Pages

Pages

Wednesday, December 21, 2016

2017 Predictions and Anti-Predictions

I am tempted just to repost last year's "10 Awful Tech-Industry Terms to Stop Using in 2016" (link) with just a revised date... it's all still accurate - and as I recently said in an interview, anyone still using words like "digital", "OTT" or "seamless" in the telecoms industry should be fired for gross ignorance & incompetence.

But there's such a lot of interesting stuff going on at the moment - and a lot of hype as well - that I decided it was worth doing a proper post on predictions, and also (in many ways more fun) anti-predictions. 


(Note: This is an edited/tidied-up version of my original post from December 21st)


I reckon that anti-forecasting - predicting what won't happen despite a "consensus", is massively underestimated in value in the telecom industry, because nobody (especially marketers) likes a negative story. My first anti-prediction is that, sadly, this won't change in 2017, and we'll be back here again in 12 months grudgingly acknowledging that we got things wrong again, and believed too much hype and spin.

As clients and regular readers/followers probably realise, I cover quite a broad spectrum of areas - from future voice / video / WebRTC / messaging applications and platforms, to 5G / NFV / WiFi / LPWAN networks - and also a smattering of cool new "telecom futurism" stuff such as blockchain and AI. I could try to structure this post into "buckets", but actually they all interconnect in complex ways, so I'm blending them all together deliberately - and also not separating the positive and negative predictions. 


So, in no particular order: 




AI gets everywhere. This is the big one. I generally don't go to big trade events like MWC or CES any more (not a good use of my time), but I'm willing to bet that both of them have a proliferation of AI, machine learning, deep-learning, speech/image recognition & associated analytical techniques as top themes. As well as headline use-cases like self-driving cars, I think what's going to be the real story for the telecoms industry is the operational uses in network management / planning, BSS/OSS, management of NFV & SDN, customer care, fraud management and more. For example, using pattern recognition to spot abnormal behaviour in network elements to indicate impending failure (hardware or software). What's going to be interesting is seeing who's had the foresight - or good luck - to have been collecting the right types of datasets to train the algorithms. Certainly, any new systems and services being implemented - or partnerships/deals being struck - should have a strong component of both "instrumentation" to collect data, and a strategy/team in place to analyse it properly. (See also my recent post on 5G vs AI - link)


5G IoT hype gets punctured. I'm a bit saddened at the moment by aspects of 5G. There has been a genuine - and laudable - attempt by the telecom industry to understand "verticals" and various use-cases upfront, in defining 5G. The problem has been the focus on the "sci-fi" scenarios - unsurprising as multiple industries look at each others' predictions and say "wow!". Everyone is getting distracted by the wildest visions, ignoring pragmatism, humdrum issues like legacy systems, and economic/practical bottlenecks in the process. Aided and abetted by governments, industry bodies and consultants trying to "5G-wash" everything in their promised "Digital Society" and "Internet of Everything" nonsense, and egged-on by telcos hoping for cheap spectrum and lax regulations, the 5G/IoT story has got ahead of itself. (At CES in Jan'17, Qualcomm's CEO came up with the most ludicrously hyperbolic prediction - that 5G would rival electricity in importance).

Meanwhile, in the real world, IoT at the low end is being satisfied today using anything from LoRA to WiFi, while 4G-based NB-IoT is "real soon now!" and won't hit the right price points anyway. A 5G variant will be many years away, and is unlikely to get to the $1-3 price needed for mass adoption. At the top end, it's far from clear there's enough latency-critical endpoints to justify the system-wide costs and complexity that will get added. QoS-managed, 1-millisecond latency flying robots sound great, but even if there's a million of them, they'd need a $10k/mo ARPU to offset the other 19.999 billion things that won't need all that extra network intelligence & machinery.

Add in the coverage issue - a lot of IoT will be in-building or on-site, in places that telcos have patchy presence and understanding of, and I think the industry is overselling itself. 5G - as I've written about recently (link) - is mostly going to be about fixed and mobile broadband once more.








Messaging-as-a-Platform disappoints. We've all heard the stories of WeChat embedding commerce and transactions elements. And it seems like Facebook, WhatsApp, and even the perennial no-hope RCS crew are trying to emulate it, plus add in chatbots for good measure. I think it'll fail, except maybe for occasional interactions with businesses you don't care enough about to install a proper app. Nobody is going to switch from a favourite taxi or airline loyalty app to a sub-standard experience inside a messenger. It's easier & makes more sense to put messaging (and voice/video) in the vertical app, than vice versa.


Network re-intermediation: Forget the term "end-to-end". 2017 is going to be about new companies, boxes, platforms and bits of software in the network. We're going to see more "multiplicity", with SD-WAN growing in enterprise, bonding together multiple Internet connections to supplement or replace MPLS. In mobile, we're going to get many new players offering various combinations of multi-IMS or eSIM-based platforms (buy my report! - link) to enable new IoT-SP or MVNO models (although I think they're going to stay very small and niche for the next 2 years). We'll maybe get a smattering of true multi-radio designs too - as seen in Apple's recent patent. In WiFi there's an interesting trend towards cloud platforms (eg Google WiFi for the home, or KodaCloud for small businesses), where multiple access points are intelligently controlled via a remote service that manages deployment, coverage, security and more. All this extra layering is also going to make life harder for the (late-to-market) NFV crowd, as it's going to mean that a user's data flows through multiple paths, and multiple core networks. (see this post of mine - link)


"Fake everything". We're used to hearing about fake news, and Photoshopped images. Expect 2017 to bring even worse things - in particular, fake videos, fake audio recordings, and fake IoT data. There will be a growing need to demonstrate that the images, sounds and other data are indeed genuine. Some of this can be done by "fingerprinting" in various ways, but I think we'll need better ways to demonstrate "data integrity" automatically. I'm increasingly swayed that blockchains and distributed ledgers might be part of the answer.


RCS - still dead. We're now on the 7th or 8th sequel to the zombie movie, and the producers still think they can find a blockbuster, even though everyone else just watches out of amusement at the hammy acting and cliched ending. Google's involvement via the acquisition of Jibe, and subsequent attempt to cajole RCS into being some telco/Internet alternative to Apple iMessage or "SMS updated" is a dead duck. I'm sure there will be some announcements at MWC, but I bet they don't quote any MAUs & DAUs (for *proper* use, not just as an SMS client). There's vague talk of repurposing it for MaaP, but nothing to attract developers or users. I'm seeing signs that the next attempt by the industry to force RCS into the market might be as a part of next-gen European emergency NG112 standards, as a platform for what's called "content-rich emergency calling". Given that Twitter, WhatsApp and other services are widely used to send pictures or video of emergencies, I can't see that one succeeding either.


Private Cellular is going to start to move higher up the vendor and regulatory agenda. I wrote about spectrum-sharing and IoT recently (link) but that is only part of the story. Many other factors are making enterprise or government cellular more plausible - small cells, cloud/NFV core networks, open-source elements, eSIM, wider availability of skilled people, MuLTEfire, moves to issue MNC (mobile network codes) to non-telcos, enterprise/scalable IMS platforms and so on. This isn't going to happen overnight, but the signs are coalescing - and even bits of government is noting, such as the UK National Infrastructure Commission report from December (link), which called for private networks in businesses and universities (see screenshot below).








LPWAN lift-off While some use-cases for IoT will definitely need (or at least prefer) managed spectrum and and networks, there are many others that will be happy with unlicensed and little/no cost connections, as demonstrated with WiFi and Bluetooth, and even USB. National and city networks are emerging, with some local technology / government organisations like the UK's Digital Catapult even allowing free use for low-scale developers (link). I expect we will see more integrated solutions for agriculture, city-management, non-critical transportation systems and more, where the connectivity is just baked in, perhaps without a subscription requirement for a formal "service". I view it as being a bit like electricity - we all need mains power, provided as a service, but we can also get our own batteries, generators or solar cells. (Or, at large scale like airports, even build our own power stations). NB-IoT should helpfully bring more telcos into play, but only for some sub-sectors that can bear the cost (and the wait) for a service-type model.


(Dis)unified Comms-as-a-service proliferates. The enterprise world for voice, video and messaging defies neat trends, beyond a continued shift away from old PBXs and towards the cloud. Microsoft Office 365 is clearly a huge driver, but so are SP-based UCaaS from 8x8, Vonage, RingCentral and hundreds of others. Some are homegrown systems, some based on BroadSoft or other platforms. New work stream-style offers like Cisco Spark, Unify Circuit and others are growing, whilst others are looking more at social/messaging approaches from Slack, Facebook (see my recent post - link) and numerous others. Meanwhile a host of contact centre and sales force platforms are integrating all sorts of communication functions (and bots) and many clever, well-designed conferencing solutions abound. In other words, it's a bit of a mess - partly because of mobile apps and WebRTC "democratising" some of the hard stuff, along with numerous open-source components. There are industry-specific vertical solutions like Symphony [a client I recently wrote a  paper for - link] that can claim differentiated roles as well. I don't expect 2017 to bring any more clarity - indeed, we will probably see even more "disunification" of both the vendor side and actual user behaviour. (Worth noting that Enterprise Connect is one of the bigger events that I still consider worth attending). 


Telcos have some decent opportunities in enterprise, both generally and in verticals. However, the enterprise unit needs to exercise increased autonomy from the core operations and network groups, except where clearly synergistic, eg maybe NFV-powered NaaS connectivity. It needs to be able to partner rapidly, adopt 3rd-party solutions, and not be "religious" about specific technologies or standards if the market dictates otherwise. There is very little synergy between UCaaS and IMS, for instance - as Mitel has learnt to its cost with its failed acquisition of Mavenir & subsequent spinoff. In future, expect a much greater need for vertical customisation, professional-services driven engagements, cloud partnerships (including Amazon, Google, Facebook, Salesforce et al) and a focus on solutions rather than minutes/trunks. There are some interesting opportunities for wholesalers and other intermediaries too. Fragmentation = purpose-specific solutions. 2017 will need to be the year that "unified" and "disunified" get embraced as distinct trends.



Blockchain: The telecom industry has been a bit of a laggard with blockchain and distributed-ledger technologies in 2016, but that should be remedied in 2017 as there is a fair amount going on below the surface. Various vendors and operators are looking at trials, or niche use-cases. I'll have more to say on this in Jan/Feb, but I'm expecting traction in data-integrity protection, use in telcos' vertical projects in eGovernment, some aspects of NFV and authentication/identity, as well as micropayments. (See my presentation from IIT RTC on Blockchain & Telecoms here - link)



NFV realism: 2017 is going to bring a set of realisations - similar to those in 5G - that NFV and network-slicing is not going to "fix world hunger". Grandiose projects aiming to transform telcos' overall service creation, deployment, control and billing will be seen as over-ambitious and fragile. More effort will go into smaller islands of virtualisation, whether for dedicated IoT core networks, particular services/functions (eg vEPC, vIMS), particular network elements' scalability & assorted others. Over time these islands will get glued together - messy and inelegant, but more viable than huge projects that risk collapse under their own weight. Lack of skilled staff is a major bottleneck that will only be partly fixed in 2017. Network slicing will be recast as an internal tool for multiple telco units on the same transport - not a customer-facing one where an IoT provider can get its own custom slice. It'll still get hyped at MWC though, but it's not the, er, best thing since sliced bread. (My post on slicing vs. hacking - link)



Regulation: I'll be honest - it's still anyone's guess what Trump means for the FCC. It's notable he's had a meeting of tech co's - but not a similar round table of telcos. I suspect that Net Neutrality will still have legs, and lawyers will still make most of the money in that area rather than vendors or operators. In the EU there is a battle shaping up between the Commission and BEREC the regulator's group about national vs. EU rules, especially on neutrality and spectrum, and how much is defined at each level. BEREC's implementation guidelines on Neutrality were seen as quite strong. Meanwhile, various aspects of the "Digital Single Market" plans are rumbling on - some good, some bad, and some window-dressing. The "Community WiFI" initiative is in the last category & is destined for failure or irrelevance - although it might take until 2018 for the wheels to fall off. Elsewhere, there are some trends to liberalisation in spectrum management and numbering - unlike most industries, telecom regulation tends to be a battle between two or three distinct groups of big players (old telcos, Internet, governments) which all have divergent/convergent interests. I'll be doing a lot more on policy stuff in 2017.



Universal Basic Income (UBI) hype
As well as my "day job" as a telecom/Internet analyst, I'm also involved in more general futurism and some aspects of policy. The last 6 months has seen a huge surge of interest in "technological unemployment" because of automation and AI, and the potential to use some form of UBI or negative income-tax to offset the predicated millions of lost jobs. Many researchers and authors have opined on this (I recommend Calum Chace's book - link). However, I am less pessimistic, especially on a 10-year view. I think tasks will get automated, but even where whole jobs get replaced, there is plenty of work to do, although re-skilling will be an issue. As for UBI, I think it's interesting and it's good that proper trials are going on in Finland and elsewhere. But I have serious doubts about its affordability and practicality. I also think that ageing and retirement may impact the workforce more than robots - many domains *need* automation to cover a shortfall of people. I think will governments will need to better support those impacted by AI or automation, but I'm unconvinced that the universal/unconditional aspects of UBI are the best approach. It's all very well saying people shouldn't need to work - but somebody has to pay for them not to, so it makes sense to be efficient about it. (AI should make administering more complex schemes much cheaper and easier, anyway). I think 2017 will see a continuing of UBI hype, but also some much-needed sover pragmatism. (I recently spoke on this at a London Futurists' / Transpolitica conference - link)


Random rants: Things I'd like to kill off in 2017 
- Conference apps. Just have a mobile website & email a PDF agenda. I don't want a personalised, privacy-invading "engagement" platform
- Link-hijacking and spam-blog social aggregation services like link.is, Crowdfire & paper.li. Just use "raw" links please!
- Survey / feedback spam. I'm getting deeply fed up with multiple "rate my X" messages and emails. Travel companies are some of the worst - Expedia sends up to 3 spam emails for each booking. Let's shame them online with the #surveyspam hashtag
- Automated social-media tools which send "Welcome" introduction messages on Twitter, add you to irrelevant lists and so on. Do it manually or risk getting blocked. 


I look forward to debating any or all of these with people in person at various events in 2017, or online either in the comments, on Twitter or LinkedIn. If you're looking for a keynote speaker, or internal advisory project in these areas, then please get in touch via information AT disruptive-analysis DOT com. Also, I run the Future of the Network research programme for STL Partners, so if you're interested in my detailed reports on 5G, NFV, spectrum, LPWAN and so on, please check this link for details.

No comments:

Post a Comment