Last week in The New York Times, Andrew McMillen boldly claimed not just that Australia had “bungled its $36 billion high-speed internet rollout” (0) but that he could actually explain how it had managed to do so.
Now it is fair to say that I am not one of the greatest fans (1) of the extremely costly Australian government-financed and -operated nationwide fast broadband policy. The National Broadband Network (NBN) as originally conceived to deliver fiber-to-the-home (FTTH) to nearly 94 percent of households was, at the original price tag of A$43 billion (2), the most expensive infrastructure project ever embarked on by the Australian government. In only a few years, the price tag had blown out to A$72.6 billion, and the scheme was running far behind the original deployment targets. Following a change of government in 2013, a review of the project was commissioned (3), which led to downscaling to a more modest “Mixed Technology Model” (MTM), estimated to incur a net loss to the economy of only A$620 per household, as opposed to the fiber-only A$2,220. Not surprisingly, given the headline, I was expecting to see something along these lines.
Bungling or bugling?
So imagine my surprise to find McMillen quoting the former (until 2013) NBN chief executive’s claim of a ”bungle” due to resiling from the original FTTH plan! “While the National Broadband Network initially envisioned high-speed fiber connecting homes and businesses directly to the network, the Liberal-led effort compromised by connecting them with existing copper wire — basically the same technology used in the earliest days of the telephone.”
McMillen’s “evidence” of the “failure” of the MTM was a hodgepodge of anecdotal evidence, ranging from Australia’s recent ranking at 51st average fastest download speed in the latest Akamai rankings to a disgruntled Brisbane video game developer complaining that it would be faster to send a hard drive by road to Melbourne than to upload the files using current network capabilities — something it was claimed “could take several days.” Australia, it was alleged, was a first-world country whose economic development was being held back as a consequence of experiencing slower internet speeds than Thailand and Kenya. One software company manager near Brisbane claimed to be paying $1,000 per month for a high-speed (100 Mbps) connection — and that the experience of buying an internet connection in Ho Chi Minh City, Vietnam, where “despite the obvious infrastructure problem, the internet there is much, much better.”
But is the Australian internet experience really all that bad?
The problem is that McMillen has fallen into the proverbial heffalump trap: When comparing apples to oranges, don’t expect to find a banana. But failing to find a banana still grabs the headline.
Australia may not have the fastest average broadband speeds recorded by Akamai, but it is a very large and, in some places, very sparsely populated country, meaning that it costs a lot more to provision broadband networks in the first place. Also, the more widely scattered the population is, the longer on average it takes for data to travel to consumers, and the slower Akamai’s “average speed” of the connections is. This is because Akamai measures average connection speeds by the time taken to send a message from its servers to a sample of internet users in each country. So a server in Sydney sampling from a national internet-using population of 24 million, 10 percent of whom live in Western Australia, some five hours flying time away, will necessarily lead to lower average internet speeds being recorded, even though the actual infrastructure may be no different in capability to that installed in, say, South Korea. The entire population of South Korea, which is twice that of Australia’s, is concentrated in an area less than one-eighth of the size of just the state of New South Wales, where Sydney is located. One can see immediately what that does for the average time taken for messages to move about.
Furthermore, the goal of the NBN and MTM has been to connect a large proportion of the Australian population. Kenya may have “faster internet” than Australia, but International Telecommunication Union data show that Kenya has only one half the number of internet users per capita and only 1 percent of the broadband connections per capita recorded by Australia. There may be some fast connections in Kenya, but not many of them — all likely located close to Akamai’s server(s). That’s nice for some, but not for 94 percent of the population as for Australia’s subsidized scheme.
Moreover, the copper connections on which the MTM relies only in part (fiber is still being rolled out where it is economically feasible) are not “the same technology used in the earliest days of the telephone.” Modern VDSL2+ copper technology used in the MTM serves almost all the broadband connections in the United Kingdom (4), which, at 16th fastest average speeds, outpaces Australia (16.3 Mbps vs. 10.1 Mbps). Once again, local geography and demographics — which are impervious to policy intervention — create limitations in Australia not experienced elsewhere.
You don’t know how lucky you are, mate
So what of the “hard luck” anecdotes in the article? As with any large-scale infrastructure rollout, political priorities mean some people will have to wait longer than others to get the newer, faster connections. Those later in the rollout may feel aggrieved about having to face the limitations of existing infrastructure for longer. But if access to fiber is crucial to a firm’s business case, why are firms not relocating to those places where such facilities have already been made available? After all, this is what firms for whom access to railways, roads, ports, and the like is imperative do.
It may just be that for most Australians; faster internet is “nice” but not “necessary.” Their (in)actions betray their words. Those who actually need a Ferrari would pay the list price up front rather than wait for the subsidized one to arrive. Complaining in the queue while waiting to get a government-subsidized Ferrari for the price of a Toyota Corolla appears somewhat ungrateful. If not for the MTM, they would likely be waiting even longer for an NBN fiber connection and paying even more for the “privilege.” Now that MIGHT be something worth complaining about!