The Connected Britain conference in London gives a glimpse into the state of our broadband economy, which is shifting from speed to user value.
Margot James MP, the Minister for Digital — actively demonstrating digitality
Yesterday I enjoyed attending the Connected Britain conference in London, and would like to offer you some thoughts on what I heard from its first day. The event is focused on the capital-intensive physical infrastructure, and the service providers who sell that capacity as broadband ISP services. There were many C-level executive speakers, and a packed auditorium of attendees in suits.
We are at the end of a wave of broadband investment in the UK with the deployment of “superfast” networks, both by the national incumbent BT, as well as by its largest rival, Virgin Media. There are also diverse new entrants and disruptors, such as CityFiber, Hyperoptic, and Gigaclear. Until now that market structure has been relatively simple, with a mix of heavily regulated wholesale local loop access, and relatively lightly regulated competitive retail ISPs.
The focus of public policy and market activity is now beginning to evolve beyond merely enabling high-speed connectivity. If we use a water utility metaphor, the industry challenge has two aspects. Firstly, reduce the amount of “leakage” of value from the “pipes”, so as to improve affordability. Secondly, ensure that whatever is being delivered is “drinkable” and suitable for its end use.
Overall this represents a shift from superfast to “supereconomic“, focusing initially on the reduction of “leaks” and losses. These value “leaks” are due many factors, including:
- a spatial mismatch of the legacy infrastructure (copper, ducts, rights of way) to supply the new demand, with “ancient” wayleave contracts, for example;
- high friction in granting new wayleaves, negotiating street works, plus deploying to new build estates;
- low user trust and slow adoption due to security being an afterthought in service design, and public trust weakened as a result;
- spectrum that is licensed at the wrong granularity of space (whole nation) and time (long tenancy agreements), resulting in low efficiency of use (<20%) and inaccessible “dark spectrum” that cannot easily be sublet;
- artificial encumbrances of business rates on physical infrastructure that distort fixed vs wireless vs satellite investment;
- one-size-fits-all rules for rural and urban (e.g. roadworks), when these have very different needs and characteristics;
- near impossibility of accessing dark fibre from BT on sensible economic terms, de facto thwarting the emergence of more local access competition to Openreach and novel market applications;
- low levels of automation in many processes (e.g. fault isolation) due to immature engineering practises;
- and slow migration from legacy infrastructure, causing duplicate cost from failure to decommission what is no longer fit-for-purpose.
Superfast is neccesary but not sufficient
The evolution from superfast to supereconomic is a normal phase of maturation, as the initial craft-based technology deployment phase is standardised, automated and industrialised. If there is a parallel to the automotive sector, broadband is at around the 1950s. The basic product template is firmly established, but everything is built by hand, not by robots. Our motorways are still under construction, and local congestion is growing. We have to get through our chrome fins and gas guzzler phases before we enjoy a quality and economy revolution.
The fully automated “self-driving network” remains a sci-fi tale, the “flying car” of broadband. Indeed, our infrastructure is optimised around the C19th telephony hub and spoke model, and hence is unsuitable to modern requirements like dense 5G cells and cloud RAN. It took many decades for the automobile industry to transcend the limits of the horse and coaching infrastructure, and there is no reason why progress should be any quicker now.
To deliver the infrastructure ambitions of the supereconomic era, we move away from a simple duality of monopoly versus competition, and enter into a more complex hybrid coopetition approach. Where there is competition, it is between more players, and rivalry is even fiercer. However, there is a countervailing need for a “collegiate” model of cooperation, so as to share resources and drive economies of scale and scope.
You can think of supereconomic broadband as being a transition zone from superfast to superfit. An example of superfast is 4G, supereconomic is 5G’s “enhanced mobile broadband”, and superfit is the desired (if over-ambitious) model for 5G’s low-latency network “slices”.
Whereas the end state of superfit has a tightly defined and assured service definition — everything is “drinkable” — with supereconomic we are still delivering traditional unassured broadband access defined by “speed”. What changes is that the focus of supereconomic broadband is a more predictable relationship between the infrastructure being deployed and the resulting cashflow. It is an economic utility, albeit not a true technical one, since the service outcomes remain somewhat undefined and occasionally “undrinkable”.
Supereconomic today enables superfit tomorrow
British MPs continue to find that their #1 postbag issue is unacceptable broadband service, whether in terms of cost, availability, speed, reliability or stability. Making the jump from “superunfit” to “superfit” in one go is too much to ask, given the sizeable tail of users struggling with even basic Web application usage over DSL. Too many people are paying the TV license fee, but still cannot successfully use even iPlayer video streaming.
Whilst the conference presenter talk has moved to “full fiber” and 5G — where the trade show is naturally pushing new equipment sales — there is now a large “technical debt” to be dealt with. BT’s prior insistence on FTTC means we have huge amounts of fragile electronics in the streets (compared to none whatsoever with point to point fibre). This “instant legacy” infrastructure imposes considerable constraints on what kinds of technical and regulatory models are possible in the UK.
The point has come where the previously unthinkable must be contemplated. Openreach has to be allowed to flourish based on its merit — or fail based on its mistakes. Its own past choices (and present ones with DLA and DLA2) are driven as much by the desire to thwart Ofcom as to support the UK’s long-term communications needs. These policies should be allowed to come and bite Openreach in the backside; it may be “too big to succeed” in its schemes. This means DCMS and Ofcom “moving the money” so that unacceptable outcomes don’t lead to continued monopolistic rents.
This demands a regulatory shift of incentives from inputs to outcomes — both at the retail user level as well as national industrial policy. It is a transition away from maximising the quantity of enabling technology. We instead emphasise the most efficient delivery of the end user service and experience. For example, if Openreach cannot enable (and isolate) multiple concurrent service providers on its infrastructure over a single line — requiring an expensive second line — then it should get paid less for it. Metrics for continuity of service, time to repair, and stability of the connection are every bit as important (if not moreso) than the peak burst speed of the line — and the money should move to reflect this end user value reality.
We also have to decouple the service outcome from its enabling network mechanisms. If that user experience is delivered using a combination of low-orbit satellite, 5G and SD-WAN, that should not be treated as any less successful than a functionally equivalent full fibre connection. If someone finds a clever protocol way to bond all the copper in the street to deliver a fibre-like outcome, good on them! The 5G national testbed in Slough is an exemplar of doing the right thing: develop supply chain integration and instrumentation, not just raw performance.
Every city street that is dug up is a deadweight loss to society, and you can only have one major dig in a town at any time. Sometimes (but not often) the Marxists are right, and competition is genuinely wasteful (or simply unattainable). The supereconomic challenge is identifying these corner cases, which are always contextual to the local market. Every network is a LAN for some locality, and what works in Bristol may be wholly unsuitable for Barra. By cleanly separating policy from mechanism, Ofcom can help a healthy regional network diversity to emerge, without chaos or loss of scale economics.
Superfit is the utility computing end game
Taking control over the current industry value “leakage” is the same philosophy as “lean” production. We need to control variability and eliminate waste, so as to reduce defects and improve resource use. Indeed, when taken to its logical conclusion, this is the essential definition of superfit: a managed supply chain, with fungible sub-components, that delivers an assured experience outcome. In other words, a supereconomic value-for-money growth phase is the necessary stepping stone to the end game of superfit.
To achieve the superfit ideal we must align supply to demand at all timescales, not just spatial demand at long ones of network planning. That in turn forces ever more granularity of regulatory and network management in both space and time. For example, packet scheduling and QoS have to be related to application value via “quality floors”. Access to ducting goes from “last mile” to “last ¼ mile” to “last 10 yards”. Spectrum becomes dynamically traded, with a futures and options market.
The regulatory emphasis is then on standardised industry technical and economic interfaces at all layers. The policy importance of what happens between those interfaces is lessened, as long as the outcome is delivered. This has previously been successfully executed with BT and its “equivalence of inputs” (EOI) to enable diverse retail ISP competition. EOI offers a policy template for application to other related problem domains.
If successful in this transition from inputs to outcomes, then we will see earlier adoption of whole new supply chains. Novel industry sectors will emerge, such as terabit access networks for genomic data, and rural 5G for agtech innovation. The centre of the local community will shift — from the church and village green in the C19th, to the shopping centre in the C20th, and finally to local data centre(s) in the C21st.
The end state is not merely a Connected Britain, but a Computed Britain. All of these broadband access systems exist in the context of delivering distributed computing services, in their many guises (e.g. mobile edge compute, cloud, blockchain). We now know exactly what needs to be done, and roughly how to do it. As one speaker quipped, delivering the “industrial internet” future “is not rocket science – just management science”.
Epilogue — Packets for pizza
It is rare for a telecoms industry event sponsor to deliver a high-quality talk, often being rather dull and self-promoting fillers of stage time. A fine exception to the rule was Nokia’s UK & Ireland CEO, Cormac Whelan, who gave a thoughtful and well-presented view of the digital economy and its key emerging constraint, which is trust and security. For example, our present networks are very poor at isolating different users and types of use from each other.
He used Dominos Pizza as a case study of industry transformation using network technology — “as much a technology company as a food one”. By coincidence the venue was also housing an Italian foodstuff industry jamboree (with accompanying aromas).
Our ethereal information services all exist in the context of delivering something else. Being corporeal, nobody can consume broadband directly, at least not unless the transhumanist nightmare is actually delivered. Whilst our highest aspiration may be superfit, it takes our customers’ industries to offer sensual superdelicious. The proof of our broadband success is always defined by the taste of someone else’s application pudding (or pizza).
For the latest fresh thinking on telecommunications, please sign up for the free Geddes newsletter.
Read It Now
