The future of telecoms

My telecoms crystal ball

I was recently contacted by an MBA student with a list of questions about the future of telecoms. To stir up some debate, I have copied my responses below. As always, your feedback is valued – please get in touch.

What’s your view of the telecoms industry at the moment?

It is on an unsustainable path, and faces four key transformations in order to become sustainable.

1. Externally, it has to restructure in order to integrate with a world of distributed computing services. “The cloud” is not about low-cost virtualised hosting, but is really about making information a cheap raw material that powers every industrial process. Our industry’s circuit voice legacy thinking inhibits it from fully understanding its role and value in this emerging ecosystem. The scale and scope of the customer problem to be solved fundamentally changes.

You don’t want to lease a circuit for a year. Instead you create an association of local and remote resources for a minute. You don’t want to spend $10m setting up an MVNO, but rather 1¢ to establish a transient local computation resource from wholesaled assets. Amazon “gets it” for compute and storage resources, but has a centralised model that is missing the “third leg” of the distributed computing platform, which is control over the delivery network.

The telco business is slowly heading the right way with SDN and NFV, which allow for more flexible and efficient resource allocation. These are just baby steps towards “utility distributed computing on demand”.

2. Internally, it requires a shift from a craft-based approach to network design and operations, to a much stronger science-based form of engineering. This is vital if operators are to deliver a much higher quality and quantity of experiences; increase predictably of QoE and cost; raise flexibly in a rapidly changing world; and do all this at a much lower cost. Without a robust model of cause and effect you can’t engineer out the costs and failures.

These benefits require fully exploiting packet-based statistical multiplexing and its mathematical properties. Today’s networking path is a road to mathematical ruin: it is a form of statistical illiteracy. This process of “scientification” has barely begun. The very idea that the industry is has weak conceptual & technical foundations will come as a shock to many. (SDN and NFV are going to hit some painful barriers as a result of this performance engineering capability deficit.)

3. At the interface between the external and internal, telecoms needs to shift from a supply-driven industry to a demand-driven one. Today’s broadband networks are characterised by speed, not their suitability for specific customer outcomes, or their associated cost and quality trade-offs. You can’t go to the connected appliance store and be sure what you buy will work when you bring it home. Understanding and characterising demand, and building fit-for-purpose supply, is the way things need to go.

This attitudinal change has not really become mainstream yet, but the increased talk of assured services enablement points in the right direction. The industry wasted a decade with IMS, re-building a circuit infrastructure on packets, rather than working directly with the fundamental properties of packet networking.

4. Any activity that is not part of this “distributed computing infrastructure platform” picture needs to be spun off into a separate consumer digital or enterprise services unit. Several telcos have gone a long way down this path already.

If you are a pessimist, then none of these things happen, and we will see consolidation to monopoly/duopoly, nationalisation of infrastructure, high bills to users and taxpayers, stagnation of innovation, and the digital economy fails to achieve its true potential. Progress is not assured, and we have entered dark ages before in human history.

If you are an optimist, then all of these happen, and thus we are on the cusp of a boom in digital services later this decade and early in the next one. The whole of the global economy and society will re-organise around the assumed existence of cheap information, and dependable and affordable means to access it. This allows for more complex higher-order systems of value to be built, and whole new industries to emerge.

What do you think are the most interesting dynamics and changes in the telecoms industry during recent years?

I would point out the following as highlights:

  • The rise of new players controlling devices (Apple), cloud resources (Amazon), user information flows (Google), component technologies (Qualcomm, ARM) and their impact on the telecoms supply chain. The corresponding exits or declines of many established players and partners (e.g. Nokia, RIM) places the telco supplier ecosystem in a weaker negotiating position in the whole digital services supply chain. The speed of some of these changes is startling.
  • The decline in relative power of the old network equipment vendors, who are no longer able to “lead” the ICT industry as they once did. They will be gobbled up by IT companies to become networked distributed computing platforms. This is already well under way: look at Oracle or Intel’s recent acquisitions. The rise of the new vendors from China and South Korea cannot be ignored.
  • The implosion of the highest-margin voice and SMS revenues in many markets from OTT arbitrage. I started a blog in 2003 called Telepocalypse that predicted this would occur. “I told you so…”
  • The emergence of vast new markets in Asia, Africa and South America, many of which have significant growth remaining and may leap-frog Western countries. New “connected mobile societies” are emerging, with a bulge of young people eager to apply these technologies to better themselves. The telecoms centre of gravity has moved firmly south and east.
  • A shift back from mobile to fixed as the natural core of the business, after a 20 year aberration. Ever-denser mobile networks are increasingly reliant on pervasive and powerful backhaul. That changes the dynamics of the business. Having two distinct markets based on access technology is making less sense over time, although significant differences do remain.

I would accept and expect legitimate differences in focus in coming up with such a list — cablecos, regulation, media, m-commerce, etc. — as there are many other important transitions going on.

Some things have notably not happened:

  • Broadband has not yet really fulfilled its promise. 3G was in many places an economic failure, albeit a user success. Fixed access remains dominated by former incumbent PTTs and a few cablecos. True competition remains weak in most places. Competition offering different levels of network quality at varying prices is often non-existent.
  • The number of digital services being delivered to the home is still below what we might have hoped for. The level of teleworking, e-health, distance education etc. still has much growth potential.
  • Cloud services remain immature. It is not possible to buy simple, packaged services that include suitable delivery and service assurance in the way we take for granted in other (physical) service and product businesses.
  • The industry has not, in aggregate, found a way to cover its costs of capital for a very, very long time. This has to change.

How do you think that the telecoms industry will evolve in the coming years?

I am an optimist on this matter. But then again, I have to be, as what I do is act as a catalyst for change. Writing “Telepocalypse” forever would be a dull and depressing professional pastime.

I expect we will see a significant shift of the balance of power and industry structure, as second-order change is typically underestimated in magnitude, but takes longer than initially expected. Physical assets with lifespans of 30+ years will eventually become part of multi-utility corporations. Their “landowners” will rent raw physical paths and rights of way. A “utility distributed computing grid” sits on top of that. Then you have the application service providers.

There is a lot of uncertainty about what effective control (rather than ownership) of networks the cloud and device providers will take. My educated guess is that we will see traditional telcos have a reduced role – they are just “trucking companies”, whereas powerful “digital logistics” platforms will lower operator pricing power and intermediate the end user relationship. (In the light of what Apple has done, many mobile operator strategic plans from the late 90s and early 2000s in retrospect look like delusional acts of megalomania.) These cloud commerce platforms will integrate hosted e-commerce and wholesale communications, and provide complete business lifecycle services for enterprises to reach their customers and vice-versa.

What happens when most digital commerce takes place in a few “cloud industrial parks” in each region, largely controlled by non-telcos? Network virtualisation centralises resource control, but who configures those resources and matches them to demand? What happens when the customers all turn up and buy cloud services via a Salesforce or Amazon platform, just like consumers go to iTunes to get media and applications?

The level of vertical integration vs horizontal structure is also highly uncertain. What Apple did with handsets could happen to networks, as radical new technical and business models take hold. We can already see potential triggers in the lab, such as Recursive Internet Architecture (RINA) and Contention Managed (CM) networks. However, there is a vast amount of pent-up technical and political friction to resist such change. Expect the unexpected.

The tough question is always timing. I have a habit of being ten years too early in my prognostications, but some of these things I’ve been saying for a number of years now. We are probably 5+ years away from this chaotic and fragmented process of distributed computing and combined commerce & communications platform-forming becoming fully visible and mainstream by gaining a buzzword. It will take many decades to fully mature. I may be drastically wrong about the structure and timing — this is all a hunch taken from extrapolating previous industrial and industry revolutions.

What is certain is that the industry will have to may a lot more attention to what its customers are actually trying to achieve. The current model is riddled with network cost and user experience hazards, and these are getting worse, not better. The pressure for major structural change is irresistible.

How do you think that consumer behaviour has been evolving?

They take it for granted that digital connectivity and services (1) exist fairly ubiquitously, (2) work fairly reliably, and (3) are reasonably affordable. You don’t have to be claiming a pension to remember a time when none of these held true. As a result, users are embedding the assumed existence of these services into their lives. You don’t keep a paper contact book any longer; there’s no city road atlas in your bag; and you’ll set off with a vague intention to meet someone, but with no actual prior arrangements for a time and place.

This is in turn triggering deeper structural changes in social and economic behaviour. A good example might be the change in the nature of middle childhood, from a “street” activity to a “bedroom” one. This is a complex topic, and not really my area of deep expertise.

What’s your personal view on the rise of OTT services?

They are taking advantage of an arbitrage opportunity presented to them by the network operators, and who wouldn’t, given the amount of fat to slice from many incumbent telcos’ larded underbelly!

That said, it is unfortunate that we’ve created an OTT networking environment where we have the worst possible set of technical and economic characteristics:

  • OTT services work by exploiting the efficiency gains offered by packet-based statistical multiplexing compared to circuits.
  • Every application has the incentive to acquire the resources is needs to operate via aggression, not (network-mediated) collaboration.
  • When an OTT application offers a load to the network, its cost is not just the delivery of those bits, but also the opportunity cost of rival flows that must now be excluded in order for the OTT application to work.
  • As a result, we have to massively over-provision access networks to deal with the resulting contention issue. This is ineffective, so real-time applications fail a lot; and inefficient, so bulk data delivery costs a lot.

These network costs and QoE risks are not allocated to their appropriate sources, which creates an unviable business in the long run. Yet there is also a silver lining for telcos to this cloud: it makes OTT businesses ultimately self-defeating. All those rival video, voice and data flows pollute and poison the common shared statistical resource on which they depend — and refuse to pay the environmental costs. The future of the Internet probably looks like the river Thames before a “managed” sewage system was built. A great networking stink awaits us.

What do you think are the biggest barriers to entry facing OTT services?

Historically it was the complexity of acquiring, installing and configuring these services. Skype made VoIP easy; Apple made software distribution easy; WhatsApp, Line, Kakao Talk, Facebook and Viber have made it easy to find other people to message with in an OTT style; WebRTC is going to take out even more friction. What’s left are managing quality of service delivery, and differentiating yourself in a very crowded and noisy marketplace.

Do you think that OTT services are a threat to the telecoms industry or an opportunity?

The short term threat is obvious, and the situation is dire. It would be easy to use inappropriate language to describe the fate of those who have failed to anticipate this change. “Wait and see” became “woe and suffering”.

One valid response is to join in and launch your own OTT services that are differentiated through integration with your core service offer. Telefonica’s “ToGo” product is a good example. Such responses can only take you so far in shoring up the old business model. Hypervoice technology can be used to create new value propositions around old services like conference calling. There are under-exploited value-added voice services like speech to text that ought to be more commonly available.

In the long run, it’s all opportunity. The telecoms business exists in order to make other people’s applications work. The framing of “OTT” is unhelpful, in that it offers a spatial metaphor for what is really a temporal business. Telcos sell “absence of delay in the delivery of information goods at a distance “. Content and application providers can’t produce this. There is plenty of scope for new custom and partnership to make everyone in the supply chain successful.

What do you think is the main value to consumers of OTT services?

Historically it has been cheapness, but as unlimited bundles have been offered to counter the OTT effect, this has been blunted. Increasingly, OTT services offer a combination of design, functionality and — let’s face it — novelty that telcos simply cannot match. Why should you expect a business that worries about “holes and poles” with multi-decade planning horizons to have competence at digital services design which has a tempo measured in weeks and months?

How can operators provide similar or even more value than the OTT services (e.g. match value and differentiate)?

Operators aren’t in the applications business in the long run. Maybe they need to put legacy services into an “Undigital Services” unit and begin to move on to where the future action is! History is littered with examples of industries that myopically failed to see second order change, whilst continually focusing on first order change.

The President of the American Locomotive Company said in 1938: “For a century, as you know, steam has been the principal railroad motive power. It still is, and in my view, will continue to be.” Diesel engines were already in service. His company was out of the steam engine business only a decade later.

Worrying about “fighting the OTT threat” is optimising your steam engine business. The future revenues are from assured service provision as part of a sophisticated distributed computing platform. Playing with the transport and trains metaphor, the end game is electric (rails, maglev or even hyperloops). Today’s broadband and cloud are more like intermediate or precursor technologies that only hint at the “universal mass transit” distributed computing possibilities to come.

Networks themselves are just large distributed supercomputers — with internal computing resources for their own operation. Combined with smarter devices and huge data centres, what we are seeing is the dawn of an era of what I call hypercomputing. I expect most current telecoms businesses to end up missing the transition, but that doesn’t mean the future is glum. To the contrary, for those who grasp the essence of what’s required — dynamically matching information service delivery supply to demand — it is very bright indeed.

Do you think that the regulators should create any regulations related to OTT services?

In general, no. These applications a form of “speech”, and it’s not the telco’s job to decide what forms of speech are acceptable on the network, and it’s not the regulatory’s authority to manage it.

There are some exceptions. For example, there are some issues “at the boundary” where OTTs interconnect to telco services. OTTs should not collect call termination fees meant to fund physical infrastructure.

Where OTT regulation is appropriate, it is the dominion of domain-specific regulators (e.g. energy for smart grids, ministry of health for telemedicine).

Some people claim that the operators degrade the OTT quality of service on purpose, such as dropping Skype calls. Do you think that the operators should use (or be allowed to use) this strategy or do you think that there is a better strategy to deal with OTT services?

There should be a principle of absolute non-discrimination for any network that uses public rights of way or resources, such as streets or spectrum. (What you do wholly on your own property is your own business). However, that does not imply “neutrality”, since no such thing philosophically or practically exists. If a telco wants to inject jitter into its broadband service to make all VoIP fail, that’s its own (stupid) choice. Creating differentiated application-agnostic transport services is fine. Picking on individual application providers is just a mafioso business model, and such information highway outlaws must be arrested..

This is a very complex area: identifying what is a legitimate “managed” service vs an illegitimate “discriminatory” one requires further industry debate and clarification.

Do you think that consumers will continue to pay for some of the traditional services when there is a free or cheaper alternative? Why?

Yes. Telephony is not voice; SMS is not messaging; MMS is not picture sharing; IPTV isn’t video streaming. Each of these telco services are complex bundles of value, where the data transmission is only one part of that. The simplicity and ubiquity of these services means they are likely to be around for quite some time.

The advantages that telcos have had with these are, however, being eroded. In the 1970s, if you got a nuisance call, you would contact your government monopoly provider and their security department would deal with it. Today, spam calls and messaging are getting more common, and telcos don’t seem to realise that this could fatally undermine vital elements of their proposition against OTTs. Telcos have been neglectful and negligent custodians of telephony and SMS, taking the profits without re-investing in the platform. Future voice and messaging harvests look bleak as a result of selling and scoffing the seedcorn.

What kind of pricing and bundling strategies do you think that the operators should use?

Rational network pricing can be re-established using a polyservice approach, which forces application data flows to declare how time-sensitive they are. Costs are driven by peak loads, and those who cannot or will not yield should face the price.

Operators should then consider tiering their offering by quality, and not just quantity, at different price points. Today’s broadband market is obsessed with (peak) speeds, which doesn’t help you when your DSL line retrains in the middle of a call, or a 4G cell hand-off fails. Users are not homogenous in their demands or ability to pay, but we treat them as if they were.

Bundling is a tried and true approach, and offering larger bundles of OTT services — more like cable programming — could be a progressive area to explore. The economic analysis of bundling and pricing strategies is not my core expertise.

How do you think that the operators can slow down or contradict the decline of revenue of traditional services?

The decline in revenue of these services is inevitable. The question is what to replace them with.

The answer is to create new network services that allow for a much wider range of quantity, quality and cost points; and can match supply to demand at a much wider range of timescales (from microseconds to months). Networks are more like option trading platforms than pipes. Those who understand this trading space and make the right (profitable) trades are the ones who will prosper.

That is a switch in metaphors — from pipes to trading platforms. It suggests that the underlying basis on which we currently conceive of value from network-enabled service delivery is broken. We use “bandwidth” as if it were a real thing that customers demand. In reality users want applications that work (well enough) at a cost that is low enough. The difference is subtle and critical.

Wastefully creating “bandwidth” that doesn’t meet the users’ actual needs is doomed to eventual failure. Turning away from a supply-push industry towards a demand-driven one is the only possible way forward.

To keep up to date with the latest fresh thinking on telecommunication, please sign up for the Geddes newsletter

Join the Newsletter

Get your fresh thinking with free Geddes newsletter
  • This field is for validation purposes and should be left unchanged.