be proved, in the sense of a deductive argument from first principles. But if we search for a cogent description of a network’s value, then n log(n) appears to be the best choice. Not only is it supported by several quantitative arguments, but it fits in with observed developments in the economy. The n log(n) valuation for a network provides a rough-and-ready description of the dynamics that led to the disappointingly slow growth in the value of dot‑com companies.

    We have, as well, developed several quantitative justifications for our n log(n) rule-of-thumb valuation of a general communications network of size n. The most intuitive one is based on yet another rule of thumb…

    In other words, “our SWAG is better than their SWAG because…” ;-)

    The rest is an apples-and-oranges argument. Since there’s no agreement about what “value” is supposed to represent, it’s no wonder different people get different answers: Jeff (and Metcalfe) are talking about “value (“usefulness”? “willingness to pay for”?) to the user,” while to the IEEE guys “value” appears to be limited to “What can we get if we do an IPO on this?” Isenberg’s “Just deliver the bits, stupid.” network will have less value (in the IEEE sense) than an identically-capitalized “controlled” network, simply because it lacks the controlled network’s rent-seeking opportunities. (Why do telcos fear net neutrality?) Meanwhile, as Jeff notes, it doesn’t matter what the market cap of myspace is: If he doesn’t use it it is of no value to him as a potential user/customer.

  • One of the interesting implications of peer-to-peer networking is the erasure of the distinction between “the network” and “the edge”. We used to distinguish these terms by a simple test: if you route packets, you’re the network, and if you don’t you’re the edge. But Skype actually routes packets through a series of end systems, creating a virtual packet network that goes from edge system to edge system.

    So it doesn’t really make any sense to get all hung-up over the old-fashioned edge vs. core dichotomy.

    Similarly, it’s a fallacy to insist that users want “control” over the network. That would imply they want to run full-blown routing protocols, network management systems, firewalls, and spam filters. I don’t think they want control as much as they want choice with respect to services, prices, and responsibilities. When you “control” a network, you take responsibility for it, and users don’t want to be sys admins, they just want to play.

    That being said, Metcalfe and Reed both err in failing to appreciate the negative effects that limit network value: spam, viruses, and idiots. Only a naive Utopian fails to include them in his value calculation.

  • Richard,

    I’m not sure that’s the type of “control” we are talking about.

    AOL or Mobile providers imposed a certain type of “control” even though they were/are using the same open internet protocols as everyone else.

    But your point is well taken.

    Perhaps it’s not control that users want, but they certainly don’t want anyone else to control the network.

    That is, they don’t want anyone putting limits on how they use the network.

    For example, if MySpace says “no more flash”, it opens up an opportunity for a competitor to “we’ll take you flash widgets!” and get some converts.

  • In network engineering, “control” isn’t a bad thing, it’s the essence of the problem. In fact, many of our best engineers are trained in something called “control theory,” a discipline that prevents networks and similar systems from becoming unstable. Control theory places the limits on things that have to be limited in order for systems to function well.

    Users therefore do want someone else to control the network, and most of us are mature enough to realize that “all you can eat” being an economic rarity, limits are what we contract for. Most users probably appreciate the fact that they can buy an service plan that limits their bandwidth because it also limits the price they pay and requires those with larger appetites to subsidize their network usage.

    If the network isn’t controlled, it simply collapses, and we have plenty of experience with that on the Internet. The Original Design by Cerf and Kahn lacked control, and consequently underwent “Internet congestion collapse” as soon as there were enough users to overload its internal links. Control – in the Transmission Control Protocol – alleviated this problem for a time, but peer-to-peer is bringing it back.

    Can we solve this problem by adding more bandwidth to the Internet? No, because the laws of physics get in the way. And oddly enough, the closer to the center of the network you get, the better you’re able to ensure stability along with fairness and all the other good properties of user-friendly networks. You can’t manage traffic strictly at the edge. Bummer.

    Are users as stupid as Jarvis implies? Clearly, many are, but by the time our pimples clear up we generally discard such silly demands for “control” that can’t physically or practically be done, and settle for a network that allows us to do what we contract with the network provider to do.

  • Pingback: Jackie Danicki » One for the metrics fetishists()

  • In any network the edge is where ever there is a user. No one is ever not on the edge.

  • Heavy, dude.

  • Jeff,

    We have a forthcoming plugin for WordPress that you might enjoy quite a bit. It is called Buddy Cards and starts to tie social nets and identity together. Here’s some more on what we are building at 30Boxes:


  • Changing MySpace to RupertSpace was inspired, but I think we should not stop there. I’d be most content if we could change Rupert Murdoch to ‘Reynard Murdoch.’
    And , voila and therefore: ReynardSpace.
    We neatly retain the ‘R” but smartly tie the former Rupert to Reynard the Fox, that rotten trickster of medieval lore, the devious anthropomorphic bugger who outwitted all the other mammals, kings included. No loss in alliteration here, a happy thing.
    The fact that Reynard runs the partisan Fox network never occurred to me untl after my aha moment. I clapped my forehed so hard that I reeled. Reynard for Rupert is so apt that it hurts. Yes, the bloviation station nicely continues the work of Reynard the Fox, one dodgy soundbite at a time.

  • (nitpick)


    you are, I assume unwittingly, propogating a misuse of the term “exponential growth”.

    While Moore’s law is about exponential growth, Metcalfe’s is about a polynomial (specifically quadratic) growth which, beyond some crossover, is vanishingly small compared to exponential growth.


    – exponential growth with respect to some input ‘n’ is described by a^bn, where a and b are constants

    – polynomial growth with respect to some input ‘n’ is described by an^O + bn^(O – 1) + cn^(O-2) + …., where a-z are constants and O (“order”) is a positive integer constant. In the particularly well-studied case where O=2, it is referred to as quadratic.

    Note in particular that for polynomial growth, the ‘n’ always appears to the left of the exponential operator ‘^’, for exponential growth, it appears to the right.)

  • Good post, Jeff, although I have one small quibble:

    Network value = the sum of the value each member of the network places in it.

    But, of course, that is incalculable (what, again, is the dollar value of love?).

    Network value as you define it may be difficult to calculate, but it is by no means a hopeless problem — in fact, the next Holy Grail of searching is to make this “incalculable” quantity as fundamental to next-generation search engines as PageRank is for Google. cf. John Batelle’s The Search for a good summary of how these highly personal subjective valuations of the Web will power search in the future. It’s pretty wild stuff.

  • Pingback: Classyfeeds » Network fluidity is the key()

  • Richard,

    You say peer-to-peer is bringing back a problem that TCP had alleviated for some time.

    Why has peer-to-peer use grown?

    Users are routing around central control.

    How can you say users settle for “what we contract.”

    My cable company blocks port 25, I use another. My IT department says no IM, I use port 80.

    Does that sound like people want control?

    You technical points are brilliant, but I think users don’t care if something is technically feasible, they are going to try and get.

    And often they win.

    Sometimes everyone loses. Like clogged University networks and MP3s.

    It’s a constant battle I guess.

  • Richard,

    I’m 340 pounds, it comes with the territory. :)

  • Pingback: Better Living through Software » Blog Archive » Idea Slaves()

  • Pingback: CaNN :: We started it.()

  • Pingback: Fudeblog by Cesar Cardoso » A rede é o…()

  • Jeff,

    Wow this network conversation is out of control. This law vs. that law. The bottom line I agree there are too many people trying to develop the next best social network. The issue I try to address for our intended market is HOW TO GENERATE VALUE. How can we provide value. To me personally it means a unique and well positioned blend of content, communications, social network strength, target network strength, credibility, verification, marketing power, interoperability, UI design / feel, brand, etc…

    For social networks focus on a target market and power it with compelling media and sophisticated technology. Run like hell and try to generate a strong base of users get acquired so you don’t die when the shake out and consilidation happens within 3 – 5 years.

    Just one entrepreneurs thoughts at 3 AM on the whole WEB 2.0 craze!

  • Pingback: BuzzMachine » Blog Archive » The internet is the social network()

  • Pingback: Design Idea » On the internet, nobody knows you’re a dog()

  • Pingback: BuzzMachine » Blog Archive » @Facebook @Shark: jump?()