Our NBN is an engineering Fail

Our NBN is, from an engineering point of view, a White Elephant and is a total failure. It is a failure because it was conceived by a Unionist and a Diplomat on the back of an in-flight air-sick bag, in order to win a political “pissing contest” with the then CEO of Australia’s largest telecommunications operator. In this process good engineering design was never in sight.

Since that time in 2007, politicians have so muddied the decision making process that the excellent engineering team hired to realise the NBN dream have been cornered into developing a model and a plan to construct a White Elephant, that will be described as obsolete before it is even half way through its 30 year business case.

This is a rant, and as such I am taking liberties with details and I am summarising my arguments. You don’t want to read an unabridged version. What I want is to record here are my predictions for what will become of our NBN and, if possible, propose alternative directions that could redirect the work of the NBN Co team into a better outcome.

I must also point out that this rant is entirely my own rant, and is not paid for by anyone, and doesn’t represent the thoughts of my employer, my wife, or anyone else who gives a damn about me. However, this rant is for sale. Based on the extent of my experience, vs. those involved in completing the McKinsey NBN document (valued at $53million), I value the recommendations contained within this rant at about $5 million. This rant is copyright to me.

I am an engineer. I always write this on my landing card when I come back to Australia. I could write: entrepreneur, consultant, salesman, analyst, at various times, but I don’t. I’m proud of my engineering skills and heritage. This is an engineering view.

The Big Picture

Engineers are often accused of over simplifying the world. A few minutes watching the TV show “The Big Bang Theory” shows the prejudice meted out by the scientific community to the one engineer represented. Part of our engineering mentality is the simplification of problems to what we call “first order” variables. These are the issues, the variables, that have the ability to significantly change a situation, or a problem, and are always addressed first by an engineering team.

Some time ago, I was involved in a futurists briefing forum on the directions for broadband networking in Australia. I remember a speaker, from our CSIRO, being asked about his concerns about the growth of Australia over the next 50 years. First order stuff; things that can make, or break, this country. Not surprisingly a ubiquitous broadband network did not enter into his answer.

Australia in 2050 will have almost 10 million more inhabitants than it does today. Where are these people going to live? How will they have enough water to drink? How will we generate enough energy to provide all of us with a sustainable lifestyle? These are the kinds of big picture questions that a far sighted politician should be considering on our behalf. Guaranteeing that we all have access to 100Mb/s by 2021 is definitely a second or third order variable in the big picture, and is barely worth considering. Having ubiquitous broadband is irrelevant if we are living with permanent crippling water shortages, and don’t have sufficient energy to light our homes and businesses, and travel as we need and desire.

The population growth targeted for Australia will require us to build 20 cities with the population of Canberra within the next 50 years. Or, as a worse solution in my opinion, we will need to more than double the size of Melbourne and Sydney, making them even more unsustainable and unpleasant than they are today. Either way, we need to make significant investment into our water resources, and into our energy generation to ensure that we are not all forced to compromise our beloved Australian lifestyle.

Prediction: By 2025, the decision to invest $50 billion into the NBN will be seen to have been a squandering of taxpayers’ money, as the mandated ubiquitous services provided by NBN Co will have been made obsolete by alternatives offered to most customers by commercial operations.

Proposal: Choose to develop five NEW sustainable smart cities, each a “Lucky City”, distributed throughout the Lucky Country, one in each State nearly, with focus of ensuring water, energy, transport, hospitals, schools, port facilities, and business relevance are properly designed. Allocate a budget of $10 billion to each Lucky City. The goal for each Lucky City should be to attract a permanent population of 1 million residents by 2030. That leaves us with 20 years to develop 5 further second generation Lucky Cities, before 2050.

But, we are building the NBN. It seems that the decision to spend our money is unfortunately irreversible. So if we’re going to ignore the big picture, and concentrate on a rearrangement of the deck chairs, then we can at least try to get that done so that they’re all lined up nicely, as we plough onwards into the darkness of the unimagined and unplanned big picture future.

Technology Decisions

The engineering team behind the NBN Co have had to deal with seemingly irreconcilable issues with regard to technology choices. They have been asked to build a system that can be built today in a very cost effective manner, and yet still be viable in 30 or 50 years time as is needed to make the business case at least minimally palatable.

Some time ago, when telephones were state of the art, people had to make decisions on how to roll out the newly designed twisted pair cables. In the very early days of telephony, mainly in the United States, the cost of long runs of overhead (open air) twisted pair was significant and prohibitive, so entrepreneurial engineers invented the “party line” telephone service.

The party line shared one telephone cable amongst several telephone subscribers, allowing all of these subscribers access to a telephone, with the limited drawback that only one subscriber could use the phone at a time (others could listen in on conversations too, if they raised their receivers carefully, which contributed significantly to neighbourhood watch efforts), and additionally that the “ring tone” was used to identify who should answer an incoming call. Hindsight has shown this option to be incredibly limiting for the subscribers impacted, although it was hailed at the time. The party line may have been the first practical Time Divison Multiplex (TDM) system.

Many years later, Telecom Australia attempted to solve a similar problem, a limitation in the number of twisted pairs available in the access network, using a technology called RIM, which enabled many telephone subscribers to share one twisted pair. Whilst this technology, developed by Alcatel if I remember correctly, was hailed at the time, hindsight has shown it to be incredibly limiting for the subscribers impacted and upon the delivery of modern broadband services. I believe the RIM system was based on Pulse Code Modulation, also a kind of TDM.

Now we are considering the roll out of an expensive optical access network, which will reach ubiquity throughout our nation, and we seem to be repeating the mistakes
of the “party line” and the RIM services, through the use of a shared optical network technology called PON, and more specifically GPON. GPON uses passive optical splitters to share the TDM capacity on one fibre between up to 64 customers.

PON, or Passive Optical Network, was invented by scientists in the early 1990s as a way to reduce the cost of rolling out optical fibre access networks, by reducing the cost of the optical fibre by sharing the fibre, or equivalently its capacity. In early 1990s optical fibre was relatively new, and was quite expensive. In 2010, and over the next 30 years of the NBN Co business case, the cost of optical fibre will continue to fall, and has already become an insignificant part of the total construction cost equation.

Already today, it is believed that the cost of rolling out a “home run” optical access network, is less than 10% more than a PON access network. A home run network is built on the same architecture as today’s copper access network, and every customer gets their own fibre.

Prediction: By 2025 the decision to build PON based access networks will be questioned by customers who’s access to holographic video services, for example, is curtailed by their neighbours sharing their bandwidth, NBN Co will be forced to revisit their obsolete PON builds and re-open the streets to add more home run fibre.

Proposal: Redesign the proposed NBN Co roll out to implement a home run fibre construction plan. Centralised GPON or other technologies can then be used by Retail Service Providers, and as each customer has their own fibre, RSPs will have access to Layer 1 and can make their own technology decisions.

Footprint & Roll out Decisions

The NBN Co has been put into a trap by the political process of mandating the ubiquity and equality of services to 93% of Australians. Simply put, it is not economically viable to provide the same services to all. Nor do all want to purchase an identical level of service at this time, or at any time in the future.

So, the NBN Co has to start building now in the easiest locations, to allow the managers to achieve their Key Performance Indicator targets, which are sure to be in terms of homes passed per month, and lock in their personal bonus payments. That is, the major building works will be undertaken in cities where passing the most homes is the easiest, but unfortunately the NBN Co fibre is least needed. And simultaneously NBN Co must also build in marginal country seats where the politicians need to secure votes to remain in office. Needing to build everywhere, and all at once, will cause significant labour shortages, and will probably cause the single vendors of technology to experience shortages in supply.

Prediction: Reaching 8,000 homes passed per day will result in significant excess costs that were not included as part of the NBN Co business case (though they were was identified as a serious risk).

So, why are we overbuilding our inner-city areas with telecoms infrastructure again? Our city streets were overbuilt by Telecom Australia and Optus in the 1990s with HFC to deliver broadband, television, and voice services. Over 2 million home have access to HFC networks installed in Melbourne, Sydney, Brisbane, and Perth by one or both of these two companies. Surely, we can reuse this technology in our NBN? But no, our politicians tell us that HFC is obsolete. But, what would they know?

Prediction: Those predicting the obsolescence of Hybrid Fibre Coax technology will be retired and most probably in their graves before this eventuates.

HFC, or Hybrid Fibre Coax, is a technology based on an optimum hybrid of coaxial cable and optical fibre, blending both Frequency Division Multiplex (FDM) and TDM technologies. There is only one reason why HFC might become obsolete, and it has nothing to do with technology.

HFC is used by cable TV operators (called MSOs) in United States and throughout Europe. In both of these significant markets the MSOs provide a competitive access infrastructure, a facilities based competition against the incumbent telecom operators, and have driven prices down and service quality up throughout their markets. So, as long as these MSO companies exist, and they have an extensive embedded base of HFC networks, then HFC technology will never be obsolete. Given the political drive to maintain and enhance telecommunication facilities based competition in all of the developed world (except Australia), the death of HFC technology is therefore over rated.

Proposal: Acquire and use the existing Telstra and Optus HFC networks installed throughout our cities as a head-start on the NBN Co build out. This would allow NBN Co to: avoid re-overbuilding sensitive inner-city areas, immediately pass over 2 million or 25% of homes,  immediately provide or maintain 100Mb/s services to any and all customers within the footprint who want these services, and most importantly allow NBN Co to focus on delivering broadband to the bush.

Proposal: Direct the NBN Co to focus their roll out exclusively into regions where the minimum 12/1 wholesale service offering is currently not obtainable. Only once Australia’s digital “have nots” have been fully provided (within the original 93% mandate) should the NBN Co begin to roll out into existing broadband enabled areas. This will achieve the important mandate for providing for Australia’s digital future, whilst minimising the potential for over investing taxpayer funds and orphaning existing investments.

Vendor Strategy

As a young engineers moving from the world of the University on to the work place we are, or were, very impressionable. In our first jobs we learned the ways of our company, and if we were lucky we developed good habits and clear thought processes through being in contact with experts on a daily basis. Over time, years and even decades, these thought processes insidiously change in our minds from being “a way” to do things, to being “the way” to do things.

When a young impressionable organisation, such as NBN Co, is being lead by two people who jointly have over 4 decades of experience in the way that a particular company does things, then there is every chance that alternative thought processes, which may reach alternative results, will not even be considered, because they are not even within scope of the experience of the leaders. I believe that one only needs to look to the employment history of the CxO organisation of the NBN Co to see this effect in practice.

Following a year of vendor selection activities by the NBN Co, and in consideration of the above discussion on blinkering of the thought process, it is easy to see that the team have pursued a strategy of risk maximisation at every decision making opportunity. There are two aspects to this. 1. Geographic Focus and 2. Single Vendor. The issue of single vendor I will address in the following section.

Looking at the vendor roster for the NBN Co, one would believe that our future is intimately bound up in the technological future of Central Europe. Both of the selected major network equipment companies originate from countries that have exchanged title on the same pieces of land now for over 1,000 years. Now, we know that the employment histories of the NBN Co management team is bound up in these companies, but surely there is enough capability within the team to see the risk associated in limiting the technological input into our NBN into one very small piece of the geography of the World.

The continents of Asia and the Americas have produced a number of significant technology companies, who’s lack of presence on the NBN Co network vendor roster is conspicuous by their absence.

China and India are now the world’s most rapidly developing nations, and are home to some of the world’s best technology companies. Their engineers have studied in every major University in the World, and have returned to an environment that promotes vigorous and sustained innovation and growth. These companies promote non-European ways of thinking and try to develop their own paths towards solving today’s issues. Specifically, I believe excluding Chinese companies as suppliers to the NBN Co is possibly the biggest mistake that Australia, with our close cultural ties to China, can make.

Prediction: By 2020, technology developments from within China and India will show the selection of exclusively European network technology vendors for the NBN build as myopic and xenophobic.

Proposal: The NBN Co should revisit its vendor selection policy to ensure that the technology investment programme of vendor companies is properly considered as an indicator of how their technology legacy will develop over the full 30 years of the NBN Co business case. Preference should be given to vendors who exhibit fully alternative methods of achieving the required results to ensure that real technical diversity is enhanced.

Vendor Management

The NBN Co team have worked hard to try to pick winners from the options provided to them by vendors during the tendering processes. However, the process of picking winners is as flawed in the vendor management process as it is at the race track, with the favourite rarely coming in first place.

Before the Americanisation of Telstra, Telecom Australia always pursued a dual or multi-vendor policy for all major technology purchases. Limiting yourself to the selection of a single vendor may be expedient in the situation that you have to collect a $13 million windfall and get out of town within 3 years. But, in the case of a 30 year business plan there are many opportunities for critical path failures that can simply be avoided by the implementation of a parallel vendor supply policy. One only has to imagine what would happen to the NBN Co build schedule if there were to be another Icelandic volcano eruption halting European air travel and air freight for three months, for example.

Yet, despite the opportunity to select several, or at least alternate, providers of network technology building blocks, the NBN Co have in every case determined to select a single company as a winner.

Beyond this, relating to the above point of Technology Decision process, there has been no effort to select equivalent alternative technologies to allow for the outcome that the basic technology selection itself could have been wrong. If there had been, then we would have several vendors for these technologies.

Prediction: By 2020 one or more of the NBN Co prime sole vendors will be involved in a billion dollar price gouging scandal, and this will be documented through evidence leaked to Wikileaks or an equivalent whistle-blowing web site.

Prediction: By 2020 one or more of the NBN Co prime technology vendors will fail, and the bankrupt assets will be purchased by Chinese or Indian interests. During the vendor bankruptcy process, the NBN Co build process will be halted for 12 months, until an alternative vendor can be found to replace the bankrupt vendor’s proprietary technology.

Proposal: The NBN Co should revisit their vendor selection process to ensure that multiple vendors are selected for each technology building block at the initial stage, enabling the technology supply chain to be flexibly managed based on price, performance, and availability of product. Vendors should be selected from different geographies to ensure that force majeure does not delay the NBN Co build schedule.

2 thoughts on “Our NBN is an engineering Fail

  1. Thanks again for your detailed comments. Sticking to first order, let’s just say that 12/1 has been offered as a minimum. So, let’s first fix the fact most people don’t even have that. Make that the NBN Co prime directive, so to say. Anything beyond that should be in a second stage of investment of our tax dollars, and not commenced until the prime directive is fulfilled. To ADSL2+ and HFC and other network stuff, I’ll only say that any signal carriage technology can be engineered to produce any capacity or contention ratio required. You just adjust the engineering design rules based on the physics. As a relevant example 1GE twisted pair is commonplace, as you’d be well aware. I’m not going to enter into a fibre versus copper versus wireless argument. It is orthogonal to the discussion, and gets into who my employer is, and who my customer is, and other messy details. But, your point on the migration to fibre over a period is certainly correct, most probably because of the saving in natural resources (copper is less prevalent than silica). IMHO the natural migration should happen in due course, rather than as a step function which is the current NBN Co build plan.

  2. Pingback: Our NBN; make it better, please. | feilipu

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s