Wednesday, September 27, 2017

Early Indications September 2017: Potpourri for $400, Alex


As I’m reading 200+ newsletters in preparation for Early Indications’ 20th anniversary next month, a few ideas are floating around. In no particular order:

1) Companies for the next 20 years

Looking back to when I started this exercise, such leaders as Digital Equipment, Sun Microsystems, Nortel, and Compaq have been bought and/or shuttered operations. Kodak and Nokia are a tiny fraction of their former size and influence. Circuit City and Borders are dead; Toys R Us is in chapter 11. Looking ahead 20 years, what companies can we bank on still existing, much less dominating their market, in 2037?  Short answer: Amazon, given a caveat, and some morphed version of Google.

Jeff Bezos is 53 years old. It’s not impossible to imagine him still running Amazon for two more decades, but if he does not, it will be fascinating to see what kind of succession plans he makes: absent its founder, I don’t know if Amazon can maintain its incredible innovativeness and creativity. I’ve never admired a company so much for a reluctance to copy industry leaders: I hate the term “best practices” because it all too often excuses managers from identifying and solving problems with original solutions. Given Bezos, I’d bet on Amazon still being a leader 20 years from now.

Even though its founder has similar maverick tendencies, Tesla is much harder to pick as a winner. Until Elon Musk and his team deliver all those pre-ordered Model 3 sedans with minimal hiccups, I can’t be confident in the operational acumen Tesla managers bring to basic execution. The scale of Musk’s bets is admirable, but as a friend of mine once said, “cash flow is as immutable as the laws of physics: get things going the wrong way, and you’re quickly swamped by the vortex.” Is Tesla a car company, a battery company, an energy company, a mobility company? Is Musk a car guy, a mass transit guy, a space guy, a social commentator guy? At some point these declarations will have to become clear. 

In the same vein, what will Dyson be without its founder? The electric car announcement is vague yet tantalizing, but is this a company for the longer haul?

As the world (including the US) becomes more urban, and new transportation patterns — including electric bicycles — emerge, where does that leave Wal-Mart? Also, the company’s founding generation is no longer in command, so how long can managers maintain the cost discipline, culture, and vision that made the company what it is today? At what point does today’s retail leader join A&P, Sears, Macy’s, and JC Penney as also-rans?

Apple is a puzzle. They reap high margins, run a phenomenal supply chain, and earn fierce customer loyalty, yet the market values the stock with a pedestrian P/E ratio (at 11.7, it’s lower than Kohl’s; Facebook’s is about 37). Once again, we see the effect of losing a visionary founder. F. Scott Fitzgerald quotably asserted that “there are so second acts in American lives,” yet Apple is on at least its third. How many more rebirths does the company have in it? One potentially bad omen: building a luxurious headquarters building. A good omen: the company’s impressive move into custom silicon, as evidenced by the insane A11 processor in the new phones.

With 5G wireless delivering the potential for Verizon and ATT to end-run the cable operators with residential broadband in urban and suburban markets, picking winners in the digital pipes business is tricky. Comcast and other cable companies are learning lessons with wi-fi that could carry over into 5G, which relies on smaller cells and thus requires more backhaul nodes in the architecture than today’s cell networks. Too soon to tell here.

I see speed and agility beating size and scale in more settings, and thus bet against IBM, GE, HP, and GM existing as independent companies 20 years from now. Much as Uber is fast and agile, meanwhile, I don’t see the company fending off all the competitors it will face, especially in the absence of a coherent and viable corporate culture: with 1.5 million contractor-drivers being the face of the company, they have to believe in and embody the company’s values. I don’t know how much this is the case elsewhere, but 90% of the Uber cars I observe here in our college town bear dual branding with Lyft. Thus Uber will be nibbled away and weakened from within by its founder’s toxic legacy.

Finally, what of today’s ad-driven eyeball aggregators, Google and Facebook? Much as I want Facebook to go away for the good of humanity, I don’t think it will. As for Google, the new focus on machine learning competence will position various Alphabet bets to win. Whether the company looks like it does today, I can’t predict, but count on a recognizable remnant to persist.
Banks have deep lobbying pockets and thus many friends in Congress, making insurgency in financial services extremely difficult. Both investment banks and retail institutions profit whether their customers do or not, and this state of things looks like it will be legislated as fact for the foreseeable future. (Recall the industry’s massive, organized resistance to anyone speaking officially for the consumer.) I expect more consolidation but have no idea which players are predators and which are prey.

Other companies I don’t think will make the 2037 opening bell: Fiat Chrysler, Sony, Samsung, Walgreens. I’m much more confident in Corning (good culture, and culture of innovation), Patagonia (along with the Swiss Mammut, one of the few independent outdoor companies not part of a conglomerate), and Canon (which has diversified far beyond cameras).

2) The place of technology in pop music

At the risk of sounding like the “get off my lawn” guy, I was thinking about how little the Internet and smartphone age has inspired musical references. Compare these world-altering technologies to airplanes, trains, radios, cars, records, TV, movies, photographs, and wireline phones, which appeared with great regularity in pop songs across many genres. We have yet to hear “Twitter blues,” “Friend me like you mean it,” or “using Google Maps to go see my baby.” 

Look just at the blues: you could fill a hall of fame with only railroad songs. Pound for pound, what else (except maybe romance, and I’ll still lay money on the iron horse) can beat this top 5: 
  *John Henry, 
  *“City of New Orleans” 
  *“The Midnight Special” 
  *“This Train Is Bound for Glory,” and, oh yeah, 
  *“Mystery Train” (recorded by everyone from Elvis, Clapton, Dylan, the Dead, and the Doors to a great YouTube nugget from Grace Potter). 

Honorable mention: “Waiting for a Train,” “Folsom Prison Blues,” “5:15” (The Who), “All Down the Line,” “Let It Rock,” “Lonesome Whistle,” “Night Train,” “Casey Jones,” “Crazy Train,” “Orange Blossom Special,” “Rock Island Line,” “Train Kept A Rollin,” and “Wabash Cannonball.” (That doesn’t even count “Take the A Train,” which is a great Duke Ellington song about a subway.) 

Automobiles are similarly richly represented. “Rocket 88” was a seminal R&B hit for many artists beginning with an uncredited Ike Turner, produced by Sam Phillips. Pickup trucks are a country music staple. Early rock and roll regularly featured cars (“Little Deuce Coupe,” “Pink Cadillac,” “Mustang Sally”). Prince memorialized his Corvette, Tracy Chapman her Fast Car, and Lucinda Williams Car Wheels on a Gravel Road. Artists as diverse as War (“Low Rider”), John Hiatt (multiple tunes), the Eagles (“Take It Easy”), Ides of March (“Vehicle”), and the Grateful Dead (“Truckin’”) all wrote great car songs. (And there are plenty of highway songs too: think of Dylan and AC/DC.)

The contrast to computer songs is striking. (Also, I distinguish between computers as technology and robots as fantasy: there are a LOT of robot songs that aren’t about computing.) First, there aren’t many, and second, the best go back a long way: Talking Heads’ “Life During Wartime” (1979) and Prince’s “Computer Blue” (1984). In a related vein, there’s also the great surf-guitar hit “Telstar,” about the satellite, dating from 1962 (UK: The Tornados) and 1963 (US: The Ventures). Radiohead released OK Computer in 1997, but how many people can name even one song from it?

Smartphones and social media have turned up in a fair number of hip-hop numbers, but I can’t assess their relative stature or staying power. Some appear to be continuations of the landline phone staples: will s/he call? Why won’t s/he call? Beyonce did address explicit videos in “Video Phone” (2009); texting shows up in more than a few scenarios, many of them drunk and/or regrettable. None feel anthemic. I couldn’t find anything apart from minor artists or parody bits regarding Facebook and Google. There are actually songs about Microsoft, but I couldn’t motivate myself to listen to any. 

I’m not sure what all that adds up to. Maybe mechanical things that make mechanical noises are more sonically evocative than silent silicon and glass. It does seem significant that such potent cultural symbols and presences have not inspired people to write songs. As to why and so what, I’ll leave that to others.

Friday, September 01, 2017

Early Indications August 2017: Reflections on 20 years


Twenty years ago, I went to work for a think tank run by Ernst & Young to lead research into online commerce. I had a robust travel budget and attended a lot of conferences: Internet World, Wall Street Journal Internet Summit, Vortex, Demo, and others. I wrote trip reports that developed a following among colleagues, so in October the Networked Commerce Update newsletter was launched. It was published monthly or twice-monthly, and featured book reviews, research updates, observations, predictions, and conference reports. It’s difficult to believe it’s been running, under various names, for 20 years come October.

So much of what we consider “internet” technology didn’t exist at the time. 

  • Google launched the following year; AltaVista was the “cool” search engine, while Yahoo was still curated by humans. 
  • Texting was virtually unknown in the U.S., though it was taking off in the Nordics. 
  • Consumer GPS was still a few years away. 
  • Wi-fi, MP3s at scale (Napster came along in ’99), and wireless cellular data were all moving from the lab into commercial reality. 
  • Digital cameras stood alone and had underwhelming resolution: the Sony Mavica stored 640x480 images on a 3.5” floppy disk but was a big hit among real estate agents. (Kyocera introduced a Japan-only camera phone a couple years later.) 
  • Wikipedia was preceded by Microsoft Encarta.  
  • Facebook and Twitter were, to a degree anticipated by AOL.
  • Computer monitors were heavy and large. The thought of whole cube farms with double 24” monitors was impossible due to both heat and weight.
  • Apple was essentially dead in the water; the stock was under a dollar.
  • Dialup modems (remember the sound?) were the primary means of connection in the U.S. 
All that said, 1997 had winners. Amazon IPO’d in May 1997: $100 invested then would have been worth about $64,000 at the 20th anniversary. eBay was cracking lots of codes followed by later success stories (including Uber) but had yet to IPO, nor had PayPal helped it scale. Mapquest was paving the way for later wayfinding services. Linux was alive and thriving en route to outlasting AIX (IBM), Sun’s Solaris, HP-UX, and others, becoming the free and open basis on which everything from a watch to a supercomputer could be built. Enterprise websites could already look to Cisco for the winning playbook; FedEx package tracking was one of the first real-time processes ported to a web browser.

Microsoft was such a big winner, particularly after Windows 95 proved to be the on-ramp to Internet browsing and email with its integration of the Internet Protocol communications stack, that the Clinton administration sought to break it up in the style of AT&T about 15 years earlier. The George W. Bush administration later reversed those filings, and the smartphone market eventually did what legal remedies did not: reduce Microsoft’s monopoly power.

Still, the winners are far outnumbered by companies that went away, got bought, or got bought then went away. 

-By 1997 Digital Equipment was in trouble even though they pioneered a 64-bit processor and their search engine was state of the art. Compaq bought DEC in 1998, only themselves to be bought by HP four years later in the wake of the tech-stock bust.

-Netscape’s superhero leadership team of co-founders James Clark (founder of Silicon Graphics) and Marc Andreesen (who cowrote the Mosaic browser at the University of Illinois supercomputing research center) and CEO Jim Barksdale (ex- ATT wireless CEO, ex-FedEx COO) was outmaneuvered by Microsoft’s bundling of Internet Explorer; it was also never clear what the browser business model was. Netscape was bought by AOL, which merged with Time Warner.  

-@Home was an early broadband provider launched by several cable operators in 1996. The company’s 1999 merger with the Excite portal (which had discussed acquisition by Yahoo a month prior) combined “pipes” and content in a disaster. The merged entity’s stock price peaked at $128 before dropping to $1 in late 2001. One of the company’s moves helped define the “Internet bubble” as it bought the online greeting card company Blue Mountain Arts for $780 million; that company was subsequently bought by American Greetings in 2001 for $35 million.

It really is hard to digest just how much of daily life is new since 1997: imagining a day without texting, Facebook, Twitter, Google, YouTube, smartphones, Apple IoS devices, glass keyboards, GPS, wi-fi, or flat-screen displays feels like harkening back to horseless carriages in 1930. 

What will be next? That is, what parts of life in 2017 will seem quaint and unimaginably primitive in 2037? I have two candidates:
  • I have said many times that the car will change more in the next ten years than in the last 80. Many automakers are declaring their intention to shift to alternatives to internal combustion engines, many cities are embracing bicycles seriously (Seattle is successfully slashing the number of single-occupant vehicles), and autonomous technologies will benefit from the leaps being made in machine-learning hardware and software. What that adds up to I don’t know, but e-bikes will be a pretty significant piece of the mix, I’ll wager, especially in the world outside the U.S.
  • Not only will the world be much more densely populated in 2037, en route to ~9 billion people by 2050, but it will be older as life expectancies increase on every continent. Thus I predict we will see new attitudes, accommodations, and applications of technology to the aging process. Exoskeletons, active prosthetics, and cognitive enhancements (possibly via augmentation similar to cochlear implants or pacemakers) will help address dementia, loss of mobility, and other consequences of long life. Rather than increase health costs with dramatic and expensive interventions soon before death, perhaps we will invest in quality of life over the longer term. I won’t go on the limb to predict the U.S. health care system’s blueprint, but do imagine that the current broken model will be replaced by something different.

Next month we'll take a closer look at those 20 years of newsletters, seeing where I missed big developments or made silly predictions.

Monday, July 31, 2017

Early Indications July 2017 Review essay: Machine Platform Crowd by Andrew McAfee and Erik Brynjolfsson

Despite U.S. CEO pay being a bigger multiple of average wages than ever before, the job is also incredibly stressful, with intense shareholder pressure. Ford, GE, and Macy’s all have new chief executives this year, in part because product companies are valued much lower than platforms: Ford stock currently trades around $11.00 a share while Tesla is in the 300s. Former Ford CEO Mark Fields is out of a job; Elon Musk is a worldwide rock star.

Academic readers who follow Internet economics have known of MIT professor Erik Brynjolfsson’s work for nearly 25 years, while general tech/business readers know his collaboration with Andrew McAfee, The Second Machine Age, that has sold well for the past three years. Their new book is very sound in its architecture, and even though I wish it went forward to use the model more (see below), I will be assigning this title in both undergraduate and masters’ classes starting in August. That makes it the first book I have ever assigned for business-school student purchase.

The book is written with exceptional clarity over a robust conceptual scaffolding. The global economy, we are told, is negotiating three ”rebalancings” as digital business practices expand their reach and impact. First, thinking machines (we learn a lot about the game of Go) can outperform humans on new kinds of cognitive tasks previously thought to be impossible for computers. Second, platform business models (Google, Apple, Amazon, Facebook, and Uber lead the list) outperform sellers of products; witness GE’s effort to become an industrial Internet of Things platform under former CEO Jeff Immelt. Finally, firms are no longer the default way to get things accomplished: digitally coordinated crowds are counterposed against “the core,” which means formal organizational structures: “the knowledge, processes, expertise, and capabilities that companies have built up internally, and across their supply chain” (p. 15).

Never do the authors go into Chicken Little mode and say the corporation is endangered or that all non-Internet stocks are “toast.” Rather, the challenge will be to negotiate the balance between each of the three dualisms. We will for a very long time have various blends of machine and human, platform and product, and crowd and core. The book’s examples of the three dualities are well chosen, hitting the right note between familiarity and below-the-radar startup, and with global coverage.

It should be clear that I find the book very valuable; it’s attractively comprehensive and accessible without being pedestrian. That said, it could be better, and I will note some little things then some bigger things that could aid in this pursuit. First, I don’t find the “core” concept as intuitively solid as the other five 5 tent-poles — readers can decide for themselves. Secondly, the book includes those annoying business-book checklists at the end of every chapter. This may be a marketing-mandated reaction to demographic change given that old guys like me might not be the meat of the book’s target audience, but the last book (that lacked the chapter end matter) sold very well for Norton, who also published the new book. I’m not sure why they were deemed necessary this time around.

More substantively, it concerned me that Uber gets let off without much of a critical reading. The company absolutely is rewriting our sense of what is possible in Internet business practice, and an incredible growth rate demands its inclusion. Still and all, nothing is said regarding the company’s extraordinary string of allegedly criminal conduct regarding Google IP (currently being tried in a California courtroom), outright deceit (proven lies told to drivers, riders, regulators, and likely investors), and a systematically sexist corporate culture (in addition to treating female employees horribly, the company has spied on female reporters who use the service and compiled the “rides of glory” analysis of men who were driven to an address not their own after some late hour at night). McAfee and Brynjolfsson are correct to include Uber as a harbinger of machine/platform/crowd dynamics, but to leave the company’s broadly and consistently appalling aspects unspoken cheapens the book.

My larger disappointment is more structural. The three rebalancings are very much taken in isolation, but combinations thereof show the power of the currents we are navigating. Google/Alphabet is a machine learning platform company, whether in its autonomous vehicles, its Nest thermostat business, or its core advertising/search operation. Amazon uses crowds (to write reviews of products and reviews of reviewers), it is a portfolio of platforms from AWS to Kindle to Echo to partner sellers, and machine learning drives everything from product recommendations to shipping time estimation to asset provisioning. In other words, what do the two MIT researchers have to say about the systematic impact of overlapping machine, platform, and crowd expertise? Not a lot. A few companies are mentioned in multiple sections of the book, but in the conclusion, where I expected to to transcend the three silos, there is nothing. In 2017, can a company aspire to platform status without machine learning dominance? I doubt it. (Look at aspiring platforms Snap or Slack for confirmation.) Why is this concurrence so? Perhaps platforms are the strategic objective, machine learning (and human talent expertise alongside it) the core competency, and crowd utilization a byproduct of the platform’s scalability?

Let me end on a positive note, for my disappointment is a product of how much value I found in the base typology. Rather than being written for Silicon Valley (“here’s how to create the next billion-dollar startup”), Machine Platform Crowd feels directed at the rest of us. That is, rather than informing the next Mark Zuckerberg or Travis Kalanick (both of whom, in my view, have profited too much on the privacy tax they levy on others), the book feels like it is sounding the alarm. Unless the likes of Caterpillar or GM or Marriott or Time Warner or Target or your favorite Big State University get the joke and build credible platforms on top of innovative machine thinking (in part by harvesting crowds’ insight and diversity of outlook), far too much of our future economic welfare, privacy, and consumer choice will be concentrated in a radically distilled winner-take-all economy. McAfee and Brynjolfsson are cognizant of the broad ramifications of their work, pragmatic but far from confident that this movie will end happily for the mass of the world’s middle class:

“Depending how they are used, machines, platforms, and the crowd can have very different effects. They can concentrate power and wealth or distribute decision making and prosperity. They can increase privacy, enhance openness, or even do both at the same time. They can create a workplace imbued with inspiration and purpose, or one that is driven by greed and fear. As the power of our technology grows, so do our future possibilities. This potential increases the importance of having clarity in our goals and thinking more deeply about our values” (p. 334).

Now more than ever, where are the relevant yet enduring resources for that pursuit? McAfee and Brynjolfsson have asked many of the critical questions, but it is up to far more of us to help answer them.

Sunday, June 25, 2017

Early Indications June 2017 Review essay: Scale by Geoffrey West

Review Essay: Scale: The Universal Laws of Growth, Innovation, Sustainability, and the Pace of Life in Organisms, Cities, Economies, and Companies by Geoffrey West

Why are things the size they are? That is, why can’t a mammal be as small as an ant? Why are cities, social and kinship networks, and elephants not 100 times their current sizes? Why are school districts, park districts, policing areas, statistical metropolises, and counties sized a) as they are and b) not consistently or uniformly? Finally, why and when do things stop growing and when do they die?

Geoffrey West trained as a theoretical physicist, migrated into biology, and then pursued network analysis of things, people, and places. Scale is a great book: West is asking important questions, at a broad scale, in almost complete opposition to current academic tendencies toward hyper-specialization. Further, the answers to the questions are not of arcane interest, as one would expect from a theoretical physicist. Instead, a parent who today gives birth to an infant who will likely have a life expectancy of about 100 years (given some early breaks in the right direction) should be asking the same ones on behalf of the newborn:

-Where does planetary population doubling each its limit?
-How fast will humans run out of critical natural resources, whether water, titanium, or animal protein?
-Will science crack the code of human aging?

For all the potency of West’s big questions, he brings plenty of fascinating new answers to this book. Indeed, the science of scaling has advanced extremely rapidly in just 20 years or so, providing tantalizing insights into the essential nature of many natural and human-made phenomena alike. All mammals have a life span of about the same number of heartbeats. The network structure of tree branches and the human aorta share mathematical proportions and relationships. Similarly, the networks underlying biology and urban infrastructure such as electricity and plumbing are uncannily similar, down to their exponent. The book is full of such causes for wonder.

The book’s chapter on the “science” of companies was for me the weakest link. As West states, streets and sewers don’t make cities: people do. Thus he wonders why cities don’t “die” the way whales or companies do, but companies aren’t organic: they are humanly created entities designed to fit within tax codes, government regulations, artificial geographic boundaries, and other constraints. Imagine one nation that requires companies to pay dividends to shareholders, for example. A second country that allows Amazon-like reinvestment of profits will give rise to very different business practices and structures. Change a parameter like differential voting rights of shares (as at Google/Alphabet), taxation of offshore profits (as at Apple), or capital requirements (JP Morgan), and once again, the “metabolism” of the corporate body will change markedly. Given the extensive and malleable artifices within which they operate, to liken a company to an organic life form is a quest with limited utility.

The material on cities, however, is compelling and original, at least to me (an admitted non-student of urbanist scholarship). I hadn’t realized the sheer velocity of global migration to cities, for one thing: over the next 35 years, he asserts, 1.5 million people will be urbanized every week. The other stunning discovery, to me, was that the increasing size of cities scales reliably with the speed of city life. That is, spatial density changes the experience of time. Finally, the analytics behind a predictive model that addresses such variables as crime, patent creation, and disease as a function of population are thoroughly impressive. This model gains some of its power by drawing on the literature of complex adaptive systems, as befits a senior member of the Santa Fe Institute: cities and bacteria are remarkably similar fractal structures, for example, as are flow rate maps of trucks from central distribution hubs and cardiovascular arteries. Similarly, West’s analysis of cities by their “metabolism” (energy use) is enlightening and concerning.

West did not write Scale to be beach reading, or to generate buzzwords with a half-life of a few months. Rather, the scaling of the planet will require more and more frequent game-changing innovations that reset the field of play (life is speeding up, remember?). The printing press and the steam engine were introduced about 300 years apart; the personal computer and the smartphone hit mass markets about 30 years apart. What will be the next innovation with Internet-sized impact? Given the impending global population increase to 9 billion people, West argues we need something in the next two to three decades, and then another breakthrough about 25 years after that. The relentlessness of superexponential growth curves was well summarized in a puzzle: If we have a bacterium that doubles its volume every minute, and it starts doubling from a single cell at 8:00 am with the goal of filling a 2 liter container by noon, at what point is the vessel half full? The answer is nowhere near 10:00 am but instead 11:59; consider that the vessel will be only 1/32 full at 5 minutes to noon. 

Thus West, in my experience, successfully resets the reader’s worldview, giving the current scattered and underpowered efforts at “sustainability” new urgency. “Continuous growth and the consequent ever-increasing acceleration of the pace of life have profound consequences for the entire planet,” West states in his conclusion. The rate of change “is surely not sustainable, and if nothing changes, we are heading for a major crash” of a sort we’ve never seen before (p. 425). Given the current lack of broad mathematical literacy (West’s book is equation-free but you have to grasp logarithmic scales), however, it’s hard to say who will read this and change his or her mind.

Before I started the book, I was thinking about scale in a different realm: representative democracy. Just as banks grew “too big to fail,” have modern democratic nation-states become too big to govern? Consider the House of Representatives: in 1804, a Congressman represented, on average, about 40,000 people. Today, the average California Congressional delegate represents 677,000 people: an impossible number. Both by objective counts of bills passed and opinion polls, the U.S. Congress is not working for its people. The European Union is hardly a robust counter-example; China is run, to the degree it has its successes, under completely different principles.

Thus the two scale cases reinforce each other in a particularly unfortunate feedback loop. At the time when science needs to be multidisciplinary, multinational, and theoretically rigorous and predictive (in a way the social sciences have rarely ever achieved), we have a system of governance that generally devalues science and scientists, balkanizes constituencies within and across nations, and accelerates rather than decreases energy intensiveness: think of how much energy is used to pump water to Los Angeles, or to cool Phoenix, or to power all the cars in metro New York. At the same time, changing climate means higher water levels, changing crop yield patterns, and new winners (ships that use the North Pole to cut transit time) and losers (many ski resorts are already rebranding as snowfalls decrease and become less predictable). If anything, as urgently and eloquently as he makes his case, West’s concern is understated to the extent that legislative gridlock, remove from constituents, and symbolic agendas render government largely incapable of leading (or even following) the discussion that needs to happen sooner than later.

Tuesday, May 30, 2017

Early Indications May 2017: Education for a world of bits + atoms


The physical world and the digital world are converging. 2-d bar codes and RFID tags serve as physical hyperlinks, connecting a physical object to its digital data heritage. Amazon merges deep data expertise with unparalleled physical logistics to promise same-day delivery in some areas. The so-called Internet of Things connects sensors on physical devices to powerful data analytic capability to increase fuel efficiency, prevent unexpected breakdowns, and smooth traffic flows. Waze, Uber, Airbnb, and many other startups are building new business models at the intersection of mobile apps and the cars, houses, and other attributes of private citizens.

The changes wrought by this combination of digital and physical attributes will be massive. My distinguished engineering colleague Monty Alger suggests two big buckets: 
1) Disruptive innovation, in which incumbents are attacked from outside their traditional competitive realm. Content industries are already in disarray: record labels from MP3s and streaming, newspapers from Google and Facebook, non-sports television from streaming, etc. Looking ahead, who knows what Google + Lyft could do to GM in time-shared autonomous electric vehicles, for example? 
2) New economies of scale and scope as digital expertise with big data and hyper-scale infrastructure encounter metal-benders and other traditional “atoms” companies. Again, recent history is instructive: Facebook photos vs Kodak; Google searches vs library reference desks; Uber’s 1.5 million drivers vs Kelly Services, the former with 3x the number of contractor/workers and 1/7 the longevity. 

We will soon see more companies or other organizations that dwarf the giants of the 20th century: AT&T, IBM, Exxon Mobil, GM. A recent example: last week Peloton achieved “unicorn” status with a $1 billion+ valuation. Its innovation is a $2000 connected stationary bike that lets people join unlimited spin classes (for $39 per month) from their homes. Note that Dorel, Trek, ASI, QBP, and the rest of the traditional bike world were absent from the effort.

How is this combined world different?

This world of bits + atoms has many characteristics that distinguish it from the world of only a few years ago. First, the speed of change has never been so rapid. Uber is valued at $69 billion (bigger than GM and Lockheed Martin, and about the same size as Goldman Sachs) after less than eight years of operation. Nokia and Research In Motion lost the global lead in smartphones, becoming irrelevant in less than five years after the launch of the iPhone. Instagram was acquired by Facebook for $1 billion in 2012, two years after being founded but with 200 million users and only about a dozen employees.

Second, innovations are building on each other; we now have a set of building blocks that can be reconfigured. Tiny high-resolution digital cameras, created by the millions for smartphones, can be used in game consoles, robots, or medical applications. GPS finds its way into everything from missile systems to social network games. The same machine learning algorithms that helped a Google company beat a Go grandmaster are being used to manage energy consumption in the parent company’s data centers.

Third, more and more domains look like Wall Street: sub-second algorithmic trading governs mobile ad placement, while high-speed digital markets are coming into play in energy markets and even at Amazon, where items can change price multiple times a day. (See this story.) Few companies are ready to manage such speeds and volumes of market information, much less to respond accordingly.

Skills shortages

The talent needed to invent, manage, and succeed in this world is scarce. The CEO of Symantec predicts there will be 6 million cybersecurity jobs in the global market by 2019, with a quarter of them going unfilled. (One scary implication: the already pathetic state of Internet of Things security will get worse.) Digital tools are becoming more prevalent at all levels of manufacturing, and many entry-level employees currently lack the skills to use CNC machines, 3D printers, and workflow automation software. Looking ahead, robot programming is beyond nearly all industrial employees, even engineers with graduate degrees. But given the speed of digital disruption, internships and entry-level hires often serve to perpetuate the skills gap rather than close it as companies recruit new workers who look a lot like the existing ones. “Big data” is a case in point: many job postings ask for a worker the existing managers cannot define, so buzzwords can replace concrete skills, software packages, and work achievements in the interview, with predictable results.

College curricula are lagging farther and farther behind. Pure bits people in computer science are doing well, as are students with pure atoms skills in domains such as hospitality management or fitness instruction. In the middle -- where many engineering fields, several business majors, and, increasingly, the hard sciences are being challenged to integrate physical and digital processes – several factors combine to prevent learning innovation to address the skills shortfall.

First, curriculum change is shepherded by the very people who succeed under the current regime. Time after time, bold changes get abraded and segmented in committee, and the outcome is incremental shifts to existing turf, fiefdoms, and reward systems. Second, in a research-driven university, professors with grants and publication hurdles often teach what they know and are investigating (push) rather than what companies are figuring out they need to hire (pull). Related to this mismatch, real-world challenges rarely align with academic silos, and interdisciplinary problem-based instruction is rare. Finally, many students bear responsibility for not digging deep or hard enough into their prospective fields of employment until an internship after the junior year, far too late to fine-tune the shape of a major, or to get mentoring in that field. If students don’t know what they need to succeed in the market, professors aren’t incented to know, and company recruiters screen for younger versions of themselves, the status quo gets perpetuated further.

What about STEM, you might ask? Parents, governors, and lobbyists from many corners of the economy hope to improve how the U.S. trains and measures its primary and secondary students. We have computers in more classrooms, Internet connections to most schools, and, slowly, teacher training to make instruction better. Certainly education in quantitative reasoning needs to improve to prepare students for the many complexities posed by the intersections of bits and atoms. But STEM programs, or even STEAM (including the arts) efforts, are necessary but not sufficient.

Because of the mixing of the physical and digital domains, and the increasing speed of change, skills for what some are calling “hybrid jobs” are proving to be in short supply. How many engineers can prepare a project budget that accounts for spare parts, currency effects, and price volatility in raw materials? How many programmers can manage a team? It used to be that only senior executives needed to learn cross-domain skills: marketers learned data analytics when a new division was acquired, engineers learned finance when they got promoted to run a line of business, statisticians learned interviewing techniques as they became responsible for instrument design, not just scoring. But no longer is it a mid-career transition to learn outside one’s educated domain. Now, even entry-level employees must cross disciplines, understand different cultures, and merge “soft” and “hard” skill sets.

Where in college or university does this happen? The notion of general education distribution requirements, in theory, should allow biologists to study materials science or architecture, or roboticists to study psychology. In practice, the combination of loose requirements, student convergence, and minimal incentive for interdisciplinary faculty collaboration means that students learn little outside their major. In one college of business, the two most popular general education courses were introductions to astronomy and to the history of popular music. You can guess the average grade earned in these courses.

Consider the field of robotics as a counterpoint. On a self-driving car team, in humanoid robotics, and in factory automation, the following disciplines intermingle:
  • computer science (architectures and programming)
  • metallurgy (what can we make this piece from?)
  • physics (lidar)
  • electro-optics (machine vision)
  • material science (batteries)
  • mechanical engineering (how can we make the bot strong and light?)
  • psychology (how do co-workers understand the robot? How does the robot signal its co-workers?)
  • math (path planning is highly complex).


That astronomy course isn’t doing anyone a favor in this marketplace.

What do we propose instead?

Let us begin with a list of objectives, then propose possible ways to achieve them.

1)    Education cannot be confined to the traditional K-12, 2-yr college, 4-year college, masters degree, Ph.D. model.
2)    Students (who can be defined as anyone with need and desire to learn) will need to learn or re-learn what they need close to when they need it.
3)    Students need to learn how to learn over a lifetime, how to express themselves in various ways (words, numbers, animations/simulations, videos, etc), and how to assess the validity and quality of both argument and evidence (classic critical thinking). That is, the core of a liberal arts education has never been more necessary.
4)    Students need both hard skills and soft skills, and the ability to tell when each is applicable.
5)    Students need to learn self-awareness, both to manage their contributions to a team and to frequently reposition themselves relative to workforce demands and opportunities.
6)    Students need to learn the differences between skills advancement in a particular domain and skills integration across domains, and how each is to be achieved.



The ultimate solution to this wish list is a decade or two away. In the interim, we propose as a starting point a model of integrated education that uses student internships as a test bed for industry-student-university collaboration. (Details are being worked out this summer.) Going forward, we foresee a model of integration moving across university departments and even colleges, in line with the robotics example noted above: when companies seek skills rather than majors, the university will have to respond with new course design and delivery models, new skills certification practices, and new funding streams. Those are outside the scope of the initial project, but we feel it is important to position the integrated internship as the initial step in a long, fundamental transition that more adequately supports the need of a global economy, our students and graduates, and the employers who will hire them.