In valuing a social technology, it is impossible to avoid the enhancement in value the network provides. Stated quite simply, social technologies benefit from an economy that awards value to the service as more people join the service. This, of course, is the network effect; a network gains value as more people join the network.
The classic example of network effect is illustrated in Metcalfe’s law. In valuing a telecommunications network, Metcalfe speculated that the value of the network increased proportionally to the number of participants in the network. As a simple illustration, consider a telephone network with three participants. When the fourth participant joins that network, the value of the network increases for all participants in the networks. Odlyzoko and Tilly, in 2005, further refined Metcalfe when they illustrated that the value of a network doesn’t actually increase proportionally, but logarithmically. To illustrate this refinement, each addition to the network adds value, but in reality, more value is added when my relative joins than someone in Dubai. My patterns of connectivity cluster around my relatives, hence Odlyzoko’s and Tilly’s refinement.
Metcalfe’s law provides the groundwork for a substantial amount of applied network effect theory. One of the foremost applications of Metcalfe is to internet technology, particularly social network technologies. One can easily see the lineage that validates the application of Metcalfe – the internet is indeed a telecommunications network, and the value we perceive from additional entrants is visceral.
Getting to the point, Metcalfe’s law holds fundamentally, but the application of Metcalfe to a wide range of social internet technologies proves remarkably flawed. To illustrate: Metcalfe’s law is built on a core assumption that entrants to the network have a limited set of options. In the 1980′s, when someone joined a telephony network, they had but two options – use the phone or not use the phone. This binary calculation assured that whenever the phone was taken advantage of, the user was getting full value. Ethernet is another telling example; when you plug your computer into an ethernet network, the only options you’ve got are to accept packets or not accept packets. Either way, when you use the network, you’re getting full value.
This notion of “full value” makes the mathematics of network value calculation quite appealing. If everyone on the network gets the same value from using the technology (everyone has the same options – i.e. call or not call on the phone), then valuing the network is absolutely possible. When using Metcalfe (or Reed, or Odlyzoko and Tilly’s refinement) to value a network, the core assumption is that the value we derive from the network is binary – this works for things like ethernet and telephony, but the mathematics prove to be overly crude when applied to social network technologies.
I’ll try to illustrate a comparison. Indeed, Myspace’s network provides two options to you – you can either join or not join the network. If we wanted to apply Metcalfe to Myspace, this is where we’d stop. However, the value in Myspace is much more nuanced than simply being on the network; you can take value from the many things you can do on the network. The network offers a myriad of associations, including friending, grouping, messaging, browsing, stalking – actions that create a compound value that is unique for each network entrant. Indeed, each new entrant to Myspace offers others in the network the chance to create these relationships, but these many types of relationship create a value continuum – which is different than a value binary.
Therefore, the fundamental flaw in applying Metcalfe to social technology is its inherent lack of nuance and granularity. When people join the network, they are given more options than simply connecting; the network is worth the sum of associations and actions that are allowed in the network. We must instead think of network value in terms of a network effect multiplier, as the actual value a network adds to an application is under the direct control of the application designers.
Consider flickr. Flickr is a socially-enabled application built around photographs. Stripped of flickr’s social tools, the service would provide a core value to its users – it would be a very high quality image host and archive. This core value is the “real” economic value of the product; this theory is consistent because flickr users have proven willing to pay for their services. Indeed, the core value provided by flickr is important, but the core value alone is not flickr’s total value. Enter the network, and network effects. Flickr is a socially-enabled tool, allowing users to connect around photographs. The social actions that can be taken in flickr are fairly limited; comments, page views, connections, groups and pools – these are fairly “commodity” social tools (in a sense, all of the social actions are native to the users as they have been previously pioneered). This is lightweight social networking, with very low barriers of entry; the network effect is light as well. To understand the final value of flickr, we multiply the core value by the network effect value (the network effect multiplier).
As a contrast, consider Myspace. Myspace’s core value is quite low. When you log onto Myspace you get a profile, a message box that doesn’t interoperate with the rest of the world, some limited image hosting, etc. However, the network effects of Myspace are tremendous. The size of the network and myriad uses of the network create a network effect multiplier that is much greater than flickr’s. However, since the final value of the network is a function of the core value and the network effect, we see a balancing function.
This balancing function is the key to valuing social technologies. The core value is the raw economic value the service provides to the user. The telephone was useless without the network; however, a service like flickr, or even Myspace would provide value stand-alone. With the telephone, you only had two options when using the network – call or not-call. In flickr and Myspace, you literally have millions of ways to use the network, each with a different value outcome. The network adds value to flickr and Myspace, but the value it adds is distinctly more nuanced than what Metcalfe proposed – and the value the network adds is in the hands of the designers.
As social networking becomes commoditized, as more and more sites make social a part of their experience, the value-add of embracing social will need to be quantified. Metcalfe’s theory is absolutely valid in context, but the applications to social technology lack the nuance that will be required to quantify cash outlays. The good news is that quantifying the value of the network isn’t overly complex. We start with the core value of the service (the non-social value proposition), and the network effect multiplier. As the network effect multiplier is contingent on the site’s design, this can certainly be quantified. Network-enabling a product does not produce a binary value-add; some sites will add lightweight social networking to enhance a core value tool (flickr), whereas others will derive almost all of their value from the network (Myspace). The key to understanding this is knowing that the value provided by the network is variable, and the outcome value of the service is contingent on the core value and the network effect multiplier.
In thinking this through, I’ve tried to focus on the value of the network. However, divorcing the actions you can take in the network from the network’s value kept leading me back to my initial train of thought. The value of a social technology’s network simply can’t be divorced from what you do in the network – the actions you can take are deeply nuanced. We’ve matured from the binary assumption of communicate/not-communicate that network effect theory is built upon. Of course, Metcalfe’s core theory still holds for things like telephony and ethernet networks. However, humans are not computers; our actions, and the derived actions of network participation have variable values. The compound effect of our actions is the network effect multiplier. As we develop socially-enabled applications, consideration of this network effect multiplier will prove useful in determining the value of our labor.
I’m going to officially call this an idea in progress. I’m really struck by the potential value of articulating this properly. At this stage, I’m just at the beginning of doing so. Thoughts, comments and feedback are certainly welcome.
Tags: economics, information, networks, social networks
There is something there in your train of thought. Stickwith it, I would however take exception to the title “Metcalfe’s Flaw”. He deserves better.
Thanks – and I assure you this isn’t a dig at Metcalfe. A number of times in this post I assert the validity of his approach for telecommunications networks. The flaw lies in the wholesale application of his law to social technologies by pundits. Indeed, a lot of his core philosophy holds up, but it needs to be adjusted for new contexts. Again, the problem here isn’t with Metcalfe, but with those applying his law without understanding the context.
I have posted my thoughts about your theory on my blog.
“If mashups continue to be the norm and social networking sites such as MySpace begin to offer open API’s , then any new participant could bring in or develop third party value beyond the control of the designers.”
more
I posted some more thinking on this topic.
One of the outcomes of this that I find interesting is that you can use this reasoning as a predictive tool for future social networking sites. The Flickr analysis shows us that sites can use social networking aspects as value-added features to arrive and something new and different, and that value can be quantified in some way.
It also points to the most interesting prediction, I think…that if we meta-network, the outcome expected from a pure Metcalfe analysis would be a many fold increase in network value. But of course, that might not be the case in real life. If MySpace could suddenly interact fully, at a social level, with Facebook, would that benefit each network? I’m almost certain it wouldn’t. At the same time, a mediated interactivity might produce a great deal of value…this is a ripe area for Web 2.0 sorts. A site that allows us to mediate between our social networks in a robust way (a more “connected” sort of ClaimID) would be an absolute winner.
Fred,
Of note, the more you write, the more interesting you get. I’ve read your blog and news about your activities for some time now. What I have a hard time grasping about your social networking ideas is, other than volume and opportunity, how have social networks changed us over time? I suggst people have not changed much, just their opportunities to join in on the network.
You mention MySpace giving you the same option as the telephone to connect or not connect, but MySpace also provides “nuance and granularity.” I think the telephone gives you many more options than to connect or not connect. Who you call, what you talk about, plans you make, etc, vary greatly with each caller as, I assume, with each MySpace member. (BTW- the preview stops here, I don’t know if you are getting the rest of this post)
I remember when telephone users were on party lines and one could get quite a bit of local news (gossip) by just picking up the phone and listening. In a way, I equate this eavesdropping with the telephone similar to someone reading your profile on MySpace. With party lines you never knew who was listening in on your conversation and with MySpace and Facebook you never know who is looking at your profile. I’m not arguing our new social networking has not changed us, I’m just not sure how.
Everyone – thanks for responding. All your takes on this have been extremely thought-provoking. Let see if I can’t respond a little. Michael and Jason – I think a lot of what you say is correct and absolutely fleshes out this idea. We have the power to control the sociality of a service. Indeed, a number of social networks have caught on in new places because of their simplicity (going back to Michael’s response). There’s a lot of value there. The actions in the network are not binary – we’ve got many different ways we can wander and connect in the network now.
Raymond, your response illustrates the challenge I feel in making this argument. At one level, both Reed and Metcalfe are only interested in valuing a network. They have divorced “content from presentation” in a sense. The phone provides the binary access to the network (call/not call), but of course over the phone you can do many things that are siimilar to things you can do in social networks. However, when you do these things over the phone, you aren’t creating new paths in a network. That you are friends with someone doesn’t mean more or less people will be able to call that person. In a social technology, the process of friending actually does add complexity to the network, increasing paths of transversal for others. Even though the actions are similar, the network (as in the actual network layer) does get more valuable only in one of these contexts. I hope this makes sense. I’ve struggled with the problem of enumeration in trying to keep my thoughts straight on this. I think I need to do some diagrams, really ;)
Just a joke, but to lighten up you day: Your actual network layer of friends can increase with the telephone if you are Paris Hilton and someone downloads all your telephone numbers and puts them on the Internet for all to see!
Thanks for your post and your comments: the idea that usage can be different is interesting, and very new. I’d love to comment, but as it is quite precisely my PhD topic, I feel a bit overwhelmed.
I was wondering: having other people use flickr or Wikipedia a way different than yours, does it improve your own experience? If not (e. g. using phone lines to send faxes) than what you have is simply two different parallel services, with little externalities; if it does (e. g. you are an amateur historian, and having developpers improving the wiki engine is usefull) then you face a completely new ploblem: I can’t think right now how to model it, but I promise to point you to any early ideas on the topic.
You are right to point out the flawed reasoning behind valuing social apps by the number of available connections.
I think we’re talking about two different “flaws”, though:
First, an applications “core” value is, of course, distinct. You make this point regarding Flickr. The social network aspect also has a variable value that seems to often be discounted, though. Simply tracking who your connections are is basic, sending messages is basic, but we also have many apps today that let us browse our circle’s content and even use the preferences of our friends as input for recommendations. These different models have distinct basic values that are then amplified by a “Metcalfe multiplier.” This difference in the value coefficient is perhaps your “nuance.”
Second, that Metcalfe multiplier is not necessarily as straightforward as the original network value law. Rather, the “granularity” is in how connections are made (and utilized) in practice. If my site has 100,000 users, but each one has only 2-3 connections, then my multiplier is much lower than if each had 10. Of course, if each user has 1000 connections, then some “nuance” values might be diluted (such as group messaging) but others (such as fashion-related affinities) might scale linearly.
Maybe the best takeaway is that network tracking is a type of value that’s easily portable. If social nets were as sticky as their creators like to tell investors, then we’d all still be using Friendster.
Hi Fred,
Thought-provoking post as usual – I commented on my blog as well.
I’m not sure I follow the point about the network effect multiplier (NEM) being under “the direct control of the application designers.” That seems like network effect is predetermined, valued during developement, without factoring in adoption by users. You clarify the statement later by saying that our actions on the network compose the NEM.
So is it our participation in the network or the features/functionalities in the network? If the former, we almost have to go back to a Metcalfe-type model. Granted it’s binary (or digital) approach is oversimplified in analyzing the analog nuances of SNS…
Looking forward to a follow-up to this WIP…
The weakness in your analysis is that you assume that people joining the network benefit the network.
Consider: 10 people on an early phone network. They swap tips and gossip.
Then a salesman gets a phone and realises sales can be made over the phone rather than at doors. Voila, telemarketing.
The 10 people who were on the network perceive a drop in the network’s utility: if they answer the phone, they might not get a tip, they might have their time taken up by someone selling them something.
To the telemarketer, it’s a great thing. 10 people discomfited, one person pleased. Is the network better for the extra connection?
Similarly for spam and email. And blogs and comment spam. And, IME, for social software tools. You have to accommodate the likelihood of a downside too – in fact, people joining the network will (I’d predict) add value according to a normal curve, centred near zero. Most people don’t add much. Some take away. A few add hugely; a few take away hugely. What’s key to the usefulness of a network is how wide the variance of those on it is – not how big it is.
On that basis, you’d expect that small will almost always be better. Social software, therefore, contains the seeds of its destruction if it expands.
I saw your website is very good, so you have the opportunity to sell space ads on you blog…
i receive 72$ every month doing nothing, there is just 6 links at my website)))And you’re website is much more better then mine.
Payments via PAYPAL and CHECK!!
http://www.text-link-ads.com/?ref=29747
try them…
After you’ll sign up here write to me… and i’ll tell you how to optimize it for well… and give you several good tips…
1. Chinese Scammers, Worldwide Business
2. Ipod world
3. GIFT CARDS
4. Mobile Phones
5. Laptop NEWS