The danger of a looming NBN divide

A friend of mine with decades of telco planning experience under his belt recently decided to plot the NBN POI Rollout Schedule as "premises served by month". Unsurprisingly, the result is a linear rollout, which basically means that NBN Co's fixed size workforce just keeps pegging away at the build for the next two years, but it struck me that many of us aren't talking about, the layered (inside to outside) approach necessary for reinventing basic infrastructure. 

In around two years, more than nine million premises will be served by a Point of Interconnect (POI), but that doesn't mean we can all connect on that day. There are many more steps in the process, the last of which is running a fibre up your street.

While NBN Co will have POI's and backhaul in-place for all premises in about two years, it will take until around 2020 for the Fibre Serving Areas (FSA) and the Fibre Distribution Areas (FDA) and pits-n-pipes to be rolled out. You don't start a freeway "everywhere at once", but from one or both ends. Similarly the NBN must be implemented sequentially, building out from the core to the edge with no cheap shortcuts possible.

People have complained about the NBN's lengthy timeframe, but I think we need to consider that this is an endeavour more than 10 times greater than duplicating the whole East Coast Highway, which links Melbourne, Sydney and Brisbane.

That little project has been going more than 20 years with no clear end in sight, but nobody complains and gets national media attention for it.

For the Coalition to complain and criticise NBN Co for being tardy and worse is at best unwarranted, at worst disingenuous. It plays well to an uninformed public, but is political grandstanding that will backfire. 

The trouble with tinkering

The central problem with Malcolm Turnbull's "Better Broadband: Sooner, Cheaper, More Affordably" mantra is identifying which parts can be tinkered with to improve the outcome. Will the Coalition forever abandon fibre-to-the-premises (FttP) for everyone, if not, when do they plan to rollout a pure fibre network? That's an important consideration both politically and economically.

The design, contracts and the regulatory approval will all have to be redone if the Coalition dumps the "only fibre" policy. That's a delay of 12 to 24 months right there. Unless you just sell NBN Co and remove the Customer Access Network as a service regulated by the ACCC. Nobody, not even Telstra, would be daft enough to buy a prematurely privatised NBN Co. Starting with the rate of return being deemed "non-commercial" at 350 basis points above the Reserve Banks' reference rate. That might be a great return for the public, but not nearly enough for big business.

Moving to the other facets, the backhaul and central facilities and systems are mainly completed or will be built/extended on-demand. So, no delays expected there. The POI's, the equivalent to central non-customer telephone exchanges, must be completed, no matter what technologies are employed between them and the customer. No delays there. We'll know by a mid-2013 election if NBN Co and their contractors are on-target or not. There is nothing, in time or money, to be saved by changing the ACCC agreed POI list.

The next two levels, the FSA and FDA, can't be tinkered with much either. You can have "pure fibre" or "pure FTTN/Cable" FSA's and FDA's or hybrid FSA's with some FDA's and FTTP/FTTN/cable distribution areas.

For the same reason that when you build a road system in built-up areas, the rights-of-way, once laid down, are fixed. You can add kerbs, pave or concrete roads, adding lighting, new lanes and road dividers if you've first allowed yourself the space to expand into. Creating a pure fibre NBN allows us as a nation to cheaply and easily upgrade our common asset, while pure or mixed copper/cable NBN is extremely limited in its capabilities. Out-of-the-box, it's running as fast as it ever can and that's exactly what you don't want in a high-demand, high-growth developing technology.

The quest for savings and the new divide

The big savings must then come at the street level: the pits-pipes, cable, lead-ins and fibre network termination units. An easy saving would be aerial cables with high maintenance and low-reliability, already ruled out by NBN Co in the five-year life design process.

Some money can be saved by reusing the existing, often badly degraded, copper access network, but almost all of the money sunk into a Fibre-to-the-Node (FTTN) network, that will ostensibly be replaced by pure fibre, is being thrown away. Will Turnbull allow fibre to be run for consumers in an FTTN/Cable Distribution Area wanting a connection faster than 12Mbps, increasing the rollout costs above pure fibre because of the multiple services? That's an equitable access, as well as an economic, issue. We could well end up creating a very strong geographic digital divide.

If the mantra is "Better, Cheaper", then Mr Turnbull has to lay down strict guidelines segregating the have and have-not broadband areas. This isn't a trivial matter: both FTTP and FTTN networks have economic design lives of over 30 years. A poor decision is liable to cause unrest and dissension for decades.

One has to wonder if Turnbull has a rabbit he can pull out of the hat that allows the creaking and groaning copper customer access network to be leveraged cheaply and reliably until 2050? I just can't see multiplying bandwidths over copper many fold will lead to viable FTTN improvements. The copper, in the main, won't be up to it.

If the Coalition is keen to play the "say anything beforehand and renege later" game with its NBN Policy, not only will they be a one-term government, they'll later suffer a voter backlash the like we've never seen.

This is an edited version of a blog post originally published on October 28. Steve Jenkin has spent 40 years in ICT in wide variety roles including large and small software projects, 7 years writing real-time Exchange software in a Telco and Admin, Software and Database work on PC's Unix/Open Source software and mainframes.