I came back from a weekend away to find that Nick Carr had picked a fight with Tim O’Reilly about Google, and whether the company’s size and market power has been a result of network effects. This argument was a sort of spin-off of an issue that came up earlier in the week, when Hugh McLeod pondered the topic of “cloud computing” and whether it could produce a giant monopoly. O’Reilly argued that it likely wouldn’t, because it doesn’t benefit from network effects in the same way that Google does.

Nick took issue with this line of argument, asking:

“Is the network effect really the main engine fueling Google’s dominance of the search market? I would argue that it certainly is not.”

In fact, he says categorically that Google’s rise to dominance has “nothing to do with the network effect.”

I don’t know why a smart guy like Nick would make an argument like that, but I think he’s totally wrong. Let’s review what the network effect means: As Tim describes it, it refers to systems that “get better the more people use them,” which is a pretty good paraphrasing of the Wikipedia definition as well as other definitions you can find in various places. The classic example used is the telephone, which becomes more useful the more people use it.

Nick says that this doesn’t describe Google at all:

“The fact that my neighbor uses Google’s search engine, rather than Yahoo’s or Microsoft’s, does not increase the value of Google’s search engine to me, at least not in the way that my neighbor’s use of the telephone network or of Facebook would increase the value of those services to me… Indeed, if everyone other than myself stopped using Google’s search engine tomorrow, that would not decrease Google’s value to me as a user.”

I think Nick is being disingenuous here, and/or splitting rhetorical hairs. To argue that Google doesn’t benefit from the network effect is to deliberately miss the point. Google’s strategic point of differentiation from other sites was the fact that PageRank does a better job of determining worthwhile links. And how does it do that? By measuring how many sites (and which ones) point to them. The more sites, the higher the page ranks. I’m simplifying it, but that’s pretty close to how it works.

That is at least as strong an example of the network effect as the telephone, and probably more so. It’s true that it doesn’t matter whether Nick’s neighbours use Google or not. That’s because Nick is deliberately focusing on Google the search engine — the little box with the search button — and trying to distract you from the other Google: the indexing engine that produces those links, and uses powerful network effects to do so.

I think part of the problem is that the network effect we’re talking about doesn’t really belong to Google per se — it’s inherent in the nature of the Web. But let’s go back to what Nick originally said: He said that Google’s rise to dominance had “nothing to do with the network effect.” That’s just plain wrong. Google may not have invented the thing that produced the network effect (i.e., the Web), but it came up with algorithms that took advantage of that effect better than anyone else.

I’m not sure why Nick wants to pretend that network effects don’t exist, or that they didn’t play a role in Google’s rise to prominence, but he’s going to have to do a whole lot better than that.

Update:

Tim O’Reilly has responded to Nick’s post, and as far as I can tell he says pretty much the same thing that I did, but better — and somewhat longer :-)

Twitter seems to be one of those things that people write about almost as much as they actually use (which isn’t hard, considering the average Twitter post is less than 140 characters). I’ve written my share of blog posts about Twitter, including here and here, and now we have a longish post from Howard “Smart Mobs” Rheingold about why he has started using the app, along with some follow-ups from my friend Tony Hung at Deep Jive Interests and from Nick Bradbury of FeedDemon.

Howard puts his finger on some important things about Twitter that he says makes it interesting as an example of a “smart mob,” including:

• Openness
• Immediacy
• Variety
• Asymmetry

And I would agree with all of those. But one of the most important factors, I think — and the one from which many of the others flow — is the first one: openness. If I send an @ message to just about anyone (unless they have blocked me), I know that they will likely see my post. I can’t think of any other app that allows that, except perhaps the ability to comment on someone’s blog. In some cases, I direct message people (d username) on Twitter because I can’t remember their email address, or don’t know which one of the many they have I should be using.

Facebook allows this kind of thing too, but to a much more limited extent. And messaging people through Facebook — which is another way a lot of people reach me — is much less immediate, since you have to click the link in the email to go to the Facebook message page, and whoever gets your response has to jump through the same kind of hoops. Twitter is always on, and is much faster, provided the person you need to reach is paying attention (my Twitter info is in the right-hand sidebar).

I think Tony puts it well when he says that Twitter is like a big group chat with a wide and varied group of people. I would definitely agree. And like many groups events — parties, etc. — not everyone is listening to everyone else, and there are side conversations going on that you may only hear one side of. But if you want to approach someone, they are as close as an @ message. And sometimes it’s just fun to listen in.

Update:

Jeff Jarvis has a column up at the Guardian about Twitter.

The outrage continues over Comcast’s jamming of BitTorrent and other traffic on its cable network in the U.S. The company has tried to clarify its position by saying that it doesn’t block BitTorrent traffic, it merely “delays” it (and apparently some other traffic as well). As James Robertson and Cynthia Brumfield have pointed out, part of the problem is that Comcast isn’t being very forthcoming about what it is doing at all — in part because the company says it is afraid that providing too much detail will allow BitTorrent users to find a way around the network “shaping.”

I know that the popular position is to slam Comcast for telling users what to do with their network bandwidth, and I know my instinct is the same whenever my ISP talks about BitTorrent or bandwidth caps. But like most ISPs — and cellular carriers — Comcast has a “terms of service” agreement that allows it to restrict what users do with their accounts, so that whatever they’re doing doesn’t impact on others using the network. That’s a fact of life.

The big issue for ISPs is that p2p apps like Skype, Joost and BitTorrent can consume a huge amount of bandwidth. According to some estimates, 10 BitTorrent users on a network node can double the delays that other users experience — and as much as 60 per cent of the traffic on some networks is BitTorrent-related. That may not be a problem for BitTorrent users, but it could severely impact those using other applications on the same network.

Obviously it would be better if ISPs like Comcast or Rogers built out their networks to provide more bandwidth, and it would also be better if there were more competition in the Internet access business. But it’s hard to blame Comcast or anyone else for trying to make sure all of their customers get the service they deserve. Now if only the company would come out and say so.

I really wanted to like the Zonbu, a small, Mac Mini-style network PC that runs a modified version of Gentoo Linux and uses Amazon’s S3 (and a small built-in Flash drive) for storage — and best of all, costs just $99. It seemed like such a great idea, as Nick Carr describes here and Daniel “Fake Steve Jobs” Lyons gushes here. Okay, it has a really stupid name, but then who doesn’t in these Web 2.0 times.

In fact, the Zonbu might even be a great idea — except for the fact that it doesn’t cost anywhere close to $99. Or rather, it costs $99 in the same sense that a computer from Bill’s PC Warehouse costs $200 because it doesn’t include a monitor, keyboard, mouse or any software (which I’ve always thought was a little like taking the wheels and the engine out of a car and then advertising it for sale at $50). As Zoli Erdos points out in his post on Zonbu, the box actually costs about $250 when you go to buy one. In order to get the $99 deal, you have to sign up for two years worth of online data storage, which brings your total cost to about $400.

Okay, $400 isn’t a bad price for a networked PC with automated backup and a small form factor — except that you still need a monitor and a keyboard and mouse. I can buy a fully-configured Acer or HP desktop with all kinds of bells and whistles for $400. It might not fit under the counter or look good tucked onto a bookshelf, but it comes with a 160-gig hard drive, and I don’t have to call it a stupid name.

A recent commentary piece by Alice Mathias in the New York Times says Facebook should really be called “Fakebook” — at least for the student users who first made the social-networking site popular. As she describes it:

“Facebook did not become popular because it was a functional tool — after all, most college students live in close quarters with the majority of their Facebook friends and have no need for social networking.

Instead, we log into the Web site because it’s entertaining to watch a constantly evolving narrative starring the other people in the library.”

She describes the use of Facebook as being like online community theatre, with users putting on different masks depending on which groups or individuals they are connecting with, and what impression they want to give of themselves. This aspect of online behaviour may be more prevalent among younger users — as sociologist danah boyd has described in her research into social networks such as Friendster and MySpace — but I don’t think it’s unique to them.

I think we all have different personas we use, depending on where we are and who we’re interacting with, whether it’s work or home, co-workers or neighbours, family or old friends — people who have only known us as adults, vs. people who knew us when we were teenage hoodlums. The tension between those different personas is why some of us feel a little uncomfortable when we run into our boss wearing a grungy old T-shirt and surfer shorts with a two-day growth of beard and a wicked hangover.

Scott Karp thinks that kind of thing is why Facebook may never really work as a business tool. I’m not convinced that’s true — but it’s certainly going to make things a little more complicated (or interesting, depending on your point of view).