The NBN Debate

It seems that every man and his dog has an opinion on the proposed National Broadband Scheme in Australia, so I guess I feel compelled to add my voice to the cacophony. Of course nobody can really speak with any sense without  a realistic cost / benefit analysis, so everything I say is non sense. Facts are always open to interpretation in a political debate . However the Federal Government is the one arguing that this  is nation building infrastructure that will enable as yet impossible to imagine benefits for all time. So it is against the idea of yet to be realised future benefits that I will limit my attention.

There is a long practiced tenet in IT architecture that dictates that you do not architect or design a system for unknown or unspecified future requirements. There is a good reason for this. In the technology industry things tend to change. Predicting IT trends is hard. It's usually easy enough to spot the reasons why particular technologies are unlikely to succeed in the long run, but much more difficult to pick the winners. Just think about Service Oriented Architecture as a good idea that failed to live up to its promise, and Twitter as a ridiculous idea that took off.

I vehemently disagree with the premise that we don't know what future applications will look like. It's a pretty sure bet that they will have digitised audio, video and data. What we don't know is what the apps will do, or how they will be used, but to a certain extent that is not particularly relevant or interesting. It's like when the road system was first built, they did not have a clue what types of vehicles might be invented over time, but it would have been reasonable to expect vehicles with wheels. You might have debated the size and weight of future vehicles, but one you started building the infrastructure you could be pretty sure that all vehicles would be designed so that they could work on the existing roads. Not the other way round.

Bandwidth is also not particularly interesting. The law of diminishing returns is very much relevant when it comes to bandwidth. For example we can do reasonably high quality voice communication in as little as 10Kb. If you double the bandwidth you don't get double the improvement in voice quality. At some point the improvement in quality becomes impossible to detect. There is a good reason why the 16 bit audio CD has remained as the standard form factor for audio, in spite of numerous attempts over the years to get consumers to upgrade to higher definition formats. At some point good enough is good enough.

As for video, we are probably close to good enough. High definition video works just fine on the current ADSL speed networks, particularly when the feed is buffered. High definition video conferencing is slightly more of a challenge because the link needs to support a full duplex symmetrical connection. But once again two or three participant  conferences are easily managed on existing copper based ADSL 2 connections.

It is unlikely that the bandwidth demands of video based applications are going to increase at an exponential rate. Compression techniques and streaming algorithms  continue to improve in line with increasing application demands. I guess you could imagine perhaps doubling bandwidth requirements with 3D video, but  at some point potential  improvements will be marginal at best. If I had to pick bandwidth which this will occur I would guess somewhere in the vicinity of 3 to 4 Mb. (And you will never need more than 640K RAM. But I digress.)

Of course it is impossible to predict technology advances in excess of a ten year timeframe. For example we could develop holographic video with massive bandwidth requirements. That is the point, we simply can't predict that far in the future, so trying to develop technology infrastructure that is expected to last for more than ten years is a delusional conceit.

As for data, well you can probably find ways to consume whatever bandwidth is available. But once again there is no great mystery running applications on networks with 100Mb bandwidth. If we want a preview of what would be possible on a high speed broadband network we only have to look at corporate LAN environments to get a hint. I can think of lots of examples that would enable businesses, but then again I don’t see businesses waiting around for  the NBN before they roll out critical applications. Of course they will use it once it is there. As long as the price is competitive. 

When it comes to actually defining potential applications the government is rather vague. A spokesman for Stephen Conroy is quoted in the Herald yesterday giving some concrete examples of the need for fixed fibre to every home.

"The NBN will open up opportunities for business to reach new markets as well as enhancing access to education and in-home health care. The NBN can also assist in keeping people in their homes longer, freeing up hospital beds and improving their quality of life by not having to move into an aged care facility when they just need monitoring. People will also be able to better monitor and manage their use of electricity and water utilities using online tools."

Once again none of these applications would appear to be particularly bandwidth intensive, or difficult to achieve with current ADSL technology. Most of the monitoring and alerting systems available today operate within less than 64Kb of traffic per managed node, and that is for monitoring extremely active systems. Monitoring household electricity usage or personal health systems would be trivial.

I guess I find it pretty ironic that at this point in history we are having a national debate about bandwidth and coverage, when we are finally getting to the point where bandwidth and coverage are no longer major limiting factors. For the last fifteen years  we have been waiting for internet connections that would allow widespread video on demand, hi definition video conferencing and cloud based applications, and now all of these services have arrived.

The greatest challenge moving forward from here is not a limitation of technology, it is our ability to innovate and develop compelling new applications and services. Sure we may get to another plateau where technology once again becomes the limitation, but for now our prerogative should be to  invest in innovation.