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

Looking for a provocative & influential keynote speaker, an experienced moderator/chair, or an effective workshop facilitator?
To discuss Dean Bubley's appearance at a specific event, contact information AT disruptive-analysis DOT com

Monday, May 18, 2015

Ad-blocking: No, mobile operators won't be blocking adverts & charging Google to restore them

The last few days have seen a bunch of articles (see here [paid], here, here)  suggesting that some mobile network operators - notably in Europe - are considering deploying ad-blocking capabilities in the network, and potentially charging companies such as Google for the right to "unblock" them via some sort of revenue-share agreement. A company called Shine has been mentioned as a possible vendor - I've spoken to the firm in the past and was unconvinced.

That said, the idea that mobile operators might block ads in their networks, or charge for the traffic they use, is not new.

In fact, I believe invented the concept myself, 4.5 years ago. (See this post  - sorry to anyone trying to patent the idea!). I wasn't entirely serious, but I could see it was a possible future.

Another thing I’ve said before is that telcos often “take 4 years to spot a good idea, 4 years to implement it & another 4 years to realise they’re too late”.

And in this case, it’s already too late. While ad-blocking might be effective for stopping in-browser ads, the technology will struggle with other formats like in-app ads (except where they are from an obvious source, or in a hybrid "webview" page in the app). It also struggles with various forms of “native” advertising embedded into content, such as video pre-roll ads or sponsored-content promotions. 

In addition, the landscape has changed massively since 2010, in terms of the way that ads are delivered, dataplan size and the large role that Google/Android and other advertising-centric companies have on the overall mobile landscape. 

There are also many possible responses and workarounds to “block the blockers” – see below. My view is that these experiments will not just fail - they could backfire spectacularly.


Regulatory & policy context

As context, there is also much more awareness – and much less tolerance – of networks “messing about” with Internet traffic in 2015 than 2010, although parallel concerns over privacy and the power of a few web players perhaps make ads “fair game” despite that.

There is also a growing war in Europe especially, against US-based Internet companies being waged by some of the more shrill “public policy” teams at major telcos. Telefonica is among those advancing the most egregious and hypocritical arguments – with its blog putting up strawman after strawman, which I sometimes try to counter in the comments.

This fight is tacitly supported by the new European Commissioners Ansip & Oettinger, who seem less capable of forming independent and coherent opinions about telecoms than their "steely" predecessor Kroes.

In a nutshell, some European telcos feel they can “get away with” harassing Google and to a lesser degree Apple and Facebook, and get air-cover from their national regulators and the European Commission. While the current trials might have the convenient excuse of “protecting users’ dataplans”, the reality is much more duplicitous – they are jealous that Google has out-innovated and out-maneouvred them, in a similar fashion to their rhetoric about “OTTs”, when they have been asleep at the communications wheel for 20 years.


Minimising the data burden on users

The point in my original post was that one of the few forms of non-neutrality that regulators and the public might tolerate is that of dealing with mobile adverts. Perhaps that traffic could be paid for by the advertisers and their channels. Advertising is widely disliked - as evidenced by the growing number of (mostly PC-based) browser users downloading ad-blocking software themselves. A recent German lawsuit by publishers was defeated, but we can expect additional action to follow if telcos try to take the same path - as well as worsened relationships between the two groups of companies.

While the morality of overall ad-blocking is dubious (it’s how useful free content is paid for), on mobile data-plans with low usage caps it may indeed be an onerous burden. Websites have a right to advertise at me, but don’t have a right to make me incur substantial extra costs for the “privilege” of being “advertised at”. Auto-playing video or audio adverts alongside static content are especially disproportionate. Various forms of pop-up or other intrusive formats are also occurring in mobile, as browsers get more capable – and there is justifiable resentment at some of these. 

Paying for ads’ data usage while roaming is especially awful, and is perhaps where the operators genuinely could offer user benefit and value by blocking them. Unsurprisingly, this will probably be the last use-case considered, given the perennial and profitable rip-off prices charged for data-roaming.

Sidenote: A compromise I’d suggest from the mobile advertising industry: ensure that ads take up no more than 10% of the total traffic needed for a given website or app when accessed over cellular networs. This would neutralise the bulk of the arguments being advanced here.

I actually believe that traffic for ads is one of the very few categories that might be viable as a source of “sponsored data” – although this will be in the form of discretionary promotions (eg bulky movie trailers downloaded for free) rather than some sort of across-the-board tollgate. I estimated the future market size of sponsored traffic for advertising in my recent report on new monetisation business models - it may reach $2bn by 2019 (see here). But the model suggested here – block ads unless the companies pay up or agree a revenue share – is the wrong one entirely.

Extortion only works against the weak.


Deploy counter-measures!

And in this case, the advertising machinery is anything but weak. The ridiculous thing about this concept is the ease with which such measures can be mitigated. I can think of 5-10 possible workarounds straight away. The most obvious group of approaches is to blend apps with content so they cannot be teased-apart, and another set involves using methods to avoid the filters in some way. There are plenty of other possibilities too.

Encryption of content is the most obvious. It is already widespread in mobile, and is growing fast – in some networks, more than 50% is encrypted. There are multiple styles, ranging from SSL built-in to HTTPS traffic, SRTP for WebRTC traffic, through to using compression and proxy servers. Some of these are still theoretically “blockable” based on IP address, but the risk of false positives increases hugely. The inclusion of Google’s SPDY technology into the HTTP2 standard has pretty much ensured this is a one-way ratchet for web traffic in future. 

Belated efforts by the telco industry such as the “Open Web Alliance” to limit crypto usage or propose “trusted proxies” for traffic management are already failing and, ironically, are undermined by “untrustworthy” actions such as this.

Google is already offering a VPN option in its new Fi service, intended to secure traffic over 3rd-party WiFi. It doesn’t take much imagination to see that this could become a ubiquitous feature for Chrome – or even Android OS – over any connection. All the operator would see would be a single stream of encrypted traffic to and from Mountain View or a local Google node – perhaps even inside its own network.

The other obvious fly in the ointment is WiFi. Most smartphone users already spend 50-90% connected to 3rd-party WiFi access in homes, offices, cafes, hotels and other places. Advertising will work normally in those places. Ignore the “seamless” carrier-WiFi hype – mobile operators’ own hotspots are a tiny fraction of the total in most countries, and will stay that way. We won’t see fixed ISPs and every WiFi operator pursuing a similar strategy of ad-blocking either – there is no rationale for it. This means that (a) Google and other advertisers will further assist users in finding WiFi, perhaps even subsidising it; and (b) Tools will be built into OS’s or 3rd-party SDKs to allow developers and advertisers to download and pre-cache advertising via WiFi, serving it up locally as-needed.

Indeed, we would likely see various new ad-related APIs, or delivery services, being rolled out in Android and iOS, plus browsers, to help work around the blockages. These developments will not have come as a surprise to many, and it is foolish to believe that countermeasures are not already being worked on. 

Of course, Fi also means that Google now has a very good view of the wholesale costs of mobile data provision, and this will impact the ridiculous "revenue share" concept still further. At absolute worse-case, it will be asked to pay fees much lower than the consumers' retail price of mobile data. There will also be no justification for "double-charging" consumers for the same data paid for by advertisers.


Actions have consequences

The mobile industry is playing with fire here. The countermeasures available to Google, Facebook and others are powerful. There is a very significant risk that rather than increasing revenues, that instead these moves will actually decrease revenues, as well as increase costs and have assorted other unintended consequences – especially more encryption, and more use of 3rd-party WiFi. 

Some of the ads that may get blocked will no doubt cause embarrassment or even legal action – I can’t imagine competition authorities being amused if TelcoA blocks ads for TelcoB, for example. There’s also a strong risk that TelcoA blocks some of its own advertising, given the typical distant and disconnected silos within a large operator. That should make for some lively boardroom discussions. The publishers vs. telcos fight won't be pretty either.

I also envision the system lacking “discrimination”. Unless there is a laborious "white list" process, it will probably block ads for charity appeals, or stop government-paid ads for voting registration or impending tax-return deadlines. That should cause much amusement in policy circles, I’m sure. There will inevitably be a host of “false positives” and “false negatives”, as you can expect companies to use the same distribution channels and “signatures” for both ads and “legitimate” content.

The biggest irony is that all this will just encourage continued advancement of native apps and OS’s vs. more-powerful browsers – the exact opposite of what many telcos would like to see happen. In-app advertising will inevitable be more block-proof than web-based versions.

It’s even possible that Google will start tiering or limiting its own services on “hostile” networks, to encourage users to churn to “friendly” ones. And in extremis you could even imagine MNOs being made to pay for subscribers’ access to certain apps or services. Most people would more happily switch telco, than switch Internet ecosystem.

My prediction is that these ad-blocking services will never see the light of day - and if they do sneak out, will cause untold amounts of pain. The idea the MNOs will get a signficant revenue-share of a business to which they add zero value is laughable.

Overall, trying to extort protection money “in case something nasty happens to your ads” is a silly and unnecessary risk. Never mind the old cliché about bringing a knife to a gunfight – the mobile industry is trying to act like a racketeer here, but is only packing a rusty spoon in its belt.


Thursday, May 07, 2015

RCS is still a zombie technology, "28 Quarters Later"

In February 2008, a number of major telcos and technology vendors announced the "Rich Communications Suite Initiative" (see here).  I first saw the details a couple of months later, at the April 2008 IMS World Forum conference in Paris.

It is now 7 years, 2 billion smartphones, and 800m WhatsApp users later.

Or to put it another way, 28 Quarters Later*.



However, unlike Danny Boyle's scary, fast-moving monsters in the 28 Days and 28 Weeks Later movies, RCS is not infected with the "Rage Virus", but is more of a traditional zombie: dead, but still shambling slowly about and trying to eat your brains. It's infected with bureaucracy, complexity and irrelevance.

To remind you: April 2008 was also a few months after the launch of the first iPhone, and a few months before the launch of the AppStore. It was also when Facebook Chat, now Messenger, was switched on in my browser for the first time - while I was waiting on the podium, to start chairing the IMS event. The world of mobile devices, apps and - above all - communications has moved on incredibly far since then.

But not for RCS.

The last 7 years have seen multiple iterations of RCS versions 1.0-5.2, RCSe as a short-lived spin-off, Joyn as the GSMA-sponsored brand, and more recently an attempt to piggy-back onto the deployment of VoLTE, as well as an attempt to reposition RCS as either "SMS 2.0" or some sort of API platform for developers.

Yet the active RCS user base is never quoted - I suspect it has fewer than 20m monthly users worldwide, and probably fewer than 5m active daily users. And by "active" usage, we should really focus on the "rich" aspect (see also below) with video or file-sharing, not just SMS-in-zombie's-clothing text. 

(Some vendors/operators are trying to replace native SMS apps on phones with RCS, to force people to "adopt" it. Most will just assume it's still SMS for text-messaging and avoid the other so-called "capabilities". Be wary of puffed-up future stats). 

The issue is that RCS has a large number of unfixable structural problems that guarantee its perpetual failure:

  • It is "engineered" not "designed". Nobody sat down to ask what human problems RCS is intended to fix. It's a random grab-bag of technical features designed to be "interoperable" rather than "be useful for a purpose". There were no psychologists, designers or behavioural specialists involved in its creation.
  • It has a bunch of expensive and complex "moving parts", using obscure and inaccessible protocols. Alan Quayle has a good run-down on its failings here
  • Nobody needs expensively-engineered network cellular "QoS" for messaging, especially when 50-90% of use will be over 3rd-party WiFi anyway.
  • It comes from an era where standardised app-level interop and federation was seen as a major benefit ("SMS only took off when networks interoperated") without recognising that 10-sec app downloads or 1-sec browser interactions now make this irrelevant. Popularity, design and purpose are keys to success - you can always interoperate later if needed, via the web.
  • It's very slow-moving, as it's standardised by a committee process and hundreds of pages of labyrinthine documentation. It can't rapidly respond to new developments - messaging as a platform, stickers, "last seen online", likes, filters, disappearing messages etc. While it is a bit more extensible today, with operators' own features, it is still hamstrung by the lowest-common denominator interop specs and unnecessary QoS and per-message charging machinery
  • RCS is often pitched as "competing against OTTs", but no reason is ever given why users should switch away from WhatsApp, LINE or WeChat, and for which use-cases. I have never seen a case-study interview of a teenager saying "I used to use SnapChat, but now all my friends have switched to Telco Messenger X"
  • There is no obvious business model, except giving it away for free and hoping it adds value to a service bundle. Given that VoLTE is already an expensive investment in a declining market, it seems unlikely that many telco CFOs are keen on throwing yet more money down the drain for zero/negative ROI.
  • Blending RCS with VoLTE overlooks the fact that VoLTE will not become "ubiquitous" for at least 10 years, and maybe never. I'm forecasting just 27% of LTE users to have VoLTE by end-2019  or in other words, less than 10% of overall mobile users. Limiting RCS to the niche of VoLTE users guarantees its irrelevance, given they're exactly the people most likely to use other apps on high-end devices.
  • There is (AFAIK) no free international interconnect model for RCS - yet an important use-case for Skype, WhatsApp, WeChat and others is chatting to friends abroad.
  • It is "any-to-any" rather than driven by buddy lists or whitelists, and so prone to spam and unsolicited marketing messages. Conversely, WhatsApp remains almost totally spam-free (helped by lack of an external spammable API, as well)
  • RCS needs an IMS, which no mobile operator really wants to deploy, unless/until they're forced into it by VoLTE. While some RCS-aaS cloud solutions exist and help manage the costs, they still require client solutions on devices
  • It's not supported by Apple and only half-heartedly by Google/Android and Microsoft, as they all have their own (better) communications products that tie into their own ecoystems (and their better understanding of user behaviour and design)
  • It's not widely supported in retail-market "unlocked" handsets, used by most of the world's majority of prepay mobile users.
  • It doesn't work very well with dual-SIM phones, or for people who have multiple SIMs and accounts.  
  • It's unclear how it will work for people who don't have data plans. It might be zero-rated, but that has its own complexities and risks. For example, are file-transfers zero'd as well? It would be ironic if RCS got pressed into service as a free way to download other messaging apps from appstores.
  • It is being pitched as a "platform" before it is a successful "product". That's not the way the technology industry works. Developers and enterprises will only sit up and listen when it has already got widespread adoption among the audience.
  • Ubiquity is earned. Whatsapp, WeChat, Facebook and others have achieved viral adoption by giving people what they want, being differentiated in some way and having them recommend it. RCS has attempted to force itself on people, with telcos assuming a sense of "network privilege and entitlement".
  • Users like fragmented communications experiences for many reasons - we see people happy with 5, 10 or more "messaging" or social apps, plus an increasing tendency to embed messaging capabilities as a feature into other apps. (Voice & video will follow suit, courtesy of WebRTC)
  • There is no relevance of RCS to the enterprise. It doesn't fit with the way businesses or their employees use mobile devices and corporate applications. (Yes, I'm baffled by the Mitel/Mavenir acquisition, too)
Perhaps the most obvious indicator of failure is the name. Nobody actually wants "richness". Users want the right tool for the job, not an overloaded Swiss Army knife made "rich" with useless gizmos. The name - much like the term "multimedia" in IMS and MMS - epitomises the muddle-headed thinking that went into its conception.  It's rooted in engineers wanting to throw in as much as possible, not designers actually considering how people might behave, and what purposes/contexts they might pursue.

It is notable that WhatsApp added voice (& features like photo-sharing) only after it was already popular for text messages, and its owners could examine user-behaviour, how they used it (and for what), and only then determined what secondary features they might want. Almost all the most successful Internet applications have started out with something simple, and then added on top of it. They are also able to roll back changes, and do large-scale A/B testing on a live audience.

By contrast, RCS has been developed as a rigid, all-singing, all-dancing application with minutely-specified details.

What the industry should have done was take SMS, and make continuous minor tweaks to the application. A better (& cheaper) SMS could have battled its rivals. Context-specific versions of SMS would have made sense, too. Trying to just swap out SMS for RCS will likely make the experience worse, and accelerate the move of users (and developers) to alternative products.

There are dozens of ways to improve the utility and value of SMS. Adding video-sharing isn't one of them. RCS is emphatically NOT the new SMS2.0, despite desperate marketeers trying to suggest it.

A couple of operators' own messaging apps allegedly have RCS as an external interface for others to connect to - but I suspect that's mostly just to tick a box and keep the GSMA's enforcers quiet. I'm not aware that Orange Libon has any active "federation" use, for example. I also think Genband's "fring alliance" attempt to white-label messaging apps to multiple telcos and then allow them to "federate" is an exercise in futility. 

RCS-based app federation is just a way to create a "coalition of the losers".

The way that RCS has been positioned and marketed has been woeful as well.

RCS has had slogans such as "It's just there, it's just works" which have been laughably false, while early deployments - including by multiple operators in some countries (eg South Korea & Spain) have faded rapidly from sight. I wrote a research report in September 2010 about its continued failure.

About 3 years ago, I heard a Metro PCS radio ad while driving in the US saying "try Joyn, the messaging service taking Europe by storm!". I assume that someone subsequently rescued the teacup suffering from bad weather.

I meet virtually nobody in the industry who thinks that RCS anything other than a joke, apart from a couple of tired-looking vendor representatives. Yet most won't say so in public, scared of pointing out the GSMA Emperor's nudity. A few bloggers occasionally put up half-baked "Will RCS finally allow telcos to beat the OTTs?" articles, but that's just as click-bait, and probably half of them hope I'll jump in personally, and start a fight in the comments. 

I see a few vendors still trying to pitch RCS as an API for developers, or adding WebRTC to it in the hope to "extend" it to the browser or other devices. Both are attempts to put lipstick on the pig zombie. The use of RCS for A2P messaging will not happen until there are sufficient users adopting it for P2P. It's doubtful anyone will want to "engage" via a medium their customers are unfamiliar with, or that they think is uncool or clunky.

In my view, the industry needs to stop messing around with RCS, and kill it loudly and clearly. It needs to die with a bang, not a whimper. Someone needs to drive a wooden stake through its heart, then we can "go to the Winchester, have a nice cold pint, and wait for all of this to blow over"

The "reset" can only come with 5G. Rather than wasting yet more time and money with the GSMA's "2020" vision and perpetuating legacy IMS, VoLTE & RCS, the industry needs to start again with a clean sheet. It needs to move towards contextualised communications, where voice/messaging/video is a feature of apps, websites and devices, not just a standalone service. It needs to decouple basic useful protocols from signalling, as in WebRTC. It needs to stop trying to define application behaviour, but leave that to the market.

Operators should pull the plug on all RCS investments now. Stop throwing good money after bad. Put the teams to work on thinking about what future communications could be - not trying to protect the obsolete "interop" models of the past, with clunky, expensive and unwanted new services. Operators needs to work individually - with vendors as and where appropriate, or with Internet/app providers if that makes more sense. They need to understand how and why their customers communicate, and predict how that will change.

The future of communications is about:

  • Context and purpose - help people achieve what they really want
  • Fragmentation & multiplicity
  • Flexibility & agility
  • Design & behaviour
  • Interop & federation only where it makes sense, not by default
  • QoS only where it makes sense, not by default
  • Differentiation
  • Multiple identity domains
  • Security, privacy & control
RCS was already dead in 2008. It's still dead now. Ignore the occasional twitches of its corpse. Instead, focus on the growing seeds of new sources of value in communications.

(As a good start, check out the upcoming Contextual Communications event I'm running with Martin Geddes, on June 15th in London. Details here)

* OK yes, it's 29 quarters since the original announcement, but 28 since I discovered the details. Who's counting.

Thursday, April 23, 2015

Google Project Fi MVNO: As expected, not a big deal. For now.

I've had a look through the details for Google's long-rumoured Project Fi MVNO. It's something I wrote a blog post trying to second-guess a few months ago, and it was finally announced yesterday (I did pretty well with my predictions - see below).

Overall - my take is that it's a mixed bag. It's got some clever features - but also some serious limitations. And it's niche. Categorically - and like Apple's SIM from last year - it's not a big deal or a strategic worry for the rest of the mobile industry, at least for now.

The most glaring problem: it's an invite-only trial, and needs a specific and expensive phone, the Nexus 6. I strongly suspect a lot will change, based on observed user behaviour and future needs. It's primarily a learning exercise for Google, and I suspect some of the things it will learn will be uncomfortable for it.

As I wrote in February: "it will likely have one or two tactical near-term wins, but will be part of a cautious, long-term experiment and fishing-trip"


The pricing & positioning

The main thing that strikes me is that it's not especially cheap. Yes, $20/month is a good headline price for US consumers who have a major-operator plan today, but $10/GB isn't really that good a deal, unless you're a mostly-WiFi user who just needs a bit of cellular data for maps and emails when you're out and about. If you're using 5GB, 10GB or more per month (as you might expect from a Nexus owner), it starts to look a bit pricey, especially compared to T-Mo US or a number of international peers. There's a good comparison with other US plans here.

Moreover, the Nexus 6 is an expensive phone - $650-700 outright, or $27-29/mo on installments. Add that into the calculation, and the plans definitely aren't quite so tempting.

The "pay for what you use" idea is nice - but not exactly new. It's basically the same way that most people outside the US pay for fixed-line telephony: fixed monthly fee, plus variable per-minute calling charges, billed afterwards. There are also a few hybrid pre/post-paid plans that look a bit like this, although the "budgeting" feature is good, as is the lack of long-term contracts and punitive overages. It's a sensible and "nice" way to price for mobile service (especially the "line rental" Basic component which has long been missing, as I discussed 5 years ago, here) - and that might nudge the other US telcos into following suit. 

The overage thing is important - it's the sort of minor-seeming detail that might change user behaviour, as it removes the "fear" factor of data usage. Overage fees are essentially "fines", and unless there's a good way to track how close you are to the limit, most people will either be cautious "law-abiders" or resentful "grumblers" who've essentially "chanced" a parking meter and got a painful ticket.

Ultimately, Google is always going to be constrained by the wholesale deals it can cut with Sprint and T-Mobile - and they're not suicidal or stupid enough to give such a bargain it undercuts their own main brand. (Indeed, Sprint reportedly has a renegotiation clause if volumes get too high). And while Google can swallow some costs if it chooses to cross-subsidise (how much is your privacy worth, per month, either to you or Google?), it only has limited wiggle-room. It also doesn't have (at the moment) the extra bells and whistles like free content, T-Mo's zero-rated music and so on.

Roaming is a different story. Even capped at 256kbit/s, $10/GB for 120 countries is a great deal. One likely unintended outcome is that some regular travellers will use this as a second "roaming phone" as $10/GB and 20c roaming calls are pretty cheap in many places in the world, and less hassle than getting local SIM cards. 1GB is enough for most travellers for 3-7 days, if they use WiFi at their hotel & avoid video. They could even tether an iPhone to it - although a $650 Nexus 6 is an expensive modem). 

If someone gave me a global 3G-speed SIM for a $100 personal hotspot at $10/GB, I'd have it tomorrow, even with a monthly fee. Basically, it translates to $10/week for mobile maps, IM, social and email, plus some local tourist guides and airline check-in. As long as you're somewhere with plenty of WiFi for heavy lifting downloads (maybe offline maps, or swipe in most of the local Google Map to the cache), it's a bargain.

So that's the pricing - interesting, cheap for some people, needs an expensive phone, user-friendly, but not that amazing unless you're a frequent traveller.


What about the technology and the services?

Most notable is what Google calls the "Project Fi SIM card", which supports multiple cellular networks. It sounds like a standard multi-IMSI SIM, which is different to the Apple SIM, which is a programmable SIM. Neither are "Soft SIMs". 

There aren't too many multi-IMSI operators, but they're certainly not unknown (eg Truphone). There's no mention of choosing operators from the home-screen like the Apple iPad's approach - you get what you're given. And although it can switch from T-Mo to Sprint, it doesn't sound like that's in realtime, per-connection - the FAQ promises not to kill your battery, so I guess it just checks periodically, or maybe uses Google Now to plot your normal journeys and switch at the appropriate time. I'm curious to know what the switch-over timelag is - I'd guess it's probably noticeable if you're mid-call or mid-data session.

There is one similarity with the Apple SIM though - a lack of reliance on the underlying phone number associated with the phone SIM/IMSI. Google Voice has its own number range in the US, and people can port-in their existing number, while Apple's iPad+SIM doesn't really need a phone number at all.

The WiFi implementation is a bit opaque. The website uses my trigger-word "seamless" which makes me have doubts straight from the offset. It's not clear exactly which hotspots go to make up its curated selection. Obviously it will have its own locations such as US branches of Starbucks, but beyond that it's vague at present. Is it an in-house database, or a third party's? It talks of "over a million" hotspots but that's small beans compared to DeviceScape (20m+) and Fon (15m). I'm wondering if it's done a deal with one of the US cable operators with a decent WiFi footprint. 

The problem with a lot of these "free, open" WiFi deals is that they're often not in the places everyone wants access. As I wrote the other day, many locations have free WiFi, but use a splash/log-on page for advertising, registration or company-specific T's & C's. Would the guest-access WiFi in a hotel, corporate office or ice-cream-sponsored bus just allow Google users to bypass the front page? I doubt it, unless Google is paying for the privilege.

All that said, there is no charge for WiFi use (thankfully, or else it really would be a dud) and the auto-on VPN is a really good idea that ought to be copied by everyone. I'm curious how much of the cellular traffic is encrypted too.

The voice/text/Hangouts thing is also a bit opaque at the moment. As Google Voice "owns" the number, it can theoretically do a bunch of cool things in the cloud, although for now it seems to just be multi-device / WiFi-access that's been enabled. There's no mention of cool contextual features or inbound call-management or developer APIs.

It's unclear when it uses the native telephony function on the device, and when it's "OTT-style" VoIP. It's also not obvious if it uses VoLTE or 3GPP WiFi-calling at all, based on the underlying T-Mo/Sprint platforms - although clearly it will need to have the mobile operator's 2G/3G circuit-calling engaged when it falls off 4G coverage or is roaming, anyway. (Sprint hasn't launched VoLTE yet anyway, and T-Mo's isn't available everywhere in the US, I believe).

In fact, I'm not aware of any MVNOs worldwide that are using VoLTE, even if their host networks have switched it on. In general, IMS and MVNOs don't play very well together - I remember once asking someone from 3GPP about this, and it was clearly the first time anyone had ever mentioned them to him in the same sentence. A bit more thought has gone in since then, but it's after-thought. Given that this would also need to integrate the Google Voice core network [which I don't think is IMS-based] with 2 different MVNO hosts, I doubt it.

I suspect that the Nexus software uses Internet VoIP where it can, and cellular telephony (circuit) where it has to, without this being apparent to the user. The "seamless handover" is promised when moving from WiFi to LTE - but the Fi website doesn't mention LTE to WiFi, nor WiFi to 2G/3G. Given that there's probably a timer so the phone doesn't annoyingly connect to WiFi every time it gets a fleeting glimpse for a moment, I suspect that 4G-to-WiFi isn't supported.

The fact that Google Hangouts is used for text & talk on secondary devices (with your primary Google login) means it is clearly VoIP some of the time. There's no mention of video yet. There's also no explicit mention of WebRTC yet, but it wouldn't surprise me if some of the versions using Chrome are exploiting it, although I doubt that the Hangouts plug-in for Firefox, IE & Safari does. Obviously, RCS will be irrelevant.

To be honest though, there's no real massive innovation I see around voice and messaging here. It's still phone calls and SMS, or near equivalents. One number ringing multiple devices, or over WiFi, is hardly headline material - it's been around for a decade.

I haven't had a chance to think through the privacy implications yet, but clearly data and analytics is going to be a big part of this. I'll leave that to others to comment on.

A lot of this is near-impossible to internationalise. Google Voice doesn't have port-able number ranges, and the interconnection regimes may work against its pricing. I also don't think it would be able to get wholesale rates low enough for $10/GB in some countries - if MVNOs are allowed legally at all (often, they aren't). VoIP isn't allowed in many places too, plus there's obviously the privacy angle to consider.


So, how were my predictions?

I did a post back in February where I had a go at predicting the Google MVNO features. (Read it here). I've had a lot of discussion about both that post, and an earlier one about the Apple SIM. On a couple of recent presentation tours I've done with investment analysts & visiting their fund-manager clients, both topics have been of great interest.

Looking back, I was pretty accurate, it seems. My big "miss" was that it's a smartphone plan, not a tablet one like the Apple iPad SIM. But beyond that, I was pretty spot-on:


  • Yes, it's got a SIM card, which I'd wondered about. I got the multi-IMSI bit right, and it doesn't seem that it's doing per-second realtime hopping from one network to the other
  • No, there's no retail network - for now it's online & invitation-driven
  • It's not "WiFi first", it's "WiFi as well" - and yes, the curated hotspot network is small.
  • Yes, it's small-scale and "fishing expedition", not a large, mainstream plan aimed directly at the big US carriers.
  • Yes, it's got innovative pricing for data, although I didn't quite get the right format
  • And no it's not SIM-only for all unlocked phones
  • Nothing fancy with voice/SMS beyond Hangouts as an extension
  • It's Neutral, as far as we can tell, with no obvious policy-management complexities
  • No balloons or drones are being harmed in the making of this MVNO

Overall... worth watching, but nothing earth-shattering

In a nutshell, like the Apple SIM, Google's Project Fi is not a big deal. For now. It will appeal to a few niches of users, has a couple of nice features, and is a cautious initial step that might have a longer-term trajectory. It likely won't work easily outside the US, either.

Like Apple, I suspect Google has a long-term view on what it would like to see around the 5G launch timeframe of 2020. It might nudge AT&T and Verizon on their pricing approach for data, much as Google Fiber nudges on FTTH. It gives Google great insight into real customer behaviour, that should help both new products and regulatory arguments. 

But is it changing the landscape? Not really. And outside the US, not at all.


Notes

Disruptive Analysis is a research firm and consultancy that specialises in assessing technology and business model shifts in telecoms, especially mobile. As well as reports on Mobile Data and WebRTC (see links at the top) it regularly presents on upcoming disruptions such as Google/Apple, the role of WiFi, definition of 5G, and the impact of Peak Telephony and the Future of Voice. For more information, contact information AT disruptive-analysis DOT com.

See also this presentation covering a range of pertinent trends and disruptions.





Tuesday, April 21, 2015

Venue-owners don't want "seamless" WiFi

One of the regular myths put forward by the cellular industry is that WiFi is a natural, integrated part of the mobile world, and that we will all soon be "seamlessly" connecting to the "best" network in any specific location.

It's hard to tell if this is arrogance or simply ignorance.

As a concept, it positions the cellular industry as the most important and influential stakeholder in WiFi, especially when it comes to smartphones. This is very much untrue.

Even leaving aside the better fit of WiFi with fixed/cable telcos than mobile, one only needs look at the growing array of locations deploying WiFi for their guests / visitors / customers. Often, for all its possible awkwardness, the "seam" (eg a splash page or logon screen) is an important property. It used to capture user data, display ads or information, or otherwise "engage" with people who would otherwise have little direct interaction with that location or brand.

They have no interest in restricting access to people using a specific mobile operator, nor phones with particular technologies, or operator-customised connection manager apps. Many may have particular terms and conditions that they wish users to agree to, before providing access. 

So airlines and airports provide access to frequent flyers, or to a page with departure times/gates and retailers' details. Some locations force users to watch a video before getting access. Some link into loyalty schemes, or collect email addresses, or ask for a social-network login or "like". Cafes and other locations also obviously need to service non-cellular users connecting laptops - and probably don't want every phone auto-connecting as they'll suffer congestion. Often, WiFi is intended for private work or personal web-access, not "offload" from cellular.

As an example, belowis a picture of a London bus, painted in the colours of some new flavours of Magnum ice-cream. It also has a logo for free WiFi. According to this article in the Grocer marketing magazine, it is part of an integrated advertising strategy. "The buses will use free WiFi to encourage consumers to view branded content recommending things to do in the capital depending on whether they are feeling playful or sophisticated".
 

 
While it's possible that the in-bus system is provided by a telecom firm to Transport for London, and presumably it uses LTE backhaul, that is totally separate from the purpose of the WiFi access given to consumers. It's for advertising, not offload. And presumably TfL is looking to monetise it as part of package, along with the garish black/pink stickers on the bus.

The same is true of other venues. They want to offer WiFi as a differentiator, or as a visible value-add. If it fades into the background, it becomes neither. In addition, as a user there may be very good reasons to want to use a different WiFi provider than your cellular operator - especially if it routes traffic through a core network which imposes policies you wish to avoid (eg blocking VoIP, or applying charges for certain content). There might be times you want frictionless access (eg one-click), but the problem with seamless access is that it may come with price, usability, privacy or security limitations or compromises. 


The phrase "always best connected" is duplicitous - it generally means "best" from the point of view of the operator, not the user, advertiser, venue-owner, device-vendor or application provider. Although there might be some workable use-cases of ANDSF, PassPoint, Hotspot 2.0, I-WLAN and assorted other automated WiFi connection standards, they need very careful assessment as often they will be more harmful than helpful.

I think there is scope for regulators to look at "WiFi Neutrality" laws, preventing people from being forced (or even differentially persuaded) towards specific operators' preferred networks - especially if those networks are also non-neutral in terms of policy management. Also, independent WiFi is an important competitive component in consolidating telecom markets - governments should encourage its role as a de facto method for keeping mobile data charges at reasonable levels. (See this post on partial competition) 

My view is that the only people who want to instigate "seamless" WiFi are those lacking the imagination or commercial capability to monetise the "seam". And increasingly, users will have multiple ways to get online with WiFi at any given location, and may wish to choose based on circumstance and brand affiliation - or perhaps the chance of a free ice-cream.

Monday, April 20, 2015

Net Neutrality vs. Not: Four Unexplained Paradoxes

In all the media and lobbying furore about Net Neutrality, most of the discussion is about either political, ideological or academic angles. We regularly hear about the anti-capitalist risks of "regulating the Internet", the importance of Neutrality as a representation of free speech, or the theoretical efficiencies of two-sided markets and so on.

Frankly, all of those arguments will go on ad infinitum, because there is no objectively "right" answer to any of them. Unless regulators do multi-year massive field trials of neutral / non-neutral rules, there won't be any real data on actual consumer welfare, rates of innovation and so on. It will also just be rhetoric and theory.

But there is a particular subset of claims and assertions in the telecom/Internet industry which are mutually contradictory, where the industry itself has created a paradox. Regulators should challenge market participants to clarify the dichotomies, as a more rigorous thought process might lead to a few problems simply disappearing.

I've spotted four of these self-contradictions:

Paradox 1:
  • "There is a data tsunami, leading to congested networks and ultimately a 'capacity crunch' & so we need more spectrum"
  • "We want to zero-rate some mobile apps & give the data away for free"

Paradox 2:
  • "We can expose QoS capabilities to 3rd party app & content companies such as VoIP & videoconferencing players, monetising QoS capabilities in the network"
  • "People will use and value VoLTE & ViLTE, because it's the only QoS-managed voice and video service. The rest are best-efforts only"

Paradox 3:
  • "About 10-15% of cell sites face congestion"
  • "We're going to make a lot of money from paid prioritisation"


 Paradox 4:
  • "We're aiming for customers to move to larger data-plans, or incur overage fees"
  • "We'll sell sponsored data to content/app companies [at wholesale prices]"

Now, in all of these there's a little wiggle-room, despite the self-contradiction. 

It might be that operators carefully craft their zero-rating plans, so they only generate more (free) data traffic in uncongested areas, or at uncongested times. Or perhaps more cynically, creating more near-term congestion is a tactical positive, as it helps sway spectrum policy with longer-term benefits.

Perhaps operators will expose commercial QoS APIs only to those companies that don't offer competing VoIP/video communications services, or will price it on a discriminatory basis, so that VoLTE has a cheaper cost basis. Or maybe it will be "semi-skimmed QoS" available to app developers, with "full-fat QoS" only available for in-house services.

Maybe paid-prioritisation (let's assume it works; many think it doesn't) will be offered at such a small uplift of data transmission price, that content companies will be willing to buy it like an insurance policy, covering 100% of locations/times/users, even though they know that the risks are only real for 10% of locations, 10% of the time, for the 20% of users who aren't on WiFi anyway. Or perhaps they'll price it at a huge premium to the baseline cost, so that app devs will just use it in a real emergency situation.

Maybe telcos with sponsored data platforms can develop a mechanism that encourages use just for new mobile apps and content, rather than replacing retail data plans paid by users, with discounted wholesale traffic bought in bulk by content firms.

In my view, a lot of the supposed "use cases" and "business models" that opponents of Net Neutrality put up are straw-men. They simply don't work - either technically or commercially. Many of them will actually lead to lower revenues, or have unintended market and regulatory consequences. 


(If selling QoS to third parties is ever to become feasible, it will probably need laws for non-discrimination, transparent pricing & maybe even structural separation)

We already see the failure of AT&T's Sponsored Data programme to make any meaningful impact. We also see a lack of any app or content co's clearly asking for prioritisation to be sold to them - nobody actually wants it, at least on mobile networks. And I'm willing to bet that the first time a regulator asks a telco's execs why they need more spectrum, when they're generating excess traffic with free zero-rated usage, it will cease to be a big deal.

That's not to say that other Neutrality problems aren't real - there clearly is blocking of certain apps, or deliberate degradation/throttling of others. But many of the "hot topics" for activists and lobbyists are simply worthless, unworkable ideas. 

Regulators should focus on these apparent paradoxes - as clarifying the answers (and making all involved think through them clearly) - will make most of the points of contention simply evaporate.

These issues and many others are addressed in Disruptive Analysis research on new mobile data monetisation & "non-neutral" strategies. Click here for details.


Wednesday, April 08, 2015

Will more network/IT vendors launch their own WebRTC PaaS?

One of the most vibrant domains within WebRTC is that of "platform as a service", PaaS. There are numerous providers of cloud infrastructure, mobile SDKs and ancillary services that allow developers to embed WebRTC functions more easily than using "raw" JavaScript. Tokbox, Temasys, Twilio, acquired players AddLive and Requestec, telcos like AT&T and NTT and so on. 

There is a particular need for PaaS to support mobile devices which use WebRTC in apps rather than browsers (eg with iOS SDKs), or where specialised cloud functions are needed, such as video-mixing. They also appeal to developers unable to cope with complex or clunky aspects of WebRTC, such as the much-derided SDP protocol for connection setup.

But one emerging category I'm seeing is slightly different - it's where technology vendors are launching their own PaaS propositions, reaching out directly to developers with hosted platforms, rather than attempting to sell gateways or SBCs or media servers as products.

The most prominent examples are:
  • GenBand's Kandy platform
  • Acision's Forge SDK & PaaS [itself based on the acquired Crocodile platform]
  • Intel's investment into CafeX
  • Digium's Respoke
  • (Primarily a virtualised IMS PaaS, Metaswitch Clearwater also offers some cloud-based WebRTC gateway capabilities)
What differentiates these from the various other client SDKs is that they are moves by "product" companies into the "service" arena, with subscription or pay-per-use business models. That runs counter to the normal vendor model of upfront product license + maintenance/support contract.

Now clearly, in other parts of the technology industry that transition is fairly common - the large network vendors like Ericsson, Huawei and ALU all make large sums from "managed services" contracts for radio networks or other bits of telco infrastructure. Cisco owns WebEx for conference services, while many mainstream software companies have transitioned to SaaS/PaaS offers for business users.

Yet it is one thing selling a "cloud" or outsourced version of a product to your existing customers (telcos or enterprises) as a service - but quite another trying to derive revenues from an entirely new audience of web or application developers. Clearly, this in an attractive idea - but it doesn't mean that it's easy to achieve in reality.

In WebRTC, it is instructive to consider which vendors are not offering PaaS propositions to developers - it includes most of the main gateway or media-server providers. While Oracle, Ericsson, Sonus, Dialogic, HP, Broadsoft, ALU et al might offer client-side SDKs, they are not offering hosted, subscription-based platforms for WebRTC developers at present. (Notably, all the previous product/service crossover vendors above also sell gateways or provide SDKs on a "product" basis, too).

My belief is that the others do not, for the most part, want to take the upfront risks of setting up infrastructures and billing systems for PaaS (especially internationally), nor incur the marketing overhead of reaching the "long tail" of developers. Not only is there a lot of competition here (with services firms having existing customers, or specialised capabilities), but there would be a risk of channel conflict if (for example) they ended up competing vs. firms that themselves are launching PaaS services, but which are also customers for core-network infrastructure or SBCs. 

For telco-facing vendors, I don't expect to see too many more launching WebRTC-based PaaS platforms, unless either (a) it ties into a much bigger NFV/SDN strategy offering telco infrastructure on a managed-service basis, or (b) it's a completely separate, rebranded initiative like Kandy, primarily targeting enterprise software ISVs. Broadsoft's Labs arm is offering an "incubation environment" for developers, but it's not the same as a full PaaS.

On the enterprise side, I can see Cisco, Avaya, Unify and others increasingly offering API access to their own cloud-based UCaaS offers. However, I'm not expecting them to offer subscription-based WebRTC gateway functionality or similar propositions aligned with Respoke. Notably, Voxeo split off its PaaS business (Tropo, formerly Voxeo Labs), before being acquired by Aspect. The deal that Avaya has done with Google for hosted contact centres & Chromebooks is a step in this direction, but isn't really a formal PaaS. It also has a platform called the "Collaboratory" which seems similar in principle to Broadsoft Labs - ie a "PaaS for prototypes" rather than a "production PaaS".

All that said, there may be some future acquisition opportunities and disruptions here over time. I could perhaps believe a major vendor might try its luck acquiring Twilio or Temasys or CafeX or another cloud player, perhaps seeing it as a way to start generating more recurring revenues, and higher-margin services. However, such actions are perhaps more likely to come from the IT side of the industry (IBM? Microsoft? Google?) than network vendors.

For now, I think the vendor/PaaS crossover phenomenon is a relatively rare one - I suspect many others will just watch from the sidelines to assess whether any of the existing batch start getting notable traction and growth, or else they might go down the route of offering selected partners a "prototyping environment" rather than a full pay-by-credit-card cloud offer.

Tuesday, April 07, 2015

We are past the point of Peak Telephony

The telecom industry has a huge and under-examined problem: lack of demand for its core product, phone calls. We often talk about oversupply - competition/consolidation, MVNOs, replacement by Skype etc - and also about updated telephony (eg VoLTE), but we miss the elephant in the room: users are simply bored with phone calls. They aren't very useful or desirable any more, when there are other alternatives available.

The UK telecom regulator Ofcom publishes very comprehensive market reports, not just covering UK communications revenues and usage, but also giving international comparisons, partly based on data collated by another research firm IHS. But while it gives charts of trends in telephony usage (outgoing minutes) for both fixed and mobile services, it only does so separately. The data in the most recent 2014 report compares 2008 and 2013 call volumes for a broad range of countries, including major European nations but also US, China, India, Japan and Korea. (See pages 27 & 40 for fixed & mobile numbers)


I've gone a step further. I've added the datasets together, to look at the aggregate use of phone calls, both fixed and mobile. As far as I can see, nobody else bothers to do this, which is odd because it hides a very clear and interesting picture.




(There are a few grey areas in terms of definitions, but the Ofcom data does include "managed VoIP" - ie telephony based on VoIP/IMS run by operators. It appears to exclude non-operator VoIP such as on-net Skype, and I'm not sure what happens to SIP-trunk enterprise traffic, SkypeOut VoIP-to-PSTN termination etc.)


In other words, between 2008 and 2013, the total net amount of outbound phone traffic in the UK, Spain, Sweden, Netherlands and US fell in absolute terms. In Italy, Germany and Korea it was flat. We are past the point of "peak telephony" in many markets.
 

Indeed, looking at the last reported quarterly stats from Ofcom, the fall may be accelerating. In Q3'2014, the aggregate UK fixed & mobile volumes fell 8.5% from a year earlier. Within that, mobile volumes rose 1.5%, but that was against a 1.1% increase in subscription numbers, plus a broad (one-off) shift to unlimited plans.

Some countries buck the trend, because of elasticity effects. Increased competition in France brought much cheaper mobile calls from around 2012, and a certain amount of pent-up demand was  filled - but in Q4'2014, French aggregate telephony volumes fell too (see page 3, here), with fixed telephony declines outstripping mobile growth. Obviously in developing markets such as India, there is still plenty of unmet need for telephony.

But in places such as the UK, cheaper phone calls don't translate to more usage any more. Almost everyone has "more than enough minutes". Instead, people are switching to messaging, apps (who phones for a taxi these days?), social networks and so forth. For many social interactions, there are better ways of doing it than a clunky, interruptive old phone call. I still remember a friend telling me a couple of years ago "the only people who call me are my parents, or people I don't want to talk to".



In fact, I suspect that if we took out pre-arranged calls (eg conference calls for business at a specific time), that the numbers would be even more bleak.

This has some huge implications & observations:

  • Fixed-to-mobile substitution of telephony is still occurring
  • Mobile calling has natural limits & saturation near or past
  • VoLTE deployments are expensive investments in a declining market
  • Telecom operators largely fail to market & promote use of their core product
  • Big "minute buckets" or unlimited plans give a temporary blip
  • The perceived value of the telephony element of bundles will fall
  • This is not just "competition from OTTs". Mobile Skype/Viber use is relatively low
  • This is a fundamental problem that phone calls aren't optimal for some use-cases
  • Users (consumer & business) pick the best tool for the job. Which is often cheaper, too
  • Regulators think in terms of "voice competition", not comms use-cases and purposes
  • Communications value & effectiveness is not best measured in minutes
  • New forms of voice communication (eg embedded into apps) may help reverse the decline. APIs - notably WebRTC - will help here.
  • HD Telephony is nice, but unlikely to solve the underlying problem of lack of demand
  • All involved in voice (& video) communications need to understand why people use realtime spoken calls - and try every trick to make the human outcome better
  • Contextual communications, analytics and new user-interaction models are critical
Above all, we need to stop thinking that voice=telephony. We are past "peak telephony". But we do not need to be past "peak voice", if we think how and why the spoken word is actually being used.

Dean Bubley will be speaking or moderating at the following upcoming public events

If you would like a private workshop on the impact of Peak Telephony, The Future of Voice & the potential from WebRTC & Contextual Communications, please contact me via information AT disruptive-analysis DOT com