Some years ago, I was involved in the deployment and management of a national network solution for a large, well-known Australian retail chain.
It involved more than 100 sites, each of which connected to the wider network using a mix of SHDSL and ADSL connections, with a small amount of fibre. In some extreme instances, microwave connections were used where no form of DSL or fibre was available.
Many of the connections had different price points, depending on which technology was used and how much upstream bandwidth was available between exchanges and the aggregation point housed in a Melbourne data centre.
It was a complicated beast.
Some of the connections were eye-poppingly expensive, so to deliver a competitive price for the customer, the price charged for some of the connections was dropped, while others were increased.
The idea behind this was that the average price charged per connection provided enough revenue to cover the costs of providing all of the connections to the customer, and deliver a reasonable return on top.
This is a common and accepted way of delivering such a solution, and there would be thousands of similar solutions deployed around Australia and the rest of the world.
In fact, the perfect textbook example of this cross-subsidy is currently under construction in Australia, and you may have heard of it.
It's called the National Broadband Network.
The NBN will see 93 percent of Australia provided with fibre-to-the-premise (FTTP) connections, four percent with an LTE wireless solution and the final three percent will get a satellite-based service.
No matter which technology is used, your internet service provider will be charged the same amount for the basic product; $24.00 per month for the ability to provide the same 12Mbps/1Mbps speeds to end users, no matter where they are or the technology used.
The cost to NBN Co of building out each of these three technologies is, of course, quite varied.
For instance, it is spending $2 billion to build, launch and operate the two Ka-band satellites serving the final three percent of Australia; all to serve approximately the last 200,000 locations.
That is almost obscenely expensive and, in a purely commercial sense, not a viable business at $24.00 per user, per month.
Again, to build the LTE network for the wireless portion, it has spent more than $1.2 billion on contracts with Ericsson and radio spectrum requirements, to service about 500,000 locations.
Not quite as expensive as the satellite service, but still not a particularly viable business unit on its own.
So how can they do that? How can NBN Co charge the exact same price across all three technologies and still make a buck?
Just like the retail network I helped build six or seven years ago, the fibre connections that make up the vast majority of the NBN will actually cost NBN Co a lot less than $24.00 a month each customer to supply.
It would not surprise me if the actual cost per connection is less than half of that amount. This difference will be used to subsidise the cost of the much smaller number of connections using the more expensive technologies.
Remember, this is a common and accepted way of delivering such a solution the world over.
Mike Quigley said at Senate Estimates last week that not completing the network would leave the company in situation without a purpose, and without hope of privatisation. Nobody would likely come forward to buy the satellite and wireless networks; they are just not viable on their own without the subsidy provided by the fibre portion of the network.
When asked by Greens senator Scott Ludlam what would happen if NBN Co was sold after the federal election next year, Quigley stated, "you would not find a buyer".
"If you halt the rollout at that point you just cannot run a business. You do not have a business to run," he said.
Communications minister Stephen Conroy added, "essentially it would transfer back onto the budget because by definition it is no longer an investment. So the entire cost would transfer back onto the budget."
This will not change, no matter who wins the next election. The contracts are, or will be, in place and the coalition has promised to fulfill existing contracts.
Further, it was revealed at Estimates that take-up of the higher speed tiers - and therefore higher average revenues per user on the NBN - is so far well ahead of initial expectations that the fibre network may well become even more viable than first envisaged.
If take-up rates continue as they have so far, Ludlam added at estimates, "you are running a much more lucrative business than the minister was hoping you would".
Should Labor win, the NBN will carry on full steam ahead.
However, should the coalition win, they may find themselves in a situation where they won't be able to financially support the wireless and satellite networks without passing more cost on to the ISPs, and therefore end users. This would be a highly unpopular outcome in the bush.
Malcolm Turnbull talks of subsidising the rural users. This would cost billions; billions that could and would be generated by completing the fibre network.
Given that it won't be viable to sell the satellite and wireless networks, will Turnbull's proposed solution - use of fibre-to-the-node technology - generate enough revenue to support them?
It won't cost as much to build but it won't return the same revenues of the fibre network either.
They just may find themselves in a situation where they can't afford to run the existing assets, where they can't find a buyer for them, where they have to honour large existing contracts, and where they cannot make enough revenue elsewhere across the network to make a reasonable return in a reasonable amount of time.
A coalition government might just not be able to do much about it at all.
Michael Wyres is a telecommunications and network engineer with more than 16 years experience in the IT industry.