The latest cost-benefit analysis of various Australian broadband proposals is out. It’s part of a report from an inquiry chaired by former Victorian Treasury head Mike Vertigan.
And it says in essence that Australia’s expected growth in demand for bandwidth is big enough to make the NBN viable, but small enough to make the government’s alternative look better.
I would have expected to hear the report’s authors out there defending it, but Mike Vertigan has never been keen to put himself forward in the public debate. So today much of the media I saw has been dominated by critics, and they’ve mostly been saying that a useful cost-benefit analysis is impossible, so we should just build the NBN. Paul Budde was making the claim this morning on ABC Radio, and lesser-known experts such as Sydney Uni’s Kai Riemer have been saying the same thing.
This claim – that we can’t usefully analyse the NBN’s costs and benefits – is hooey.
We can’t do a precise cost-benefit analysis, given how much Internet use is likely to change over the next decade or two. And whatever analysis we do should be up-front about how much guesswork is involved. But cost-benefit analyses are not just helpful; they’re also inevitable. Indeed, everyone who says “we should just build it” actually is doing a cost-benefit analysis. Typically they’re just doing a really sloppy cost-benefit analysis in their head, and setting their median estimate of the benefits at, approximately, Unimaginably Huge.
And Unimaginably Huge is almost certainly an overstatement.
“We can’t begin to imagine what people could do with upload speeds on an industrial scale,” Riemer told News Limited.
But of course we can begin to imagine that. Here’s how.
Places like Japan and South Korea have had high-speed broadband in place, with high upload speeds, for years now. And as the report’s demand projections document sets out, nothing magical has happened.
At right is a graph showing broadband traffic in Australia, Japan and a few other countries. In the finest traditions of the Australian Internet, I stole it from the Vertigan report. Sorry, Mike.
You’ll notice that Japan, which has had widespread fast broadband for a decade now, is not seeing an explosion in demand. Instead, Japanese demand has been growing quite slowly. The Japanese are not alone. To quote Andrew Odlyzko, perhaps the world’s leading analyst of Internet traffic: “There have been many warnings of an impending ‘exaflood’ that would swamp networks. They have not taken place, and instead we have seen a deceleration in wireline Internet traffic growth rates.”
If high-speed broadband was ready to work some new magic, we should already be able to see it at work in Japan – and South Korea, and Hong Kong, all places where high-speed broadband has been commonplace for the better part of a decade. We should be able to its benefits in Australia, too, because like most other developed countries we have had high-speed broadband for many years – not everywhere, but certainly between CBDs, many inner-city areas, and the universities. If video-conferencing is poised to explode, it should already be happening between bandwidth-rich Collins Street lawyers and their bandwidth-rich Macquarie Street clients. As William Gibson wrote long ago: “the future is here; it is just unevenly distributed”.
So Riemer’s claim that “we can’t begin to imagine” the technological future is a half-truth at best.
As it happens, Australian traffic has been exploding recently. But it’s not because of any revolutionary new productivity-driving breakthroughs. It’s down to video consumption: ABC iView, Apple TV, high-def YouTube videos, people Chromecasting Game of Thrones, and so on. (It’s essentially the same story in Hong Kong, where per-capita demand is around 100GB/month – too big to fit on this graph.)
Malcolm Turnbull has been lampooned for claiming that broadband demand is mostly about movies and TV. But if you restrict your gaze to recent demand for raw bandwidth, he’s right.
And hardly anyone wants to say we should build the NBN so we can watch True Detectives or remastered Star Trek episodes, fun though that may be. So supporters keep talking about all the other fantastic advancements we’ll miss out on if we don’t build it.
Trouble is, most of the innovations we have come up with recently don’t use all that much bandwidth. Facebook and LinkedIn and Instagram are only medium-bandwidth even at their most intensive. Twitter and smart electricity grids are low-bandwidth. Networked games like Halo 3 use surprisingly little bandwidth too, with local hardware doing most of the work. And beyond a certain point, the speed with which you see Web pages has little to do with bandwidth; it’s mostly about server responsiveness and network latency.
Most projected e-health applications, including your latest x-rays, won’t use that much bandwidth either. Even fairly decent video-conferencing for education and medical consultations and business meetings uses perhaps 2 megabits per second, according to the demand document. To the extent that something is limiting growth in the use of such technologies, that something is generally not bandwidth.
The real policy problem with the NBN is that high-speed broadband just isn’t that much of a revolution. And to justify the cost of universal provision, it needs to be.
Addendum: Note that it is possible to believe that the NBN is sub-optimal policy without believing the hyperbole that it is a disastrous white elephant. The Vertigan report explicitly finds that the NBN would have delivered benefits of around $2 billion to the economy. That’s less than the unsubsidised and government-preferred alternatives, but it doesn’t make the NBN a boondoggle, either.
Addendum 2: A long time ago I had a bit to do with Mike Vertigan. My impression then was that he was one of the best public servants going around. Others apparently agree, including at least one who disagrees with his report’s conclusions.
Addendum 3: Joshua Gans, always interesting on broadband issues, approves of the report’s willingness-to-pay methodology while highlighting some problems with applying it to likely future broadband demand.