Wednesday, March 2, 2016

Consumer Electronics technologies : need for global warming


Introduction



I was fortunate to attend both CES (Consumer Electronics Show in Las Vegas) and MWC (Mobile World Congress in Barcelona) this year and I came back twice with mixed feelings and similar conclusions. This short post will explore some of them. The title is a provocative summary: the visible part of technologies in consumer electronics, such as screens or device size, has reached a plateau, while the “inside technologies” continue to grow at Moore’s law rate, which is cool but not necessarily impacting. You end up visiting the TV stands of CES, or the smartphone stands of MWC, without the excited feeling that ”this year generation of device is so much better than last year”, which has been the case for the past 15 years.                

The abundance of technological innovation and progress, which has not slowed down, is finding an outcome with the multiplication of gadgets and accessory devices. However, most of them are “cool” but not “warm” : they do not address a user pain points nor demonstrate an immediate benefit for our daily life. What I mean by “global warming” is the necessary massive embedding of design thinking and customer centricity for the next generations of consumer electronics.

This blog post is organized as follows. I will first explain my double tweet about the Bazar (which is thriving) and the Cathedral (which has stalled). I will then focus on the smartphone and explain why I see a plateau in its current evolution. The third section is a refresher on a theme that is common to my two blogs: the need for story telling and design thinking. The last section is a follow-up about the coming of AI in our smartphones to transform them into smart assistants.

 1. The Bazaar and the Cathedral   

      

I am borrowing the metaphor from Eric Raymond’s bestseller. In his book, the bazaar is the world of open source software, compared to the cathedral, the world of commercial software sold by ISV, independent software vendors. In this post, the Cathedral is the set of large, expensive booths from the well-known brands of Consumer Electronics. They have always been the stars of shows such as CES or MWC : very large, very crowded, beautiful and innovative displays, entertaining hosts and hostesses, gifts and joyful excitement. The Bazaar is the grid of tiny booths rented by startups and small technology players - most of the time, less than 10 square feet and no special-effect-displays. A few years ago, one would spend most of the time in the Cathedral – there was so much to see – and do a quick visit to the huge bazaar (thousands of small booths) in the hope of serendipity: to detect an early product or startup innovation that could complement the rising tide of CE products.

In 2016, the Cathedral has stalled and the Bazaar is thriving. This is very striking for someone who has been visiting CES over a decade. The huge booths of the Cathedral are surprisingly similar to what they looked like last year or two years ago. The flagship products, TVs or smartphone, are also very similar. Some booths are actually making this very clear: Samsung in Barcelona used a tiny fraction of the space for the new S7 flagship and most of the booth as a retrospective of past innovations. The crowd is still there, but there are no huge lines to try new smartphones or see new TVs. The “shows within the booth”, a trademark of cathedral organizations, are much more scarce and much less joyful than in the previous years. On the other side, the Bazaar is bursting with newly found energy and vastly improved self-organization. These actors have always been there, but it is clear that the ecosystem is changing: Internet of things (explosion of sensor technologies), ubiquity of the smartphone, faster and easier access to computing power, etc. Many of the small players now come with innovations that are much closer to: (a) user needs, (b) easy delivery to customers, than what we would have seen in the past. The common lore that “it is today much easier to build a high quality product with less resources”  is clearly shown to be true if we consider the quality of what small companies are able to present at CES or MWC. There is also a much better organization, either from a geographical or topical perspective: pavilions have emerged to create hot spots, such as the FrenchTech, Israel Mobile Nation, or standard-focused associations.

The combination of these two trends still makes for both exciting 2016 issues of CES or MWC. At CES, new cathedrals are being built with the explosion of connected cars. The technology innovation stream still produces a continuous exponential increase of raw computing power – as show in CES by the Drive PX2 board from NVidia for embedded smart car computing with the CPU/GPU power of 150 MacBook Pro on a single board, ready for embedded deep machine learning. Similarly, the continuous exponential improvement of video processing capabilities is quite spectacular with examples such as 360 real-time video stitching. The constant improvement of sensor capabilities is also pretty amazing. Smart objects for e-Health are now embedding medical-quality sensors (in response to previous concerns such as with Fitbit) with impressive capabilities (for instance, the electrocardiogram wrist band Qi from Heha). This improvement of sensing goes hand in hand with miniaturization which fuels the IoT explosion, which was very visible both at CES and MWC. New domains such as smart clothing are bursting, while more usual CES IoT sections such as e-Health or Smart Home are bubbling with new energy. This continuous growth is fueled by constant progress with the silicon, as well as the emergence of de facto API-based ecosystems such as IFTTT, Alexa (Amazon), Smartthings (Samsung) or ThinQ (LG) … with many cross-fertilization such as this or that.

However, this explosion of cool technology does not leave the visitor necessarily with the warm feeling of usefulness. The idea that IoT exponential progress is “cool” but not warm enough is not new. I made a similar comment in 2013 when visiting the “Smart Home Conference” in Amsterdam. I was already quite impressed by the availability of all the connected objects that are required to make one’s house smartly heated, lighted, filled with music, more secure, etc. However, what was cruelly lacking then was, and still is, the availability of a true “user-centric proposition” delivered by a credible brand. What I mean by “user-centric” is not a surprise:
  1. to solve a real pain point,
  2. to deliver clear and immediate benefits with a promise from a brand,
  3. to assume the promise and help to setup the “smart system”,
  4. to face the customer (accessibility) and not to shy away from troubleshooting and customer service.
As a Uber or AirBnB user, it is not the technology that impressed me, but the fact that there is a reachable company who fulfill the promise and who takes accountability.

2. The Plateau of the Smartphone Platform



We all know what we expect from a smartphone : great battery life but light to carry, beautiful screens with great resolution and vibrant colors, thin and elegant but robust, fast so it responds to touch instantly and runs our preferred apps. There are obvious conflicts in these goals, which makes the engineering problem interesting. It looks like we are reaching a plateau in smartphone evolution for two reasons:

  • For some the goals, we are near the optimal performance that the human user may appreciate. It is very clear for screen resolution (this is why the “retina display” label was proposed by Apple) and it is also true for screen size (I find it interesting to see that 5 inches seems to be the “average optimal size” since this is the value that Intel announced more than 10 years ago, before the iPhone was introduced, as the result of an in-depth customer study).
  • Some of the constraints are between “click and mortar” disciplines, between exponential technologies and mechanical engineering, and the rate of improvement is much lower for the second group. For instance, it is clear that the improvement of CPU/GPU comes with an electric consumption price and that battery capacity is slowing down the evolution. Similarly, thinness comes at the expense of sturdiness.

As a consequence, 2016 smartphones are no more exciting than those of last year. 2015 was full of really thin smartphones (from 5 to 6 mm), they are all gone. Following Apple who made the iPhone6S heavier than the iPhone6, manufacturers are shying away from the ultrathin design. On the contrary, Samsung new flagship (S7) is 152g and 7.9 mm thick, compared to the S6 138g and 7.9 mm. This is the price to pay to get a better battery life coupled with faster processors. Consequently, when you play with the new models in 2016, there is not much excitement compared to the same MWC 2015 visit last year. Sure, the processors are slightly faster (although 8 cores seems to be enough for everyone, and they were there last year already), but it does not make a big difference (yet … cf. last section). I could tell exactly the same story for smart watches: they are strikingly similar to last year’s models, still too thick, with battery life that is too small and app performance which remains sluggish. I am a great believer in smart watches and I live with one, but we are still in the infancy of the product. Today it remains a geek product.

Because I am conjecturing a plateau, I would not be surprised to see (Android) smartphone prices go down sharply in the next years (when the feature races stalls, commoditization kicks in). This is in itself a joyously disruptive piece of news since it means that the smartphone will continue its replacement of the “feature phone” in all markets, including emerging countries. Meanwhile cathedral brands need new products to keep us (richer customers) spend our money and the star of the year is clearly the VR (virtual reality) headset. I have been, like all visitors, quite impressed with the deep immersive capability of these devices when playing a game. I can also see a great future for business augmented reality applications. On the other hand, I do not see these headsets as a mass-market / everyday use product, no more than 3D TV convinced me 5 years ago.

3. Fed up with data, looking for stories


I have been really excited about quantified self for the past five years. I have bought and enjoyed a large number of connected gadgets, adding the apps and their dashboards to my iPhone. I still appreciate some of them, because they helped me to learn something about myself, but most of them are standing in my own “useless gadgets museum”. I developed while still at Bouygues Telecom the theory that, in its current state, “quantified self” addresses a unique group of people with (a) a “geek mindset” to cope with the setup and the maintenance of these gadgets (b) a “systemic” interest to see value in dashboards and to learn from figures and charts (c) a good measure of ego-centricity – not to say narcissists :) . This is a hefty niche where I find myself comfortably standing, but it is still too small to scale most existing “quantified-self value-propositions” linked to wearable connected devices. We know that many connected objects companies have not met success in the past 18 months. I have heard many times the same story from some of the “quantified self players” that I have interviewed: the launch starts well with a small group of enthusiasts, but does not scale.

For most people, there are two missing ingredients to fulfill the promise of quantified self: a story and a coach. There is indeed a great promise, since science tells us everyday that we can improve our health and our well-being by knowing ourselves better and changing our behaviors accordingly. The need for story telling, in the sense of Daniel Pink in his best-seller “A Whole New Mind”, and coaching (personalized customer assistance) is not limited to e-Health, it is also true for other fields such as Smart Home or Smart Transportation. For most of us, self-improvement needs to be fueled by emotions, not dashboards. To paraphrase Daniel Pink, I could say that connected devices must adress our right-brain more than our left-brain :) The importance of design thinking is one of the reasons startups seems to have an innovation edge. Focusing too much on data and data mining fail to recognize the true difficulty of self-change – one must read Alan Deutchman’s excellent book “Change or Die”. I will make “Lean User Experience” the theme of my blogpost next month to speak about this with more depth. It is very clear when wandering around CES booths that too many connected wearable are still focused on delivering data to their users, not a clear self-improvement story.

This is not to say that there is not value in Big Data for e-health or well-being, quite to the contrary. As explained earlier, science shows that we have a wealth of knowledge in our hands from these life digital tracers. If you have any doubts, read “Brain Rules”, the bestseller from John Medina.  You will learn what science says about the effect of exercise on your brain capacity (IQ) through better oxygenation, just to name one example. However, delivering value from data mining requires customer intimacy, from a large number of reasons ranging from relevance to acceptance. This obviously requires trust and the respect of data privacy, but it also requires a strong and emotional story to be told. I have no doubts that big data will change the game in that field of e-health, but as a “systemic reinforcer”. One needs a warm story and a clear value proposition to start with.

4. Waiting for smart assistants?


I strongly recommend reading Chris Dixon post on “What’s next in computing ?”. I share a number of his ideas about the current “plateau state” and the fact that the next big thing is probably the ubiquity of AI in our consumer electronics devices. He also points out the coming revolution of miniaturization / commoditization of computers, and the impact of the smartphone as a billion-units computing platform. The image of a “plateau” is not one of an asymptotic limit (i.e., I do not mean that we have reach a limit about how smartphones may be improved). I have no doubts that engineering constraints will be resolved by technological progress and that new generations of smartphone will delight us with considerably longer battery lives, thinner designs and reduced weight. The image of the transparent smartphone in Section 2 shows that there is still much to come about this ubiquitous device. My point is that we have some hard engineering / chemical / mechanical constraints in front of us so it will take a few years to solve them.

Meanwhile, the exponential growth of smartphone performance will come from within, from the software capabilities (fueled by hardware performance exponential growth). Both because GPU are getting amazingly powerful and because we see deep learning specialized chipsets becoming available to be embedded into smartphones in the next few years, it is easy to predict that deep learning, and other AI/machine learning techniques, will bring a new life to what a smartphone can do. We may expect surprising capabilities in the field of speech and sound recognition, image recognition, natural language understanding and, more generally, pattern recognition applied to context (contextual or ambient intelligence). Consequently, it is easy to predict that the ability of the smartphone to help us as a smart assistant will grow exponentially in the decade to come.
Artificial Intelligence on the phone is already there. “Contextual intelligence services” are already the heart of Siri, Google now or Microsoft Cortana. To the large players I should add a large number of startups playing in this field such as SNIPS, Weave.ai or Neura. But two things will happen in the years to come:  first, the exponential wave of available computing power is likely to raise these services to a new level of performance – starting with natural language understanding – and the capabilities of the smartphone will make it possible to do part of the processing locally. Today the “AI part” of the service sits on the cloud, leveraging the necessary computing power of large server farms. Tomorrow, the smartphone itself will be able to run some of the AI algorithms, both to increase the level of satisfaction and to better respect the privacy.  This is the application of the multi-scale intelligent system design, which I have covered in my previous post. Neura, for instance, is precisely positioned on how to better protect the privacy of the user’s context (personal events collected by the smartphone or other connected/wearable devices). By running some of the pattern recognition activity on the smartphone, a smart assistant system may avoid moving “private events” from the smartphone onto a less secure cloud distributed environment. This is precisely the positioning of SNIPS: to develop AI technology that runs on the smartphone to keep some of the ambient intelligence private. I start to recognize a pattern which I call “Tiny Data”: the use of data mining and machine learning techniques applied to local data, kept on the smartphone for privacy reasons, using the smartphone as a computing platform because of its ever-increasing capacities.

I have selected the movie “Her” as an illustration of this section because I think that it is remarkably prescient about what is to come : the massive arrival of AI into our personal lives, the key role of the personal assistant (and all the dangers that this may cause) and the combination of smart device and cloud massive AI.


 
Technorati Profile