Early Indications April 2019: Review essay: Deep Medicine by Eric Topol

Note: I neglected last month to announce that MIT Press has published my book on 3D printing,  and that an article on some managerial implications of same came out in the Journal of  Organizational Design.  ***** Eric Topol is a cardiologist who is very visible on the front lines of the medical field as it intersects  with smartphones, distributed sensors, and other emerging technologies. His most recent book  looks at "how artificial intelligence can make healthcare human again," in the words of the subtitle.  While I have plenty of honest disagreements with Topol's book, I very much respect it, and hope  that it jump-starts some way-overdue discussions about the future of medicine, particularly in the US. Given that I don't work in medicine, I'm extrapolating my lay experience here. The phrase  "personalized medicine" conjures up, for me at least, micro-targeted pharmaceuticals: my dose of  heart, anti-depressant, or chemotherapy medication will be fit precisely to my weight, age,  metabolism, and holistic health picture. Topol makes it clear that I am off base here. First, time  -- of day, of season, of life -- matters a lot: precision dosing entails understanding time-series  data in entirely new ways.  Second, precision medicine may likely begin not with pharmaceuticals but with diet. Recent  research on the role of gut flora and related participants in digestion helps explain why nutritional  research lacks any consistent consensus: meat was good, then bad, now it's sometimes good.  Butter was normal, then evil, but now it's not nearly as bad as margarine. Sugar was portrayed as  benign (so was cigarette smoking, for the same reasons), but now the "a calorie is just a calorie"  fiction is being exposed. One reason that dietary advice is so unreliable is that the degree of  uniqueness in our digestion and metabolization patterns is only now coming into focus. A key factor in that biomedical research into the gut biome and related processes is a massive  increase in the size of the data sets being researched. Much as with Facebook photos or Google  searches, machine learning algorithms need enormous volumes of training data to become reliable.  If for no other reason, Topol's book is valuable for chapter 11, where he explains the rapidly  changing literature of diet and digestion research. One point is made quite emphatically in the book, though I don't recall the author saying so directly.  Given the vast number of AI-infused startups for medical applications, we are nowhere remotely  close to general AI: every company he discusses, some of them in depth and/or with great  enthusiasm, is a point solution. Diabetes-related readings are the specialty of one company, while  mammograms are graded somewhere else, and retinopathy is diagnosed somewhere else. In short,  AI will not replace doctors anytime soon. This is not to say, however, that physicians' jobs won't change. Certain tasks -- image recognition  in particular -- are done quite capably by machines. Note, however, that image recognition is not  disease diagnosis (aided by up-to-date knowledge of a vast literature), treatment planning, end-of- life counseling, arguing with insurance companies, or any of the dozens of other things doctors are  called upon to do. The challenge ahead will be to team doctors and computers in the most effective  possible proportions, letting each contributor provide inputs that will lead to the whole patient  experience amounting to more than the sum of its parts. Topol also provides useful context for the role of AI in employment: there are plenty of subspecialties  in the US that are far understaffed, and globally, many nations have staggeringly few doctors. AI  provides the quite realistic possibility, in dermatology care for example, of enhancing family doctors  and nurse practitioners as they address one condition among many they see in a routine workweek.  Up-skilling generalists is entirely different from threatening the livelihoods of the relatively small  number of specialists called upon to address the major issue of deadly melanomas in the U.S.  and elsewhere.  Compared to a Watson-like digestion of medical research (something that seems plausible, if not  yet realized), for machine learning algorithms to read patient records remains a daunting task. Even  with electronic medical records, the lack of standardization in people's lives, descriptions of their  symptoms, health professionals' acuity in observation and notation, interoperability between  institutions (despite potentially using the same software package), and, again, the time of  observation means that mining EMRs is for practical purposes impossible for the foreseeable  future. In the meantime, EMRs are linked to physician burnout and decreased patient satisfaction  -- doctors navigating screens and typing with their backs to the patient in the exam room are  hugely troublesome for all concerned. Speech recognition powered by machine learning could  help alleviate this issue, and Topol ends the book by imagining a "super-Siri" personal health  assistant that helps each of us live our best life, medically speaking, through an advanced form  of what the military calls data fusion. For all the many drawbacks of current EMRs and other technologies, however, it's time to  address the elephant in the room: managed care. To the extent that my genomic, diagnostic,  behavioral, and other data becomes a tool for prediction of my health (and by extension, the  cost to maintain it), there will be a fundamental conflict of interest between me and my insurer.  Topol nods at this issue but never engages it (he is, however, a consultant to some of those  insurers). As the old saying goes, "follow the money," so my conviction is that the initial large- scale deployments of AI in health care will not be to reduce diabetic retinitis or arthrofibrosis,  but rather in a reimbursement arms race between insurers and providers. It's entirely possibly  that both Aetna and a health network will have deployed IBM Watson systems as the two  parties dispute a billed transaction, so we may see algorithms at war with themselves. It seems far fetched, but perhaps the EMR + AI convergence will be the beginning of the end  of the U.S.'s unique model of health care, in which more is spent per patient than anywhere  in the world while most major outcomes rank well down the list. That is, the fundamental conflict  of interest noted above will lead to such complex statutes, regulations, and/or case law that the  current U.S. arrangements become unsustainable. Already far too much effort is expended in  paperwork that does nothing for patient care, and the implications of predictive algorithm-driven  medicine for privacy, doctor-patient confidentiality, and ethics make HIPAA look positively  antiquated and inadequate.  If Topol can envision a new "virtual medical coach," perhaps the American Medical Association  can help construct a 21st-century alternative to its previous objections to single-payer health  care. Such a design for medicine would acknowledge the new roles of genomic science, data  analytics, robotics, and machine learning, along with addressing end-of-life, opioid, and other  political hot-button issues. I'm not convinced AI can give physicians time and tools to become  "human again" (beware anyone selling a golden age fallacy), but medicine absolutely needs  to become individualized, beginning from a data foundation, for the first time.

Early Indications March 2019: Facebook Act 2?

On March 7 Facebook founder, CEO, and majority shareholder (for voting purposes) Mark Zuckerberg announced “A Privacy-Focused Vision for Social Networking.”  You might want to read it, if only to see that what he does NOT say is blindingly obvious.

There have been many commentaries on this missive, but I want instead to discuss a post from last April by Andreessen Horowitz’s Benedict Evans, called “The Death of the Newsfeed.”

There are many dots to connect here, with rather staggering implications. Let us begin.

1) Zuckerberg states that people value privacy, so Facebook will build on its WhatsApp franchise and offer end-to-end encrypted 1:1 chat: “I believe the future of communication will increasingly shift to private, encrypted services where people can be confident what they say to each other stays secure and their messages and content won't stick around forever. This is the future I hope we will help bring about.“ 

But wait: bad people can discuss and plan bad things in private, so Facebook has 

“a responsibility to work with law enforcement and to help prevent these wherever we can. We are working to improve our ability to identify and stop bad actors across our apps by detecting patterns of activity or through other means, even when we can't see the content of the messages, and we will continue to invest in this work. But we face an inherent tradeoff because we will never find all of the potential harm we do today when our security systems can see the messages themselves.” 

Whoa. Either communications is private or it is not.

2) Facebook stopped many copies of the New Zealand massacre video from being reposted, it is true, but there were still hundreds of thousands of copies that were circulated and of course recirculated. Facebook should not be congratulating itself on how much evil it has prevented any time soon. Recall that AT&T shared its bulk network traffic with US legal/intelligence sources via the infamous “Room 641A” in San Francisco. How will Facebook provide similar courtesy/compliance? Will Indian, or Iranian, or Russian governments get the same access? Whatever Facebook says about privacy, how much will educated observers believe? To whom is Facebook held accountable?

3) Even though Mr. Zuckerberg is the only Facebook shareholder whose vote counts, he has a fiduciary duty to the other investors. Not once does he mention what happens to ad revenue in these encrypted channels. If it goes away, does Facebook attempt to become the AT&T of the 21st century? If Facebook users stop being the product sold to advertisers, they likely become customers of a very different kind of service, one for which I predict there is limited paying appetite. 

4) Why is Facebook re-examining privacy given its steadfast dismissal of same? The Benedict Evans post from last year provided a useful prediction, in nearly Haiku form: 

"All social apps grow until you need a newsfeed

All newsfeeds grow until you need an algorithmic feed

All algorithmic feeds grow until you get fed up of not seeing stuff/seeing the wrong stuff & leave for new apps with less overload

All those new apps grow until..."

The massive scale of Facebook means the average user (a dubious notion, but I’m merely quoting their figure) is eligible to see at least 1,500 posts per day. If published figures are true — I’ve seen extrapolations of 35 minutes, 27 minutes, 41 minutes — that means a user would have to click once per second for 30 solid minutes merely to stay current with the newsfeed (30 minutes x 50 clicks/minute).

Enter the machine learning solution: we can algorithmically determine what people want to see out of that impossible pool of 1500 eligible items. Alas, algorithms can be gamed, or stuffed (with echoes of the SEO arms race), or lag user sentiment, which can be expected to evolve over time. Hence the “pivot toward privacy” can be read at one level as an admission of defeat: machine learning cannot curate the newsfeed model indefinitely (or even temporarily).

5) This is an important moment. As industries from CPG to defense to pharma to transportation seek out consultancies and new hires to help embrace “big data” and “AI” (whatever that is in operational terms), an acknowledged leader pivots away from its core AI-driven product. The fact that Facebook is seeing an exaggerated departure of top executives of late underscores how much this matters: having made their first (or at least most recent) x million dollars inside Facebook, these people who are in a position to know are betting they can make their next x million _outside_ the social network leader.

6) Building on messaging is not Zuckerberg's idea. Chinese services such as WeChat, Viber (popular in former Soviet satellite states), and LINE (Japan/Taiwan) can include payments, gaming, GPS, and, yes, encryption. Consultants including Forrester have been predicting the rise of messaging for about five years. One must give Facebook credit for its acquisitions: Instagram and WhatsApp have proven to be brilliant buys. But can/will they cannibalize the original product?

The larger question involves revisiting Clayton Christensen's Innovator's Dilemma from 20 years ago: can Facebook shift its reliance from the revenue-rich but overbloated and mistrusted newsfeed toward lighter-weight and potentially ad-resistent messaging tools? If so, how will Wall Street value the company? Can Facebook leadership persuade investors, regulators, and users (both advertisers and consumers of the service) of the wisdom of moving with the platforms to new definitions of relevance, convenience, and trust? What happens to the future of news media, given Facebook's role in the current crisis of an informed body politic? Whatever happens next, this feels like the beginning of the end of an era.

Early Indications January 2019: The organizational context for machine learning

Much attention is paid to the necessity for innovation, typically at the level of a technology: a molecule, a machine, a circuit. While these types of innovations matter, they cannot succeed without institutional “scaffolding,” as it were, the organizational infrastructure that gets the molecule made into pills and dispensed through pharmacies or gets the electric car assembled, distributed, recharged, and maintained over its entire product life. We posit that this organizational infrastructure will be a critical factor in the adoption of a powerful new technology in the coming decades.

We are at the cusp of a period of broad adoption of machine learning in more and more technical, professional, and commercial fields. While the labor shortage of data scientists and employees with related skills is well documented, less appreciated is the need for the institutional scaffolding mentioned above. Three factors have contributed to the boom in machine learning since 2012 or so: larger scales of training data sets, improved algorithms, and broader availability of powerful computing capabilities, often through (or at) cloud providers including Google and Amazon. The need now is for business models, institutional processes, and deep functional knowledge to be coupled with the technical advances.

This has long been the pattern of technological innovation. Powerful steam engines drove the need for bigger, stronger ocean vessels; retrofitting schooners and other wooden sail-powered craft with the new engines was insufficient. The rise of the automobile would have stalled without consumer credit, adequate paved roads, and a network of gas stations. AT&T’s Picturephone never achieved market acceptance as a landline device, but video calling apps such as Skype and FaceTime are extremely popular as smartphone add-ons. Technological innovation without organizational infrastructure rarely succeeds.

Within an organization, this cultural and organizational scaffolding is often invisible, making it difficult to change. A classic example is provided by the transition from steam and water power to electric motors in textile mills and similar facilities in the late 19th/early 20th century. Historically, water wheels and then steam engines powered overhead drive shafts that ran the length of the facility. Individual machines were thus located in relation to this overhead power source. Electric motors were initially used to drive the same shafts, slowing adoption: the case for new investment was weak. It took roughly 30 years for “group drive” to be replaced by “unit drive” in which individual machines each had their own electric motor(s) on board. Cost savings accrued as unused machines no longer were driven by the centralized power source, energy efficiency was improved with the removal of slippage in the belt-driven transmission system, and variable-speed motors allowed for each machine to run at its optimal RPM. Productivity improved dramatically as factory engineers repositioned machines, now freed from the proximity to the drive shaft, to facilitate workflow. In addition, removing the drive belts and overhead shafts allowed for the installation of overhead cranes, improving productivity further.

As machine earning is brought into mature organizations (as opposed to algorithm-centric companies such as Google or Facebook), managers will likely be tempted to overlay the new technology onto existing business practices. In the US, cellular phones were initially treated as mobile versions of stationary voice devices. In countries without a well-developed voice-device network (most of the world), such practices as texting took off much faster: the capabilities of the new technology were unconstrained by previous preconceptions and habits. More recently, 3D printing is often compared to mass-production molding and milling processes with which managers are familiar, and not surprisingly, 3D printing is found to make little business sense for long runs of simple geometries.

For machine learning to be applied to problems and processes where it can add distinctive and unprecedented value, it will need to be understood in the context of a particular market, a particular business model, and a particular business process. This level of understanding requires a rare combination of deep functional expertise, an open mind, adequate historic data for training the algorithms, and technical acumen. (Many efforts we have seen focus only on the latter aspect.) Weather forecasting — where IBM can claim deep expertise — has been reshaped by new algorithms and models (including ensembles) over the past 20 years, but medical reasoning has proven much more difficult to automate. Ad placement has been transformed by machine learning; production planning remains variable, volatile, and largely manual.

What kinds of scenarios are we seeing as machine learning is adopted in legacy business processes?

1) Unquestioned assumptions can be challenged by a naive algorithm

Because many more variables can be correlated across much larger samples than humans can cognitively manage, seemingly random inputs can be found to drive outcomes. At one hotel booking site, photos of guest rooms with the window shades open drove higher sales, but only in specific circumstances. In other scenarios, images with the shades closed performed better. One size rarely fits all, and bigger data sets can help identify which sizes fit which customers under which constraints.

2) Machine learning can cheat, often creatively

In one image-recognition application, the algorithm was supposed to be identifying sheep when it was in fact using grass as its decision tool, so a photo of a green field with rocks was scored as containing sheep. A race-car simulator found that the application’s physics engine did not penalize crashing, so the car would careen around the track, bouncing off the mid-point on the barrier wall of the straightaways. An algorithm designed to create locomotion simply built a tall, unstable tower that fell forward; a refinement allowed the tower to somersault for greater speed.

3) Knowledge of ground truth matters more than ever

As algorithms will be trusted to do more and more activities either unattended or with minimal human oversight, people will need to spot errors in data, logic, or translation into managerial levers. We have all been served online ads that are irrelevant, offensive, or otherwise a waste of the advertiser’s money. The “flash crash” of May 6, 2010 that shut down the US stock market was caused by a confluence of mischievous human trades and runaway algorithms. Our heavily digitized automobiles illuminate “check engine” lights that are as often a sensor error as a mechanical fault.

4) Ethics needs to catch up to the technology

Google, Facebook, and other machine learning leaders are realizing the potential for algorithmic harm and have funded both internships and full-time positions for ethicists. The costs are only beginning to be counted: of the speed and scope of algorithms that cannot be understood at human scale, or the “cheats” that may not be immediately obvious (as with the sheep/grass image recognition glitch), or the migration of computing from beige boxes (from which electronic money could be stolen, for example) to physical devices in people’s homes that both invade privacy and control door locks and other safety features.

These are in fact exciting times: machine learning has incredible potential beyond spam detection, credit card fraud monitoring, and energy efficiency gains. For these benefits to accrue, however, managers, investors, and institutions of training and education need to recognize the need for the organizational contexts, from budgets to risk mitigation, in which the technical gains will be situated. This path from recognition to action will take decades, be marked by many false starts, and unfold with an often frustrating sequence of relearning lessons that have already been taught elsewhere.

Early Indications December 2018: 5 Weak Signals

2018 was a challenging year for much of the tech world, and not only because of the depressed stock market. A number of high-profile failures taught several lessons:

• maintaining momentum (and revenue growth) at vast scale is truly difficult and ultimately impossible: nothing can grow forever.

• finding hard but solvable problems is tougher and tougher, in part because so much has to go right to win in billion-person-sized markets.

• whether in Moore’s law, pharmaceutical drug development, or social software, after you solve the initial hard problem, the road only goes uphill and gets rockier. Apple delivered the iPod then iPhone/iPad, and that may be the extent of their core market. (Cars? Healthcare? I’m not holding my breath.) In hardware, Intel entirely missed the mobile- and smartphone segments after dominating the desktop, and Nvidia is seeing a major slowdown in sales as bitcoin miners and self-driving cars are not providing the updraft the company expected. 

Here are five weak signals I'll be watching in 2019 and 2020.

1) Hardware robotics lacks key knowledge of how people understand machines

Over the course of 2018, three high-profile robot companies shuttered operations. Kuri, a social robot built on a platform Bosch hoped to commercialize, was priced at $700. Jibo, founded by MIT affective robotics pioneer Cynthia Breazeal, had raised upwards of $60 million and was designed to sell for about $1,000. Finally, the biggest fish in the pond of dying robot companies was Heartland, later renamed Rethink Robotics. It was founded by MIT AI/robotics superstar Rodney Brooks and had raised $150 million to make user-friendly (no safety cage required) light industrial robots that could be simply programmed and work amidst people. These so-called “cobots” (collaborative robots) called Baxter and Sawyer were aimed at small manufacturing and shipping facilities that needed an occasional helping hand: flexibility and adaptability were the bots’ calling cards rather than power, durability, or accuracy.

In the case of the household bots (Kuri and Jibo), the functionality was largely achieved by Amazon Echo and Google Home smart speakers at 5-15% of the price. Jibo employed a complex and expensive 3-axis motor system to give the table-top device an emotionally appealing “body language.” It was billed as a companion but couldn’t keep a shopping list, told jokes but couldn’t shoot decent photos, and joined other prominent crowdfunding projects that over promised and under shipped. Jibo started out global and was retrenched to North America only. And so on. (For those who are interested, here’s an excellent post-mortem on the Jibo.)

In all 3 cases, figuring out what people value in a robot, how they value it, and what they will give up in exchange still appears to be beyond the state of the field: even deep-pocketed giants including Sony and Honda have found consumer robotics to be an impossible mass market to crack. In part, engineers tend to design robots that can’t be produced at sufficiently low price targets. Further, it’s unclear how many people want to bond with an electronic device: even the Roomba, a relative success, does not remotely approach Apple/Android sales volumes. In short, it’s not at all clear what widely shared problem Jibo solves that Alexa can’t.

2) Social media giants quite simply do not value their users’ identity and privacy (in the US especially)

Every few weeks, Facebook is found to have done something that demonstrates a complete disregard of users as people. Twitter continues to condone personal abuse, often by troll bots, while presenting “innovations” unrelated to the service’s core mission. Google+ was a privacy disaster (here’s a good article), starting with a “real names” enforcement that did not follow the core concept of Google’s own Circles product, that people relate to different other people using different identities: Adam in a classroom, Adam the book author, Adam the scout troop leader, Adam the poker player, Adam the church deacon, Adam the AA member. Internationally, Alibaba’s Sesame Credit was for a time connected to the larger Chinese governmental social credit initiative, and it’s possible that Alibaba’s cloud computing subsidiary is involved with running the latter given that it frequently serves as a contractor to governmental agencies. 

If a US citizen travels to Europe and logs on, meanwhile, Google helpfully presents all manner of privacy tools that are either hidden or unavailable stateside. (I have no idea what Facebook logins look like in either the US or EU.) Google leads US corporations in lobbying spend, and it’s easy to see why, given the larger dynamics at play in the ad/media landscape. It’s staggering how little people know or comprehend about what’s happening: students regularly assert that “Facebook sells our data” when a) that’s not literally the case and b) the truth is that Facebook largely constructs then monitors and markets these students’ digital identities, which is something else entirely. 

There is no analogy in economic history (“data is the new oil” is wholly insufficient) that adequately explains or even suggests what is happening, and Google is right to be worried about regulation that could be written ham-fistedly and have many unintended consequences. At the same time, who outside of maybe EFF is lobbying on behalf of the users? Someone on Twitter made a great point about poor people who use Facebook as a utility: white elites who step off (Walt Mossberg being the latest) are in a very different situation compared to many millions of Facebook users, given their many sources of social capital. I’m not at all sure what _should_ happen here, but am watching all the same.

3) How long can Netflix abandon the long tail for being a neo-Disney hit-maker?

Long ago and far away, Netflix rented every possible DVD, satisfying customers’ need for choice and curation (“people who liked X also liked Y”). Beginning about ten years ago, with streaming supplanting the mailed discs, Netflix began investing vast sums of largely borrowed money on original content. All those old movies people used to rent went into mothballs, to be replaced with Netflix originals and a few select chestnuts to be trickled out then pulled back. If you or your children want Disney titles, meanwhile, those will all likely be gone in a year as Disney shifts its content to its own streaming platform. 

Netflix is no longer an online Blockbuster Video; it’s competing head to head with Disney and Viacom not only for viewership but increasingly for creative talent. Longtime Hollywood writers and producers are getting contracts from Netflix in nosebleed territory: Shonda Rhimes (responsible for hits including Grey’s Anatomy and Scandal) landed a production deal reported in the $100 million range. 21st Century Fox lost Ryan Murphy (Glee, 9-1-1, The People v. O.J. Simpson); Netflix gave Murphy a $300 million deal, according to whisper numbers. Finally, Netflix has spent still more money on production infrastructure in both Los Angeles and New Mexico. It’s a simple question: how long can Netflix borrow the money to pay more people to build more content to satisfy its viewers who live in more and more (and more diverse) countries every year?

4) Who will pay for 5G?

As 5G wireless rolls out in the coming years, the confusion will be considerable. AT&T is calling its 4G improvements 5G Evolution, which is not true 5G. 5G is not backward compatible with 4G, so users will need new equipment. What this equipment might include will be interesting. True 5G features higher transmission speeds and lower latency; gamers should be very excited. It also requires more antennas because the cells are smaller. Thus stationary broadband (home and small business) looks to be an opportunity for wireless providers led by Verizon and AT&T to steal market share from wired broadband providers including Comcast and . . . AT&T. 

Besides watching Netflix and playing Fortnight, what else is 5G good for? True enterprise-grade broadband is frequently mentioned as a prerequisite for “smart cities” in which everything from water mains to traffic lights to drones and other surveillance cameras can be instrumented and eventually programmed to perform better. Traffic gridlock alone is often mentioned as a substantial opportunity.

There are many opportunities, some sinister (either China’s social credit scoring or enhanced police and government surveillance, possibly without adequate oversight, here at home). But for our purposes, the more relevant issue was illustrated this summer during the California wildfires. Verizon was throttling a firefighting team that had exceeded its monthly wireless data limit, the fire chief linked the behavior to net neutrality, and the PR hit to Verizon was significant. The more important point for our purposes is that state and municipal governments are cash-starved: property tax receipts drop every time Sears closes a store, costs for health care rise every year, and longer life expectancies mean longer pension payouts to retirees. While enhanced gunshot location, faster infrastructure repairs, and other sensor-driven municipal scenarios sound appealing, there is a real question lurking: who exactly will pay for the bandwidth and related equipment to make them happen?

5) Last year it was Bitcoin, this year it’s supply chain blockchain

Back when I worked in consulting, there was a saying: “in mystery there is margin.” Conversely, once enterprise IT buyers figured something out for themselves, its price dropped rapidly into commodity territory. Thus we see consulting firms spending lots of money, time, and effort on “thought leadership” built for the express purpose of being able to charge for something the client doesn’t fully understand.

In 2019, when more people realize that “enterprise blockchain” is a modern implementation of a distributed database, some of the irrational exuberance may die down. Yes, faster invoice reconciliation, more granular product recalls, and less fraud are benefits, but those benefits do not accrue because blockchains are built from some kind of magical fairy dust. Note that IBM’s definition of blockchain never once mentions the relationship of the new kid to the grandfather:

https://www.ibm.com/blockchain/what-is-blockchain 

*****

There you have it. 2019 will be an important year for many tech companies to stabilize revenue growth, deliver meaningful innovations, and get their operational houses in order. Of course we will see surprises (Uber and Lyft’s IPO valuations?) but there don’t seem to be any hot new startups to track right now: the old order seems to be relatively stable, with Dell, IBM, and HP (among others, obviously) needing to reinvent themselves for the age of mobile device supremacy and eventual commercialization of AI. I don’t see a Netscape, a Salesforce, a Google, or a Facebook on the horizon, which probably spells trouble in the long term — where are the new innovations going to come from? — but makes for a relatively calm landscape in the short run.

Speaking on a personal note, I hope the new year brings health and happiness to each of you, along with your loved ones.

Early Indications October 2018: More

This month’s news cycle featured news that a single ticket matched the numbers necessary to win a $1.5 billion prize. South Carolina, where the ticket was purchased, allows winners to remain anonymous, but as they say, good luck with that. I’m struck by the famous study by Brickman, Coates, and Janoff-Bulman that found that lottery winners, shortly after the event, were less happy than people who had recently become paraplegics. While it’s impossible to do a controlled study, the number of lottery winners who have made a complete mess of their lives is long.

My point here has little to do with lottery excess per se. Rather, I’d like to connect the fascination with “more” to the digital age. I can’t claim either uniqueness to the US or global applicability, but it’s easy to see examples of mismeasurement: specifically, we use numbers that are easy to derive to compare things that are much more subtle. Clayton Christensen addressed this tendency in How to Measure Your Life. At the end of my life, how good a parent was I? How careful a scholar? How effective a citizen? How inspirational and responsible a leader? Numbers don’t measure any of those particularly well.

Distributing computing broadly among the human population, then networking lots of it together, feeds this tendency. Back in its early years, Google posted the vast number of webpages it had indexed. Facebook friend counts distort many aspects of true human connection. Amazon boasts “earth’s biggest selection.” Having our fingers on the screen for hours every day is changing us and our kinship in insidious ways. Infinite availability of gossip, shopping, sports talk, or anything else is probably not a long-term win, but humans are easily hacked and we absorb this stuff under the guise of self-determination even as we are being powerfully and constantly gamed by any number of bots and psychological techniques.

Robin Dunbar famously hypothesized that a typical person can maintain about 150 meaningful relationships, Facebook’s counter notwithstanding. Our attention spans, memories, and dexterity are finite, yet screens, game controllers, and other information flows are overwhelming us. The NYU academic Clay Shirky proposed that we don’t have data overload; rather, the technologies of information filtering are insufficient. That may be so, but in any event, our technologies are altering us, often in the name of “more” rather than “better.”

Where does this lead? There are relatively tiny efforts, parallel to the Slow Food movement, to resist the “more” of data: fewer numbers, better chosen, reflected on over time can be incredibly powerful. By analogy, firehoses can be great for extinguishing burning buildings but they’re worthless as nourishment. Vast selection, whether in selection of mates or of toothpaste, ultimately doesn’t enrich us: Jim Gilmore (author with Joe Pine of the Experience Economy book 20 years ago now) memorably told me that people don’t want infinite selection; they want what they want. Restaurants with thick menus rarely inspire confidence among foodies: it’s impossible to master crepes, lasagna, and stir-fry under one roof.

The food analogy begins to lead us to the main point: software may be eating the world, as Mark Andreessen asserted in the post of that title, but curation wins over sheer volume. Netflix delivered vast selection of DVDs in its first decade and for all the star ratings and recommendation engines, its share price remained flat. Now that Netflix delivers far fewer titles, and more of them are optimized for its viewers based on the combination of algorithmic analysis and in-house production, its investors are exuberant, the company’s debt habit notwithstanding.

In their book The Second Machine Age, Eric Brynjolfsson and Andrew McAfee of MIT posit that humans and robots, teamed together, can perform better than either computers or people alone. Curation seems to be a task that largely awaits this approach to take hold more broadly: digital platforms love the scale of automated curation but stumble on the execution. Facebook is efficiently algorithmic but prone to being manipulated, as the 2016 election proved; Mark Zuckerberg and his senior management appear to be struggling with the human side of the content selection issue. Google image search had to stop offering results for some queries after machine learning presented an image of people of color as gorillas. Amazon is replacing some of its category management (demand forecasting, price negotiation, and ordering inventory) with algorithms, freeing retail white-color workers for new tasks. If the company performs well this holiday season, it may validate the experiment.

Where else might we look for joint computational-human curation as an antidote to the overwhelming flood of information coming at us? I would hope the education segment takes a leadership role but see few efforts in that direction. Booksellers are making a comeback for many reasons, recommendations prominently among them. Spotify hasn’t mimicked Netflix in production of original content; I can’t speak to the role of curation in its success to date but it doesn’t seem to outperform a good DJ. Pandora’s origins in a “music genome project” (automated curation, begun in 1999) did not translate to market dominance. Choosing a vacation destination, a retirement portfolio, or a job candidate is still not well enabled by software + people; lots of error and wasted effort persists in these systems. 

Part of the reason for this slow progress is a skills deficit: the number of people who can understand a domain of knowledge (art history, auto mechanics, arthritis) then can (and want to) translate that understanding into software-friendly form is extremely small. Jeff Bezos, a Princeton electrical engineer by training, is rare in this regard and unique in his entrepreneurial translation of that skill: Amazon Web Services, the Kindle, Alexa/Echo, and the company’s supply-chin practices all derive, I would argue, from similar roots in being able to bridge the gap between physical practice and software instantiation thereof. 

Going forward, perhaps we will see software tools, possibly variants of IBM’s Watson suite, that ease and accelerate human + computational information management. Perhaps a new computer language, cognitive model, or interface will enable the aforementioned domain experts to amplify their expertise to a larger audience. Or perhaps we will continue to tread water, barely staying afloat in an inexorably rising tide of noise.

Early Indications September 2018: Why can’t your organization innovate like DARPA?

Following up on last month’s newsletter, which asked who was going to generate the big-scale innovations required for a growing world, I recently read Sharon Weinberger’s 2017 study of DARPA entitled The Imagineers of War. The effort was well worthwhile, if only for the nuanced explanation of the origins of the Internet. Paul Baran’s famous mesh diagram of a survivable communications architecture for command and control of the US nuclear forces is only part of the story. J.C.R. Licklider, a key figure in the history of human-computer interaction, was also involved: he saw far earlier than most of his contemporaries how having access to computing can change how we think, so his connection of computing to psychology was significant.

More generally, psychology was a hot area of defense and intelligence research in the late 1950s and 1960s, in part because of the interest in and fear of “brainwashing,” that is, some form of mind control. The bestselling book The Manchurian Candidate was but one manifestation of this fascination; Stanley Kubrick’s still-brilliant Dr. Strangelove also captures many aspects of the era with searing insight: mine-shaft gaps, purity of essence, and of course Peter Sellers’ brilliant one-sided phone conversations with his Soviet counterpart on the hotline.

Thus we have in some ways come full circle as the Internet was successfully used for psychological manipulation by Russian entities and surrogates in the 2016 election. History in this case did more than rhyme. For all the historical interest I had in the book, there are some concrete lessons: for all the attention DARPA gets for its successes — Stealth aviation, GPS, drones — the organization is likely to spawn few imitators. I posit that there are at least seven reasons for this.

1) Your organization isn’t motivated by defense of American interests

Being charged with preventing future surprises like Sputnik means that very little subject matter is out of bounds. Budgets are much bigger than private industry typically can mobilize. Patriotism can motivate devotion and behavior that cannot be simply hired.

2) Your organization doesn’t pursue enough whackadoo ideas

When it was first proposed, stealth aviation did not sound much more plausible than ESP between mother rabbits and their bunnies (the former were thought to know when harm befell the latter, even at a far geographic remove), telepathic spoon bending (accomplished by none other than Uri Geller), or antennas mounted on elephants to aid in radio communications through Vietnamese jungle foliage.

3) Your organization has short time horizons

Notwithstanding the common critique of corporate focus on quarterly numbers, even a 5-year plan is sometimes too short. GPS took 20 years between theoretical proposal and first satellite launch. ARPAnet was more than 5 years in the making when one of four connected computers sent the message “lo” to a second unit (it was supposed to be “login” but the system crashed), and the World Wide Web launched another 30 years after that.

4) Your organization can’t bury failures as secretly

The author of the DARPA book noted in the endnotes that classification has obscured both successes and failures from being publicly viewed. A box of documents related to a James Bond-like jetpack remains classified many decades later. DARPA hasn’t undertaken an institutional history since 1975, when it was about 18 years old.

5) Your organization has more moral prohibitions

ARPA was deeply connected to the US war in Vietnam, attempting to use everything from fortune-tellers and soothsayers (who were paid to predict a Communist defeat) to Agent Orange (one of an entire family of air-sprayed chemicals designed both to cut off the food supply and deprive the guerrilla forces of jungle cover). GE proposed a mass galvanic polygraph to be used on entire villages. John Poindexter’s Total Information Awareness project, mass surveillance with little human or institutional oversight (AI was supposed to protect privacy), ran on DARPA money. Somewhere in the DARPA robotics budget there most likely exists a human-out-of-the-loop autonomous robot with lethal capability.

6) Your organization has more conventional hiring processes

Related to 2), Darpa has a long history of hiring rogue, unconventional, or outlandish individuals, then giving them long leashes. In the age of post-Sputnik fear, one Greek physicist proposed creating a defensive shield of high-energy electrons trapped above the earth’s atmosphere in the magnetic field. Multiple nuclear explosions were detonated high above the earth’s surface to try to validate the concept, which of course did not work in practice. One former DARPA director held the biannual agency conference at Disneyland not once but on multiple occasions. Program managers in the parapsychology field held the beliefs and credentials you might imagine for such a post.

7) Your organization has some nominal and procedural objectives

DARPA rarely gets too specific on what its mission and objectives are. Over the institution’s history, they have ranged from investigating counter-insurgency to counter-terror to space-based weapons to lots of secret stuff the public can’t see. Regimes have been supported with cash, expertise, and other assets, scientists (and psychics) have been funded, and numerous contractors have been generously enriched. Most of this activity is only loosely connected to an overall remit. 

As Weinberger notes in her conclusion, “the dilemma for DARPA is finding a new mission worthy of its past accomplishments and cognizant of its darker failures.” (p. 371) After failing massively in trying to win the hearts and minds of Vietnam’s people with as few US ground troops as possible, and after 30 years of a bi-polar (US-USSR) world order, the age of Al Qaeda and related entities has proven more difficult to fight with technology. After tens or hundreds of millions of dollars of investment, for example, the best weapon for fighting against improvised roadside bombs remained . . . dogs’ noses. Very few of our civilian or even DoD organizations would be allowed to spend so much and come up with so little.

Early Indications August 2018: The Next Big Innovation?

Few books have stuck with me the way Geoffrey West’s Scale (reviewed here last summer) did. I don’t fully buy the book’s argument for the applicability of natural scale laws to human structures such as cities (here’s a much smarter review than mine), but he did put the planet’s projected population in sharp perspective for me: worldwide, 1.5 million people will be moving to cities every _week_ for the next (now) 34 years. West argues, plausibly in my view, that we will need step-function innovations on the order of the Internet to feed, employ, cure, and transport all those people.

 

There’s a quasi-debate running between several economists and management scholars. Eric Brynjolfsson and Andrew McAfee at MIT argue that human organizational structures have lagged, as they historically do, technological development. Robotics per se doesn’t put people out of work; rather, corporate, taxation, labor law, education, and other structures don’t yet create a place for these new machines and humans in a larger, functioning economy. On the other side stands Tyler Cowen of George Mason, who says that we have harvested all the “low-hanging fruit” (his words) and that compared to the 20th century, our era’s record of groundbreaking innovation is thin.

 

All three views hang together in my mind: we are due for another massively important innovation — including in the “rules of the game,” as it were. Since the iPhone launched the age of mass smartphone use 11 years ago, it’s hard to find truly important ideas: Uber and Airbnb are both about 10 years old, as is blockchain (in which China now leads the world in patent applications ), which has yet to solve a truly important problem. Autonomous vehicles, meanwhile, are looking like less of a near-term bet (as recent news from Waymo illustrates). What am I missing?

 

Before looking ahead, let’s look back and see where the last few world-changing innovations came from:

 

-The Internet began at DARPA (in 1969, ARPA) but key components including the World Wide Web came from elsewhere (Europe’s CERN, in the case of the WWW). AT&T famously passed on the contract to build the Internet, because their substantial expertise in the existing circuit switched regime made it clear the technology would never work.

 

-Malcolm McLean owned a North Carolina trucking company and died worth about $350 million. His innovation? Containerized shipping: in 1956, when he piloted the idea, hand-loading a ship cost $5.86 a ton. Containerization dropped that to 16 cents per ton. “Globalization” and all that implies, including increased standards of living in many locales, rely heavily on his invention.

 

-Norman Borlaug earned a PhD at the University of Minnesota then spent most of his professional life in Mexico, cross-breeding crops. He has been credited with saving a billion people from starvation and won the Nobel peace prize. His so-called (by others) “green revolution” was critiqued from several angles: input-intensive agriculture made seed, fertilizer, and tractor companies rich and famers indebted. Large-scale farms (including road-building and other infrastructure) destroyed cultural practices and institutions associated with subsistence faming. Pesticides and monoculture have negative long-term environmental effects. All of that is true, but feeding a billion people who most likely would otherwise have starved deserves a healthy dose of credit.

 

Thus we see an entrepreneur, an individual humanitarian, and large-scale government agencies all making decisive contributions. Absent are corporations: yes, the Toyota Prius is 20 years old, but it has not (yet?) shifted the global auto industry off of fossil fuels. Even pure electric vehicles rely on a power grid that most likely begins with the burning of gas, oil, or coal. The great innovators of the past — GE, HP, IBM, AT&T, Xerox — no longer pack the research punch they once did. Innovations at Facebook, Google, Netflix, and Amazon are heavily tilted to the realm of consumer behavior, in which advanced algorithms are used on the relatively easy task of manipulating purchase and viewing patterns, one person at a time. 

 

What about big Pharma? In an age when economic rationality means $500 Epi-pens and 5,000% price increases on off-patent drugs (see Shkreli, Martin), it’s hard to see the sector as currently constituted solving a capital-B Big human challenge. Meanwhile, as antibiotic-resistant bacteria get tougher to combat with every passing month, it’s not impossible to imagine that penicillin and its offspring may not matter (or matter much) 100 years after the drug’s discovery in 1928. As science cracks the code of the biome, particularly regarding the gut, entirely new modes of treatment may become feasible. If (very broadly speaking) the 19th century was the dawn of surgery, and the 20th belonged to the birth of entirely new categories of pills, perhaps we will see the potential of genetics and related science realized for the remaining 80 or so years of our century. There’s no guarantee the Pfizers and Mercks of the world will be the relevant parties for these to-be-built treatment modalities. Recall that Sports Illustrated did not launch ESPN, nor did Sony introduce the iPod.

 

Zooming back out to the larger issue of the innovations required for the planet we are rapidly populating, two key questions will have to be answered: 

 

-what kind of organizational structures will help envision and develop ways to feed, move, educate, and/or employ large numbers of people?

 

-in what domain will the truly big innovations reside?

 

It’s easy to perceive the trajectory of history as moving upward: higher standards of living, as measured by money. Longer life expectancies. Farther reaches of sea and space explored. I was reminded today, though, that part of a 9-billion-person planet will be doing with less: less animal protein, fewer square feet of housing per person, less social mobility in a given country even as the broader population does better on the whole. 

 

Thus the new innovation might be in the arrangement of social order: both the limited-liability joint stock corporation and republican democracy are human inventions (as are human slavery, dictatorship, and monarchies). The next big thing might be “social technology,” designed to organize large numbers of people, along with their wants and needs, just as we saw with the pre-Reformation Catholic church in Europe, or Pax Britannica from about 1815 to 1914. (Technology matters a lot for these social arrangements, as witness the printing press’s role in the decline of the former or the place of steam power in the latter.) Alternatively, there might emerge some innovation as we more traditionally define it: technologies to move people in cities, process and distribute nutritional protein, or teach people how to earn a living. In either case, time is getting short: according to United Nations projections, global population will hit the 8 billion mark in about 6 years.