Pages

Pages

Tuesday, February 03, 2015

Inevitabilities, adjacencies and anti-forecasting



I get annoyed by “unintended consequences”. Too often, they are only “unintended” because they were not predicted. Similarly, many forecasts fail to become real, as they overlook predictable problems – or perhaps, distant external factors that cut the ground from under them. Another category of “predictable failure” comes from wishful-thinking “visions” that ignore other, unstoppable trends that make them impossible.

When it comes to analysing the future direction of technology markets – be it my normal stamping ground of telecoms, my broader futurism work, or even politics – I am constantly aware that companies, industry-bodies and self-appointed visionaries fail to look outside their narrow silos. Consistently, and near-universally.

Now, obviously, nobody has a perfect crystal ball about what will happen. But it is often possible to determine what won’t happen, or at least has a vanishingly-small probability. And it is also possible to identify other factors which will almost-certainly happen in the same time frame – and see their possible inter-dependence.

That’s all quite abstract, so a current real-world example: the use of online encryption, and the recent reaction from telcos and governments.

It’s been pretty clear to me for years that once Moore’s Law meant that data could routinely be encrypted with minimal cost (monetary, power, latency, inconvenience etc), then it would be. It was locked-in. Pretty much inevitable. All other things being equal, people like increased privacy and security – or at least, the perception of it. And even if “people” didn’t want it, then it seemed likely that a lot of companies would, on their behalf.

Encryption – like any security measure - can always be foregone, in specific circumstances, where it makes sense and all actors make rational decisions. It’s safer to make non-encryption the exception, not vice-versa.

The exact timelines, technologies and architectures were less-clear (to me at least). And it certainly wasn’t obvious that the largest catalyst would be fears raised by a whistle-blower about state surveillance – although the growing use of VPNs, Tor and other techniques in repressive regimes were pointing that direction anyway. Continued examples of hacking of apps and servers, data leaks, credit card databases stolen and so forth also made “more encryption” a safe bet as a generic forecast. Invasive actions by telcos inspecting or modifying data traffic have also been a contributor – albeit perhaps less than I expected in pre-Snowden days.

Now given that “more encryption” was what I’d call an “inevitability”, you might have thought that companies impacted by it might have started preparing long ago. Instead, the last 12 months have seen Governments and telcos panicked by the rise of HTTPS and SPDY, as well as proprietary encryption techniques used in apps, peer-to-peer technologies and VPNs. (I suspect that many security services had been quietly predicting this, but politicians seem to be treating it as a sudden surprise too).

Now I’m not going to make a call on how to “deal with” encryption or not here, whether it’s a good or bad thing in certain circumstances – I just want to point out that the situation has been predictable. Yet nobody seems to have run a filter of “hmm, what happens when it all goes dark?” over their existing products, services and practices, over the last five years or so.

The same concept of “inevitability” has also been the curse of many other technology domains. It has been inevitable since at least 2005 that, sooner or later, somebody would realise that sending 160 characters of text was pretty simple and cheap, and not worth 1-10c per message. Yet we had to go through years of vendors saying “mobile data can be worth $10000/MB, look at SMS!” without many people considering the inevitable conclusion that it wasn’t really “worth” that, when decent competition finally emerged. Instead, industry groupthink tried to pitch “value-based pricing”, when in reality it was “grudging-acceptance pricing”.

The fact that the mobile industry has probably pocketed an extra trillion dollars in profits from over-priced SMS over the last decade is a fortunate accident. The emergence of Whatsapp, LINE & WeChat was a predictable – nay, inevitable – eventuality. In many ways, it was overdue. It was only some fairly clunky UIs and low penetration which stopped it happening during the Symbian/J2ME era, as there were plenty of early pre-Whatsapp attempts. The signs were there.

Yet once again, that inevitability was ignored. Rather than making sensible attempts to defend SMS by adding value, the golden goose was ignored. Rather than reinvesting 5%, 10%, 20% in service innovation, SMS revenues were classified as “data” in operators’ financial reports and used to help justify 3G/4G licences and investments. This despite the inevitability that faster networks would make the risks even greater.

Telephony is next up. We already know the inevitabilities there – and although some vendors and software developers are (finally) trying to make phone calls more useful and “friendly”, that message hasn’t percolated through to many in charge of investment and service-innovation at network operators. Instead, they are focused on recreating Telephony 1.0 and putting the bulk of investment into things like VoLTE.

But telephony – and messaging – also have to counter effects other than just the “inevitability” of free/low-cost VoIP and IM. They also have to factor in the power of adjacency – things going on in the “silo next door” – or perhaps the silo down the street or over the horizon.

For telephony, adjacencies come from various use-cases for WebRTC-type contextual communications, as well as concepts like hypervoice. But they also come from changes in human communication more broadly – the replacement of some “voice” tasks with apps (eg booking taxis), or perhaps richer forms of interaction like augmented/virtual reality.

In the networking and broadband space, some adjacencies are starting to become visible – such as competition from new platforms such as satellites, drones and balloons, as well as direct peer-to-peer communications between devices. Others are less-obvious, such as the slow move of governments and large non-telecom companies into the domain of network ownership, as well as cloud services. Only today, bank Santander announced that it would offer online storage to businesses. Expect automotive, utility, healthcare and other providers to take prominent roles in IoT development – potentially including network ownership.

As another example, there’s a lot going on in the arcane world of SIM cards. Everyone remembers a couple of weeks of excitement around the Apple SIM last year. But that’s just the tip of the iceberg. Have you heard about the liberalisation of MNC codes, and what that might imply in future, for example? What about downloadable IMSIs? Blinded by the acronyms and obscurity? Well, that’s where some of the disruptions are potentially coming from. Fore-warned is fore-armed.

The trick here is to think not just in terms of projections and forecasts – but in terms of intersections. What other lines are coming up to meet your beautiful curves stretching out to the future? What happens to your assumptions when those lines cross?

This gets much harder when one tries to apply the same principles to more general forecasting, or futurology. It’s easy to get caught out by automobiles when you try to predict the evolution of the horse-and-cart. The “paperless office” failed to take account of cheap printers, better online document-publishing – and human psychology and behaviour. Many, many predictors of “convergence” have completely missed other trends which actually favour “divergence” and fragmentation.

There’s a lot of predictions about AI around at the moment – both utopian and (especially) dystopian. But few factor in other parallel trends, such as enhancement of human cognition, whether by software, pharmaceutical means, or even genetics. A whole host of societal trends also take on a new complexion, if one factors in increasing longevity, biomedical advances, robotics, nanotech, 3D printing and so on. What happens to our security (and encryption) when the smart lightbulb reads your email on your screen over your shoulder, or listens to your conversation, before you even get to tunnel it through a VPN?

The story here is to look beyond the upbeat, positive predictions and hockey-stick curves. They’re seductive – but you also need to have a Devil’s advocate view, trying to pick holes in the narrative. Ideally, the ideas are not just “robust” to criticism, but as per Nassim Taleb, they are “antifragile” and strengthened by the challenge.

I haven’t mentioned politics much in this post, but that’s an important domain for this type of analysis too. Many of the more populist agendas fall prey to the “unintended consequences” flaw. For example, imagine an anti-capitalist agenda that inadvertently breaks Moore’s Law, or the investment case for new factories to make chips, smartphones or PCs. Or perhaps, penalises entrepreneurs who make huge exits when they sell startups. At one level, it might be seen as preventing “planned obsolescence”, or reducing inequality. But if a knock-on effect is a slowing in technology needed for climate models, environmental sensor networks, design and development of new clean energy technologies – then it will have been a Pyrrhic victory, with severely negative consequences.

Visionaries tend to think in terms of clean, idealistic utopias. Or of one over-arching metaphor like a “personal AI”, or a centrally-determined allocation & orchestration of processing or networking resources in a perfect cloud/NFV/SDN telecoms industry. They forget about legacy technologies, second-order effects, human behaviour, regulatory/political concerns and practical issues getting from “here and now” to the sunny uplands of the future. If the route goes via predictably-dangerous territory in between, the idealists have a duty to scrutinise it in advance.

So in reality, the future is messy. And analysis of inevitabilities, and the practice of “anti-forecasting” (what won’t happen) need to form a part of any visionary’s or forecaster’s arsenal of weapons.

The quote that the future is “already here, but unevenly distributed” comes from William Gibson, whose awkward, heterogeneous, sometimes-jarring worlds of the near-future are much more realistic than the beneficent, techno-utopian, AI-assisted Culture envisioned by my other favourite author, the late Iain M. Banks.

Beware of “elegance” in technology (or socio-political) predictions – it’s almost inevitably wrong. Wishful-thinking is a useful thought experiment. But it’s not a “vision” – it’s just a screenplay or fictional plot, and usually a rigid one at that. Hybrids, overlaps, complexity, gaps, inefficiencies, blurred definitions, political and human realities, flexibility - those are the signs of a realistic forecast or prediction.

If you are interested in due diligence, Devil’s Advocacy, or an open workshop/brainstorm on possibilities, inevitabilities or anti-forecasting, please get in touch, at information AT disruptive-analysis DOT com.

1 comment:

  1. Davide2:16 pm

    "I get annoyed by “unintended consequences”."

    I believe it would be fair to put a couple of references to Mr. Nassim Taleb publications in your post.

    The application of his ideas into the Telco world are yours, you mention him in the middle, but I think more references to his work should be added to this post. It would also be useful for your readers unfamiliar to the concepts of anti-fragility, second-order effects, iatrogenic.

    ReplyDelete