Speaking Engagements & Private Workshops - Get Dean Bubley to present or chair your event

Need an experienced, provocative & influential telecoms keynote speaker, moderator/chair or workshop facilitator?
To see recent presentations, and discuss Dean Bubley's appearance at a specific event, click here

Wednesday, November 25, 2020

Interoperability is often good – but should not be mandated

Note: this post was first published via my LinkedIn Newsletter. Please subscribe (here) & also join the comment & discussion thread on LI

Context: I'm going to be spending more time on telecom/tech policy & geopolitics over the next few months, spanning UK, US, Europe & Global issues. I'll be sharing opinions & analysis on the politics of 5G & Wi-Fi, spectrum, broadband plans, supply-chain diversity & competition.

Recently, I've seen more calls for governments to demand mandatory interoperability between technology systems (or between vendors) as a regulatory tool. I think this would be a mistake - although incentivising interop can sometimes be a good move for various reasons. This is a fairly long post to explain my thinking, with particular reference to Open RAN and messaging.

Background & history

The telecoms industry has thrived on interoperability. Phone calls work from anywhere to anywhere, while handsets and other devices are tested & certified for proper functioning on standardised networks. Famously, interoperability between different “islands” of SMS led to the creation of a huge market for mobile data services, although that didn't happen overnight in many countries.

Much the same is true in the IT world as well, with everything from email standards to USB connections and Wi-Fi certification proving the point. The web and open APIs make it easier for cloud applications to work together harmoniously.

Image source: https://pixabay.com/illustrations/rings-wooden-rings-intertwined-100181/

 But not everything valuable is interoperable. It isn't the only approach. Proprietary and vertically-integrated solutions remain important too.

Many social media and communications applications have very limited touch-points with each other. The largest 4G/5G equipment companies don’t allow operator customers to mix-and-match components in their radio systems. Many IT systems remain closed, without public APIs. Consumers can’t choose to subscribe to network connectivity from MNO A, but telephony & SMS from ISP B, and exclusive content belonging to cable company C.

This isn't just a telecom or IT thing. It’s difficult to get different industrial automation systems to work together. An airline can’t buy an airframe from Boeing, but insist that it has avionics from Airbus. The same is true for cars' sub-systems and software.

Tight coupling or vertical integration between different subsystems can enable better overall efficiency, or more fluid consumer experience - but at the cost of creating "islands". Sometimes that's a problem, but sometimes it's actually an advantage.

Well-known examples of interoperability in a narrow market subset can obscure broader use of proprietary systems in a wider domain. Most voice-related applications, beyond traditional "phone calls", do not interoperate by default. You could probably connect a podcast platform to a karaoke app, home voice assistant and a critical-communications push-to-talk system.... but why would you? (This is one reason why I always take care to never treat "voice" and "telephony" synonymously).

Hybrid, competitive markets are optimal

So there is value in interoperable systems, and also in proprietary alternatives and niches. Some sectors gravitate towards openness, such as federation between different email systems. Others may create de-facto proprietary appoaches - which might risk harmful monopolies, or which may be transferred to become open standards (for instance, Adobe's PDF document format).

And even if something is based on theoretically interoperable underpinnings, it might still not interoperate in practice. Most enterprise Private 4G and 5G networks are not connected to public mobile networks, even though they use the same standards.


Interoperability can be both a positive and negative for security. Open and published interfaces can be scrutinised for vulnerabilities, and third-parties can test anything that can be attached to something else. Yet closed systems have fewer entry points – the “attack surface” may be smaller. Having a private technology for a specific purpose – from a military communications infrastructure to a multiplayer gaming network – may make commercial or strategic sense.

In many all areas of technology, we see a natural pendulum swing between openness and proprietary. From open flexibility to closed-system optimisation, and back again. Often there are multiple layers of technology, where the pendulum swings with a different cadence for each. Software-isation of many hardware products means a given system might employ multiple layers at the same time.

 Consider this (incomplete and sometimes overlapping) set of scenarios for interoperability:

  • Between products: A device needs to be able to connect to a network, using the right radio frequencies and protocols. Or an electrical plug needs to fit into a standardised socket.
  • Within products or solutions (between components): A product or service can be considered to be just a collection of sub-systems. A computer might be able to support different suppliers’ memory chips or disks, using the same sockets. A browser could support multiple ad-blockers. A telco’s virtualised network could support different vendors for certain functions.
  • Application-to-application / service-to-service: An application can link to, integrate or federate with another - for instance a reader could share this article on their Twitter feed, or mobile user can roam onto another network, or a bank can share data access with an accounting tool.
  • Data portability: Data formats can be common from one system to another, so users can own and move their "state" data and history. This could range from a porting a phone number, to moving uploaded photos from one social platform to another.

There’s also a large and diverse industry dedicated to gluing together things which are not directly interoperable – and acting as important boundaries to enforce security, charging or other functions. Session Border Controllers link different voice systems, with transcoders to translate between different codecs. Gateways link Wi-Fi or Bluetooth IoT devices to fixed or wireless broadband backhaul. Connectors enable different software platforms to work together. Mapping functions will eventually allow 5G network slicing to work across core, transport and radio domains, abstracting the complexities at the boundaries.

Added to this is the entire sphere of systems integration – the practice of connecting disparate systems and components together, to create solutions. While interoperability helps SIs in some ways, it also commoditises some of their business.

Coexistence vs. interoperation

Yet another option for non-interoperable systems is rules for how they can coexist, without damaging each other’s operation. This is seen in unlicensed or shared wireless spectrum bands, to avoid “tragedies of the commons” where interference would jam all the disparate systems. Even licensed bands can be "technology neutral".

Analogous approaches enable the safe coexistence of different types of road users on the same highway - or in the voice/video arena, technologies such as WebRTC which embed "codec negotiation" procedures into the standards.

Arguably, improving software techniques, automation, containerisation and AI will make such interworking and coexistence approaches even easier in future. Such kludginess might not please engineering purists who value “elegance”, but that’s not the way the world works – and certainly shouldn’t be how it’s regulated.

In a healthy and competitive market, customers should be able to choose between open and closed options, understanding the various trade-offs involved, yet be protected from abusive anti-competitive power.

A great example of consumer gains and "generativity" in innovation is that of the Internet itself, which works alongside walled-garden, telco or private-network alternatives to access content and applications.

Customers can have the best of both worlds - accelerated, because of the competitive tensions involved. The only risk is that of monopolies or oligopolies, which requires oversight.

Where does government & regulatory policy fit in this?

This highlights an important and central point: the role of government, and its attitude to technology standards, interoperability and openness. This topic is exemplified by various recent initiatives, ranging from enthusiasm around Open RAN for 5G in the US, UK and elsewhere, to the EU’s growing attempts to force Internet platform businesses to interoperate and enable portability of data or content, as part of its Digital Services Act.

My view is that governments should, in general, let technology markets, vendors and suppliers make their own choices.

It is reasonable that governments often want to frame regulation in ways to protect citizens from monopolists, or risks of harm such as cybersecurity. In general, competition rules are developed across industries, without specific rules about products, unless there is unfair vertical integration and cross-subsidy.

Governments can certainly choose to adopt or even incentivise interoperability for various reasons – but they should not enshrine it in laws as mandatory. If you're a believer in interventionist policies, then incentivising market changes that favour national champions, foster inward investment and increase opportunities can make sense - although others will clearly differ.

(Personally, I think major tranches of intervention and state-aid should only apply to game-changers with huge investment needs - so perhaps for carbon capture technology, or hydrogen-powered aviation).

Open RAN may be incentivised, not mandated

A particular area of focus by many in telecoms is around open radio networks. The O-RAN Alliance and the TIP OpenRan project are at the forefront, with many genuinely impressive innovations and evolutions occurring. Rakuten's deployment is proving to be a beacon - at least for greenfield networks - while others such as Vodafone are using this architectural philosophy for rural coverage improvements.

Governments are increasingly involved as well - seeing a possible way to meet voters' desires for better/cheaper coverage, while also offsetting perceived risks from concentrations of power in a few large integrated vendors. This latter issue has been pushed further into the limelight by Huawei's fall from favour in a number of countries, which then see a challenge from a smaller number of alternative providers - Nokia, Ericsson and in some cases Samsung and NEC or niche providers.

This combination of factors then gets further conflated with industrial policy goals. For instance, if a country is good at creating software but not manufacturing radios, then Open RAN is an opportunity, that might merit some form of R&D stimulus, government-funded testbeds and so on.

So I can see some arguments for incentives - but I would be very wary of a step to enshrine any specific interop requirements into law (or rules for licenses), or for large-scale subsidies or plans for government-run national infrastructure. The world has largely moved to "tech neutral" approaches in areas such as spectrum awards. In the past, governments would mandate certain technologies for certain bands - but that is now generally frowned upon.

No, message apps should not interoperate

Another classic example of undesirable "forced interoperability" is in messaging applications. I've often heard many in the telecoms industry assert that it would be much better if WhatsApp, iMessage, Telegram, Snap - and of course the mobile industry's own useless RCS standard - could interconnect. Recently, some government and lobbying groups have suggested much the same, especially in Brussels.

Yet this would instantly hobble the best and most unique features of each - how would ephemeral (disappearing) messages work on systems that keep them stored perpetually? How would an encrypted platform interoperate with a non-encrypted platform? How could an invite/accept contact system interwork with a permissive any-to-any platform? How would a phone-number identity system work with a screen-name one?

... and that's before the real unintended consequences kick in, when people realise that their LinkedIn messages now interoperate with Tinder, corporate Slack and telemedicine messaging functions.

That doesn't mean there's never a reason to interoperate between message systems. In particular, if there's an acquisition it can be useful and imporant - imagine if Zoom and Slack merged, for instance. Or a gaming platform's messaging might want users to send invitations on social media. I could see some circumstances (for business) where it might be helpful linking Twitter and LinkedIn - but also others where it would be a disaster (I'm looking at you, Sales Navigator spamming tools).

So again - interoperability should be an option. Not a default. And in this case, I see zero reasons for governments to incentivise.

Conclusion

Interoperability between technology solutions or sub-systems should be possible - but it should not be assumed as a default, nor legislated in areas with high levels of innovation. It risks creating lowest-common denominators which do not align with users' needs or behaviours. Vertical integration often brings benefits, and as long as the upsides and downsides are transparent, users can make informed trade-offs and choices.

Lock-in effects can occur in both interoperable and proprietary systems. I'll be writing more about the concept of path dependence in future.

Regulating or mandating interoperability risks various harms - not just a reduction in innovation and differentiation, but also unexpected and unintended consequences. Many cite the European standardisation of GSM 2G/3G mobile networks as a triumph - yet the US, Korea, Japan, China and others allowed a mix of GSM, CDMA and local oddities such as iDen, WiBro and PHS. No prizes for guessing which parts of the world now lead in 5G, although correlation doesn't necessarily imply causation here.

There's also a big risk from setting precedents that could lead to unintended consequences. Perhaps car manufacturers would be next in line to be forced to have open interfaces for all the electronic systems, impacting many automakers' potential revenues. Politicians need to think more broadly. As a general rule, if someone uses the obsolete term "digital" in the context of interop, they're not thinking much at all.

I've written before about the possible risks to telcos from the very "platform neutrality" concept that many have campaigned for. Do they imagine regulators wouldn't notice that many have their own ambitions to be platform providers too?

In my view, an ideal market is made up of a competitive mix of interoperable and proprietary options. As long as abuses are policed effectively, customers should be able to make their own trade-offs - and their own mistakes.



As always - please comment and discuss this. I'll participate in the discussions as far as possible. If you've found this thought-provoking, please like and share on LinkedIn, Twitter and beyond. And get in touch if I can help you with internal advisory work, or external communications or speaking / keynote needs.

Note: this post was first published via my LinkedIn Newsletter. Please subscribe (here) & also join the comment & discussion thread on LI

#5G #openran #regulation #telecom #mobile #interoperability #competition #messaging #voice #innovation


No comments: