Over recent months, I have engaged with Luigi Gambardella, the outspoken head of the European Telecom Network Operators' association,
ETNO, primarily via Twitter. He is currently on a mission to persuade the ITU to change its regulatory stance on the Internet, ostensibly in the name of creating a "level playing field" or to enable a "sustainable Internet value chain". Other terms that often crop up involve applying a telecom-style model of "cascading payments" to the Internet, and enshrining some form of "sender pays" principle.
The Devil, as always, is in the detail.
My contention is that ETNO is being duplicitous, and that its proposals
are flawed, ignorant and dangerous to the Internet, telecoms and the
wider global economy. I believe that the hidden agenda is to perpetuate legacy telecom-industry business models, and to inhibit consumer choice, despite claiming the exact opposite.
My hope is that ITU delegates at the upcoming WCIT
conference in Dubai see through this organisation's transparent
lobbying, get past the hypocritical soundbites, and recognise the true implications
of what is being promoted and kick it firmly into touch. (As I write this, another related event
called the Internet Governance Forum, IGF, is ongoing in Azerbaijan, with ETNO and EU participation)
Firstly, it is important to recognise that ETNO persists in promoting a "them and us" philosophy, dividing the world between so-called "OTTs" and telcos, ignoring the 130+ Telco-OTT initiatives and other evidence to suggest that they are (or should be) all mutual peers. It apparently seeks to maintain the 100-year legacy of access/service integration. Such a polarising view, while making good headlines, is woefully inaccurate. It also propounds the very old-fashioned view that the telecoms industry is only made up of "services" based on "subscriptions", rather than applications, features and functions with a multiplicity of business models.
A good example of our exchanges is
here . It's worth going through it (use the "read next page" buttons) to get a feeling for the style of discussion. When pushed on precisely which "rules" or which "we" his referring to (eg access vs. services side of telcos' businesses), he typically fails to give coherent answers.
ETNO has been doing something of a roadshow of its opinions in an effort to drum up support. For example, a September
presentation to African WCIT participants referenced "
creating the best enabling environment to
foster commercial agreements based on Quality of Service delivery and to
avoid any regulations that can hamper the development of these
agreements."
In some of our exchanges, Gambardella has tried to imply or accuse me of representing US-based Internet interests. In return, I have called him out on his blatant obfuscation and soundbite-ridden lobbying stance, as well as various inconsistences and logical fallacies such as straw-man arguments. More recently, ETNO appears to have backed off from engaging me in Twitter debates, perhaps after realising that my clients are
actually predominantly its
own members and others from the telco/vendor establishment. (Unfortunately, I've never been able to convince Apple, Google or Facebook to pay me for anything. Lobbyists can't quite manage to get their heads around the fact that I'll happily bite the hand that feeds me, if they're making a mistake).
My professional opinion is that many of its proposals need to be ignored - and to understand why,
regulators and ministers need to understand much better how the Internet
and applications work - and more widely, how the telecoms industry will likely function in the future .
I am not averse to QoS - far from it - but based on my observations and analysis, I think that applying differential QoS to the Public Internet is an extremely harmful concept. I am a believer in
Internet Neutrality, but not overall Network Neutrality.
(I will note upfront that a few ETNO ideas - for example around privacy and data protection - are much more reasonable than its muddle-headed stance on Net Neutrality and interconnect. I strongly agree that Internet companies need to have much more stringent controls on their ability to collect, store, use & abuse personal data).
Much of ETNO's argument concerns a rather vague concept of open and unregulated "commercial agreements" between multiple parties in the Internet value chain. While in principle, it is clearly good to have more flexibility in terms of connectivity and application business models, there need to be safeguards where significant asymmetry exists. There also needs to be an awareness of the consumer benefits of the current Internet model, and ensure that those are not threatened.
What is entirely absent from ETNO's proposition is a clear
distinction between "public Internet access" and other managed Internet-like access services. As I have
argued in the past - and as Ofcom and European regulatory bodies have recognised, there is clear blue water between what users perceive as the "open Internet" (ie neutral), and other broadband services, which include differentially QoS-managed offers such as carrier VoIP, IPTV, eHealth, smart metering and so on which
do not transit the public Internet.
It is absolutely critical that ITU and WCIT enshrine that difference in global telecoms regulation, and that the current shape of the
Internet value chain, accessed via
public Internet access services is not mistakenly conflated with other broadband connectivity and services. In most countries it is still up to the individual operator whether or not to offer Internet access: it is not forced to, except in Finland where it is now a human right. The important factor here is that Internet users buy access, not end-to-end "quality".
Despite the telecom industry's heritage and vendor rhetoric, it appears that end-users are much less concerned with quality and ubiquity than has previously been assumed.
The success - and economic and social contributions - of the Internet to date suggests that this "best efforts" and net-neutral model is valuable and important to protect. That does not preclude *other* non-Internet broadband services being offered on a QoS-managed basis, and indeed there are many attempts to create the necessary platforms to do so. It is also wrong to conflate "middle mile" technologies such as CDNs with
access neutrality. It has also been possible for certain app/content providers to offer "differential quality" simply by buying more/bigger servers, or better access connections from those servers to the public Internet in their data centres. They also have differential quality through use of better codecs or application design That is
entirely different to
active discrimination by the user's own access provider
in the network.
(
Note that this is also an area where I disagree with collaborator Martin Geddes. My view is that the quality-unmanaged Internet ain't broken in general, so don't fix it. Hundreds of millions of new people are year wouldn't be buying Internet access if it was useless. If bufferbloat or other issues become critically problematic, fix them in isolation and attempt to otherwise keep the Internet structure & model as close as possible to today's. If something else radically different to the Public Internet is better, let it succeed on its own, and take over disruptively from adjacency)
The ITU and national regulators need to ensure that any new regulations do not incur opportunity costs - or disadvantage citizens and consumers - by allowing non-competitive or non-transparent broadband and Internet access markets to inhibit "Real Internet" application usage or innovation. In particular, the experimental / "freemium" model of the Internet, which allows a "million flowers to bloom" has self-evidently been successful in allowing a huge variety of new and useful applications and services to evolve. It has generated billions - perhaps trillions - in value (either revenue, consumer welfare or shareholder value).
Successful Internet companies have emerged from all parts of the world, not just the US and China. More importantly, the more forward-thinking telecom operators are finally embracing the open Internet through their various
Telco-OTT initiatives. However, some Internet-based offers compete or substitute with other services offered by telcos and access providers (eg VoIP), and have provided incentives for these to be blocked or commercially inhibited, either implicitly or explicitly.
The fear is that the ETNO proposals for ITU WCIT will be implicit inhibitors of this type of innovation in future. Certain providers may be more capable of paying for quality than others (eg startups), while operators may choose to offer enhanced quality on a discriminatory basis. It is incumbent on all regulators to do thorough due diligence on the likely impact and implications and come to their own decisions.
In my view, ETNO is clearly self-interested, because of the other business activities of its members
beyond Internet access, such as legacy telephony - and it has some highly questionable external support from consultants ATKearney, who had previously published a laughably-inaccurate and unbalanced
report on Internet economics. (
Note to ATK's lawyers: a) I laughed personally & so did others, b) happy to point out the inaccuracies for an appropriate fee).
A particularly specious strand of argument is that "investment costs" need to be shared beyond the telecoms industry - in other words, ETNO is suggesting that Internet content and application companies directly pay for "sending" data, irrespective of the fact that they already pay for access for their data centres and servers, and that users also pay access fees. This hackneyed topic is regularly trotted out by operator executives, aggrieved that their access networks are being used by their paying customers for valuable third-party services. There is a sense that they should be paid double, and extract some sort of "tax" from so-called "OTT players", despite the fact that is why customers are buying Internet access in the first place.
This argument is wrong in multiple ways, but three in particular stand out:
1) The concept of the "sender" is obsolete in modern telecoms and Internet usage. It is the wrong metaphor, rooted in a medieval view of sending parcels by donkey or messenger. If you look at the way that the web and broader Internet works, there is a complex mix of sending, fetching, requesting, distribution, delivery, subscription, caching, signalling, re-sending, broadcast, multicast, P2P and many other methods for data to get from point A to point B. Increasingly, applications and browsers "fetch" data in the background. Users themselves "send" a lot of data (eg email attachments, video uploads, browser-to-browser comms). Often, the process is automated and unpredictable. The concept of sender-pays is unfair, unmeasurable and unworkable, especially as we move to HTML5. It is one of the many reasons why
1-800 "Tollfree Data" concept doesn't work in general, except in a couple of isolated exceptions like the Kindle.
2) The argument that Internet companies must "share the investment costs" for building networks is only reasonable if those same companies have a direct involvement in determining the investment direction and priorities. At the moment, many telecoms companies invest in over-expensive and over-complex infrastructure (eg IMS, unnecessary levels of QoS, roaming platforms). If the ITU expects Google, Apple & co to somehow pay for those investments, it also needs to ensure that they are able to help direct them technically too. That means direct participation in bodies such as GSMA and 3GPP and the ITU's own standards committees, rights of veto over unnecessary equipment used solely to prop up vendors' margins and so forth. Otherwise, it will be in operators' interests to over-invest in infrastructure, knowing that it will be subsidised by Internet/content firms. It will be interesting to see if Google's work on Kansas City fibre build-out allows it to suggest changes to the more broadly-accepted economics of infrastructure build-out.
3) Internet companies have their own infrastucture costs and R&D (data centres, dark fibre, power plants etc). Applying the same logic, it would be entirely reasonable for this to be part-funded by operators whose customers "consume" those resources. "You can't use my pipes for free" translates directly to "You can't use my servers for free". I suspect that telcos would end up on the wrong side of a "balance of payments" negotiation, especially as Internet players start to explore alternative access methods such as WiFi more.
In my view some simpler rules and premises will suffice for ITU and WCIT when considering Internet governance and regulation:
- Branded "Internet access" should be as neutral as possible,
barring local legal requirements such as illegal content, or "in
extremis" management of threats such as denial-of-service or damaging
levels of P2P traffic. Any management techniques should be "de minimis"
and transparent. Covert management should be a criminal offence roughly equivalent to fraud.
- Managed or "other" broadband access can be non-neutral, as long as it is not branded as "Internet". This is already practised widely for both consumers and businesses.
There may need to be safeguards to ensure that managed access does not
reduce Internet access on shared networks to unusable performance
levels.
- Certain hybrids are possible but must be closely watched for anti-competitive behaviour. For example, some Internet-centric services may in fact be delivered through parallel non-Internet mechanisms, and prioritised / QoS-managed. For example, web video could be "on-boarded" and delivered via a telco's IPTV platform. There needs to be as great-as-possible physical and logical isolation vs. real Internet access to ensure that positive discrimination / QoS does not squeeze open-Internet connectivity unduly.
One possible compromise is to suggest that WCIT adopts the premises that:
- Basic, open, neutral-as-possible Internet access is enshrined a human right. It is left up to local regulators to determine appropriate minimum parameters of quality, performance & availability. A good recommendation would be for this to be sold (and monitored) on the basis of average achieved throughputs and latencies.
- Operators are free to offer separate non-Internet managed services using any business model, as long as they are clearly distinguished from Internet Access in terms of branding and marketing.
- Any mechanisms to optimise/accelerate Internet-based services via parallel "non-Internet access" must be made available on a totally non-discriminatory basis, in a similar fashion to other wholesale telecoms businesses such as local loop unbundling.
- Wherever possible, regulators examine the option for full structural separation of access and service provision, both for fixed and mobile operators. This would reduce the risk of anti-competitive behaviour and lead to the generation of multiple new (and fair) business models.
- Encourage lower-cost and more open/flexible telecoms infrastructure investment and standards, rather than those which perpetuate old business models and concepts of telecoms.
Overall, it is important that ITU delegates to WCIT drill more deeply into the ETNO proposals and understand better the motivation. The superficial soundbites are compelling, but they mask worrying undertones that will harm the Internet - and, indirectly, citizens and Internet users around the world. Quality-of-service is of much lower importance than "quality of experience" - and overwhelming evidence suggests that end-users are much more tolerant of variable-quality / higher-choice options than the telecom industry would prefer. It is incumbent on Telecoms Ministers, and the regulators they influence, to recognise this and enshrine user-friendly policies in future ITRs.
