Scott Cleland of Precursor has posted a very interesting analysis of Google’s usage of bandwidth and the associated costs. He claims that Google is underpaying for its bandwidth by a factor of 21 based on a variety of calculations and estimates. The analysis is sound up to a point but it then makes the mistake of conflating two things that are really separate and don’t make much sense being treated the same. I posted a comment on his blog but since it hasn’t appeared (neither have any others) I’ve posted it here too.

In essence I think Scott’s doing a solid job of representing his clients - the telcos - but he also repeats a trope that began, I think, with Ed Whitacre - that Google is somehow using telco bandwidth for free when it should be paying for it. I use an analogy below to critique the analysis because this stuff is complex enough to benefit from it. Let me use another here to critique this idea that Google somehow ought to be paying its fair share. Say a store in a certain area suddenly starts doing great business, and customers are flocking to it on the local bus system. Would it be reasonable for the bus company to start charging the store to recoup some of its costs when all its customers are already paying the prices it has decided to charge in order to ride the bus? No. If it is unable to fund its costs from the prices currently being paid then it needs to charge more or seek ways to reduce its costs. The store isn’t the problem - in fact it’s doing good by creating more demand for the company’s services.

The telcos have no business asking Google to fund the costs of consumer broadband connections any more than the bus company has any right to ask the store owner to subsidise bus tickets. With that, I’ll leave you to the comment I posted on Scott’s blog.

Scott,

You’ve done some very interesting and useful analysis here. Thank you for sharing it with us.

However, one criticism is that you conflate two things and treat them as if they were the same and part of the same category: namely, consumer broadband spending and service provider bandwidth spending. These two things happen at opposite ends of the internet value chain and are entirely separate.

In chart VI of your report you act as if consumer broadband and dial-up internet access spending and Google’s spending on bandwidth were the only chunks of money being spent on bandwidth/broadband in the US. This is, of course, not true. Google’s spending should properly be put in the context of overall service provider spending on bandwidth, not treated as part of consumer internet access spending.

Measuring Google’s spending as a proportion of consumer internet access spending is meaningless - it’s like asking how much it costs the Yankees to drive their players to the stadium as a fraction of how much it costs all the fans to get to the stadium. You’ll get a number of out that but it won’t mean anything.

I would suggest calculating how much Google pays for bandwidth as a portion of all the spending by service providers on bandwidth used to serve US consumers. Your numbers might be just as stark, but at least then you’d be measuring the right thing.

The study attempts to push a theory that AT&T under Ed Whitacre but also others among the broadband providers have attempted to push for some time, which is that consumers and Google and others should all just pay their “fair share” of the costs of the Internet. However, this simply isn’t the way free markets work: the fact is that there is a value chain and different players pay for different parts (as they do in any other free market).

Google pays less than it otherwise might because it has so many peering arrangements (entered into voluntarily by the various parties to them) which it doesn’t pay anything for. That’s the way the system works, and large broadband providers benefit from it too. AT&T, Verizon, Qwest and the cable companies are perfectly free to develop their own business models to compete with Google and are entirely within their rights to sign whatever agreements they want to. No-one is forcing them into anything. They can also charge their customers less or more if they think that will solve the problem. The real issue here is that bandwidth use is skyrocketing and broadband providers don’t want to pass the costs on to their customers, but those customers are causing the increase in costs and should rightfully pay for it.

I’m not a stooge for Google or the broadband providers (though the broadband providers are clients of mine) but I think this analysis needs some tweaks before it becomes really meaningful. Thanks again for some very interesting groundwork though.

Note: I’ve heard Scott argue against net neutrality at a couple of industry events and I think he actually makes some really good arguments (although I think there - as here - he sometimes overplays his hand). I have a lot of respect for the work he does and I’m grateful for the analysis he’s done here too.

Note 2: Google has posted its own critique / response here.

I just re-read this recent post from the Google Public Policy Blog, and I still think it’s a lot of pie-in-the-sky nonsense. It really feels as though whoever wrote it either doesn’t know enough about the subject or has dumbed it down for readers to the extent that it makes no sense. Although the example cited is apparently real, the model described is far more complicated than it at first seems, and the chances of it being implemented on any large scale are virtually zero.

When I first saw the headline, “What if you could own your broadband connection?” I assumed that it was going to be about Google’s plans for wireless services - a little late perhaps given that they failed to secure any licences in the 700MHz auction, but it would have been interesting as an academic exercise. But no, it turns out they’re talking about fiber connections:

It may sound strange, and it’s certainly not what we’re used to. Today we have a “carrier-centered” model; phone and cable companies spend billions to build, operate, and own the “last-mile” connection — the copper, cable, or fiber wires that come into your house. Individual consumers then pay for particular services, like phone service or Internet access.

In turn, we tend to think about broadband deployment in carrier-centric ways. If we want to see super-fast fiber connections rolled out to consumers, the main question appears to be whether carriers have appropriate incentives to invest.

But there’s no law of nature that says this is the only possible model. Many businesses, governments, universities, and other entities already own their own fiber connections, rather than leasing access to lines. It may also be possible to find ways for consumers to purchase their own last-mile strands of fiber.

Here, as anywhere, there would be certain advantages that come with ownership over renting. No one necessarily needs to own skis or a car, but many of us do. If you owned your own fiber, you’d be able to connect it to a service provider of your own choosing. Over time, you might save money, and it could make your house more valuable to have a fiber “tail.”

I think the examples used are disingenuous - fiber cables owned by businesses or universities are often for private networks, whereas the whole point of a broadband connection is connecting to the public Internet. Even where Internet access is “owned” by someone other than the carrier, that makes no sense until you put equipment at both ends which allows the cable to be more than just a piece of hardware. And you need a carrier willing to both connect to the business end and to provide you with the appropriate equipment at your end to make that cable work. And of course, the cable itself just connects you to the carrier, which still connects you to the Internet, so they still own the vital connection even if you own the piece of string between your place and theirs. You therefore have no more real ownership over the key piece of the puzzle than you do today.

Then there are all the technical issues involved with maintaining and fixing such a cable. Even if you can get a service provider to hook you up to the Internet, you still own the last mile, and would be responsible for fixing it if something went wrong. Your service goes out - how do you figure out where it’s broken? If it’s someone digging up the road, how do they know who to notify before they do so? And how do you exercise any authority over them to get them to fix it quickly? What if something else goes wrong? Who’s going to fix it for you? Certainly not the local service provider you’ve deliberately bypassed…

I could go on and on - only three commenters have bothered so far on the post itself, so it seems most people haven’t taken it too seriously. But this feels like another one of those occasions on which someone has over-simplified a complex situation in a way that says, “now why in the world do we do this the way we do? Look how easy it would be to do it differently - and better! More freedom! More control for customers!” and so on. It’s also a favorite tool of politicians selling quick fixes to intractable problems usually caused by other politicians…

Ironically, I think there’s a lot more potential for the kind of model the Google blogger is talking about in the wireless sphere. There, no cables are necessary so ownership is a non-issue. It really is about simply having the right hardware at your end and a provider willing to hook you up at the other. With multiple wireless providers being able to serve the same area without digging up the streets there’s potential for real competition with none of the hassles associated with a wired local access network. You’re still going to need a service provider to hook you up unless you’re willing to become an Internet node in your own right. But there is at least the potential for greater competition and more choices for consumers.

Google, of course, merely participated in the 700MHz to try to force the existing carriers to create this kind of model, backing out of the bidding themselves when they thought they’d achieved their aims (possibly erroneously). Perhaps if they’d stayed in they’d have been able to make this kind of model a reality.

The standard story told in the US media, by certain politicians and by consumer rights groups is that the US lags the rest of the world in broadband, with studies often placing the US well down the international rankings. This week there was a report from the INSEAD / the World Economic Forum which contradicted somewhat those glum findings and accords more closely with my own views on this topic.

The problem with many of these reports is that they focus on price and availability of high speeds without investigating the negative effects associated with heavy government intervention in the market. As an example, Japan is often cited as the beacon of international broadband, often closely followed by Korea, but in both of these countries the government has intervened in a heavy-handed fashion to achieve the results seen today. European countries often also score highly, often because of the competition introduced via local loop unbundling regulations - a less intrusive form of intervention than in Japan and Korea but nonetheless a much more aggressive form of regulation than that which applies in the US.

Another flaw is that it is assumed that faster speeds are always a good thing, and that diminishing returns never set in. The fact is that, beyond a certain point (currently around 10Mbit/s or so) extra bandwidth is just that - extra. There is almost no application in existence today which requires more than 20Mbit/s when in peak use, and so 100Mbit/s is a senseless benchmark. Getting more of the population to 5, then 10 and ultimately 20Mbit/s is a reasonable goal, but beating up on the US for the paucity of 100Mbit/s connections is an exercise in futility that could lead to bad investment decisions. Verizon’s FiOS infrastructure is certainly capable of delivering that kind of bandwidth, although AT&T’s U-Verse probably isn’t under the present architecture. But the point is that it doesn’t matter.

Another thing that’s rarely examined is the pricing side of the equation. Providers in Korea in particular have a very spotty financial history due to the suicidal price wars they’ve engaged in. But that competition has been spurred by the fact that they - like the US - have inter-modal competition between various infrastructures, not just regulation-based service-level competition on a common infrastructure. The latter gets quick results in terms of number of providers and price competition but it rarely foments real innovation because the underlying wholesale services everyone is using are the same. With a shift from bitstream to local loop unbundling products that changes somewhat, but competing infrastructures - especially ones built on fiber - are much more likely to provide real differentiation.

Hence the massive speed and price competition that’s been triggered in the US in areas where fiber has been rolled out by either AT&T or Verizon. In time this will reach more and more of the population and provide a further boost to both the speed and price sides of the equation.

The biggest issue for me is that these reports rely a lot on the question of timing. Where the US is today, other countries either were yesterday or will be tomorrow. We’re all heading down pretty much the same path, just at different speeds. The impatience that often accompanies the criticisms of US broadband deployments is misguided too. It usually leans on an argument about competitiveness and the ways in which broadband can transform the way we work by providing more opportunities for home working / teleworking and so on. But guess what? The measly 5Mbit/s so derided in these studies is just fine for most homeworkers and is available to almost everyone. The biggest barrier to adoption of home working is the cultural change involved, not the technology. Many companies and individual managers are still uncomfortable with the idea and suspect that a home worker is a less productive worker. That attitude needs to change more than we need more government intervention or sackcloth and ashes about the parlous state of the US broadband market, and it’s great that we finally have a study that seems to get that.

I attended Qwest’s analyst day today in New York. New CEO Ed Mueller and CFO John Richardson spoke, and were joined by fellow executives John Yost and Stephanie Comfort for Q&A at the end.

It was a brief meeting - just two hours of presentations - followed by individual one-one-ones by Ed and John for some of the attendees, including yours truly.

The company’s come a long way since I first started following it, digging its way out of massive debt and big losses and back to reasonably respectable, if not stellar, profitability, and roughly stable revenues. And the company is forecasting more of the same, with an outlook of either stable or slightly declining revenues in 2008, with slightly higher profits.

The most telling thing about the event is that Qwest is taking a markedly different tack from the other big US carriers, and possibly from most other carriers in the developed world. Telecoms has always been a growth business, which is part of what drives valuations of tech stocks, and Verizon and AT&T certainly appear to be chasing that goal. But Qwest appears, at least temporarily, to have abandoned that goal, or at least to have set it aside, in favour of a focus on margins and free cash flow.

A lot of the talk during the meeting was about partnerships, but with the main focus on two partnerships that don’t drive a lot of revenue for the company - DirecTV and Sprint. Qwest is building out a fibre to the node network in 23 markets, covering 1.5 million households, in 2008, at a cost of $300 million, but it has no plans to deliver TV services over those cables. Instead, it will focus on faster broadband speeds, which it hopes will deliver higher prices and therefore raise ARPU, providing it with a revenue lift, while relying on the DirecTV partnership for TV services. Since it only makes around a 15% commission on those DirecTV sales, compared with Verizon and AT&T’s full revenue recognition from their TV sales, this is a pretty different strategy, and unlike to change substantially, although Mueller is planning to explore services like video on demand to integrate DirecTV and Qwest capability.

On the wireless side, the company is planning to rethink its partnership with Sprint and form a new partnership (possibly with Sprint again but likely with someone else) which would provide deeper integration but also a portfolio for Qwest that would more closely mirror its competitors’. Mueller appears confident that he can get this, but given that Sprint has historically been much more aggressive about MVNO activities than the other major wireless carriers, and Verizon and AT&T have very little incentive to play ball, I’m not hopeful. It looks like Mueller may be a little naive in this respect.

For what growth Qwest is chasing, it’s going to rely on that boost in broadband ARPU, more bundling to reduce churn in access lines, and better business revenues, from gaining share and starting to see results from the Networx contract. This doesn’t seem too unreasonable, although other than adding 250 salespeople it wasn’t obvious what was really going to change. Mueller appears to believe that the investments made by his predecessor are just now becoming aligned in such a way as to provide a platform for growth, but all the execution still lies ahead, and this may be another area where his optimism is ahead of reality. All that remains to be seen, however. I wish them luck, at any rate.