A Social Science Agenda for the Tech Economy
This is a talk I gave in Fall 2017, at the Work2017 Conference in University of Turku, Finland
My job at Intel is, at its most basic level, to support foundational product and business innovation by conducting, and translating, social science research to technologists.
Why? Left to their own devices, engineers, like the rest of us, draw on the cultural and institutional practices of their training – as a famous sociologist would say, they are normatively isomorphic. That’s a fancy way of saying that they usually want to build something cool.
I currently work as an “advanced strategic planner” in the Datacenter Group (DCG), another odd position I had no idea existed. My work today is focused more at the intersection of social science, technology, and strategy (not design). And my work is more concerned with competitive threats, transformational ecosystem changes, and basic business model innovation. There are more acronyms and code names than you ever imagined existed. My team validates the LZ requirements and tech readiness for core platform features to pass through the POP on their way to PRQ.
In this talk, I’d like to accomplish three goals. The first is to say something about the movement towards software and data, among technology firms in the global economy. Sociology was formed in the crucible of modernity, the shift to industrial production and urbanization that characterized the turn of the 19th to 20th centuries. We are now living though a change of arguably similar importance. We are in the soup. Let’s talk about what we should be doing to understand, and as importantly, to influence that change?
My second goal is to say something about the ways that social sciences are being mobilized at Intel. I want to emphasize the path from insight to being ‘finished’, because it is very different from my world as an academic. Many academics find working with industry frustrating; and vice versa. Maybe I can help you to understand why that is.
And possibly ambitiously, I would like to end with a dialog around the ecosystems for social scientists. There are high-uncertainty, high-risk, high-reward opportunities for you to do meaningful work outside of our traditional locations.
For those not familiar with Intel, it is a global company of just over 100,000 employees, producing the processors that make your devices work (in the case of laptops and desktops), or else make your devices useful (in the case of data centers). It earns roughly $50 to $55 billion dollars of revenue a year, with gross profit margin of roughly 60%. While I and my colleagues worry about and debate its future, this is a very successful company. I want it to remain so.
Intel manufactures “things,” and the things we manufacture form the physical substrate of the information age. Intel is a semiconductor manufacturing company, where, quite remarkably, we make computers out of sand. The processes by which we accomplish this feat are sophisticated: lithography, ion implantation, etching, insulation, electroplating. We take sand, purify it into cylindrical silicon ingots, slice those ingots into wafers, and then draw impossibly-small, architecturally-specific designs on them. Those designs form the basis for transistors. The transistors in turn draw power to run the calculations that make your phones, computers, appliances, city infrastructure all work.
This is pretty much magic at any resolution, actually; but as the transistor size becomes smaller, the processes become increasingly complicated. For example, at some point, light becomes too “thick” to use for just your normal impossibly-small lithographic process, so you need to find ways to narrow it, first with mirrors, then by firing it through drops of water. Or through molten metal. You may hear things like “extreme ultraviolet lithography.” At Intel, “Moore’s Law,” that is, the observation that the number of transistors you can fit on a chip will double every couple of years, is literally our founding narrative. And yet, despite the sophistication of product, these manufacturing processes and their attendant organizational practices remain rooted in a decidedly industrial logic.
The economy is shifting, and the dynamics are impressive and scary. It is comparable to the transformations around industrialization. Intel sits at this intersection of ‘old’ and ‘new’ as the material substrate of the technology economy.
But here is the problem: the basis for value creation in the global economy is undergoing a tremendous transformation, and I am beginning to suspect that we: a) under-appreciate this shift; and b) don’t really yet understand either the contours of this shift, or its effects.
We have entered a new era, what my colleague Tony Salvador calls the Digocene Era. The evidence for this transition is striking. At the individual level, we are seeing a splitting off of opportunity structures at the top end of the employment ladder, particularly in the US, that is fairly staggering. At the organizational level, there is a new concentration of wealth along monopolistic, technologically-networked vectors (Google and Facebook regularly scoop up 90% of the growth in advertising, for example). At the economy-level, whole industries (economies!) are now built on the constellation of cloud, mobile, and digital, in a way that has transformed the lives of a sizeable majority of people on the planet.
As the tech analyst Horace Dediu has noted about mobile, 2 billion people are using Facebook every day. Our mobile devices are looked at for 2 hours a day; unlocked 80 times a day; holding all our memories, conversations, secrets. These are, “epoch-making technologies. They shape the fiber of society and the definition of quality of life. They obsolete entire economies and change the balance of political power. They shift the center of gravity of society.”
Apple’s iOS ecosystem is set to pass US$1T in revenue this year. And the iPhone is 10 years old. WeChat and its 700 Million users is 5 years old. AWS, Amazon’s public cloud, is 11. Facebook (1.7 Billion users) is 13 year old. Google will turn 20 years old next September. In the US, Google’s not old enough to drink! It is hard to understate how far, how fast we have really come over the past two decades.
This “era” is increasingly marked by a particular cluster of characteristics: first and foremost, a confrontation with things that are “digital” and new monetization of these digital products. Think about the iPod. It is not simply a music player. It created a top to bottom re-orientation of music, shifted relations among industry players, created (and cratered) massive amounts value across and outside the music industry itself; changed the outlets for and expectations of artists and a host of attendant policies. Something happened when we moved to CDs and MP3s that didn’t happen when we moved from vinyl to cassettes.
In the face of this transformation, what is it that ‘old’ economy companies, rooted in manufacturing and industrial organization, should do to be successful? This is the fateful question for me nowadays.
Because is not just technology companies that are embracing, or at least confronting, these changes. IBM’s strategic initiatives, most notably its Watson platform, continues to move that company further away from manufacturing hardware, and towards artificial intelligence. Intel is redefining itself as a ‘data company’, moving us into artificial intelligence, machine visioning, and so-called “smart” manufacturing. GE is building what it calls a ‘digital twin’ platform, turning physical objects into digital simulacra, applying analytics to those objects, and monetizing them. Goldman Sachs, as investment banker-ish as a company can be, has recently, and with a straight face, declared itself to be a technology company.
Central to this data-fication, and this is the second piece of the puzzle, is platforms. Platforms in the technology economy can be built upon, allowing users to iterate and experiment in the marketplace at extremely low cost. They are not necessarily new: Intel’s X86 is a platform; operating systems too. But the new constellations of mobile, cloud, and “everything as a service” is accelerating the reach of platform economies. Some of the characteristics of platforms that we are beginning to recognize, include concentration of winners and losers into an 80/20 structure; strong ecosystem, or network, effects; and a particular form of highly- concentrated capitalist accumulation.
The iPhone, and the whole “app economy” did not exist 10 years ago. That’s $30B/year just in App Store transactions, not even including the future apps, services, industries built on that innovation platform. Today, something like 70% of online purchases in China are made via mobile devices. AirBnB, Uber, Snapchat, Instagram (and its accompanying economy – we have been studying women on the Arabian peninsula making money selling things through Instagram) – these companies don’t exist without mobile, without cloud, and without a set of platforms to bridge the two.
Platforms drive massive benefits for the companies that orchestrate them, and headaches for even entrenched industries. A vivid example comes from financial services, an industry arguably protected by high regulatory and capital-requirement barriers to entry. Well Fargo has used these barriers to create multiple, profitable services: lending, credit, wealth management, retirement, home mortgages. This is banking!
But platforms in these spaces – AWS to build, but also securitization as a service, credit rating as a service, identity as a service, even current-accounts as a service – are now cheap, easy, and able to plug into your mobile offering as a set of APIs and pay-as-you-grow business models. This is allowing start-ups to compete on every single one of these value propositions.
These startups are not competing with Well Fargo “as a bank.” Virtually none of these startups is a bank at all! Like piranhas, they are picking off the most lucrative value-added value propositions, turning incumbent banks essentially into big dumb pipes. The largest digital payments processor in the US? Starbucks. Incumbents are struggling existentially to keep up.
What should social sciences look like (or could they look like) in the era of Tech economy? This is not a replacement for primary knowledge creation, but a plea for more kinds of social science. Our work requires more engagement, and that requires new ways of thinking about our work.
What should the aims of social sciences be, as we attempt to understand, critically assess, and importantly, to engage with the technology industry?
Sociology, like many other social sciences, was formed in the crucible of modernity – the systematic rationalization of a manufacturing society, the domination of concentrated urban living, the rise of the individual in society. Understanding that shift and managing the consequences of those dramatic changes arguably formed the working basis for much of the social sciences over the past century and a half. What we are seeing today is a transformation no less momentous than that of modernity. I say this with the love of a former insider, but the critical eye of a now-outsider, it appears that the social sciences are not keeping up. What are the disciplines that are forming in the face of digital?
Let’s start with what social science looks like at Intel. You may find it unimpressive! What I hope to emphasize is that the path from being clever to being finished is different from what I experienced as an academic. If social scientists are going to have impact on the unfolding of the technology economy, if social scientists are to be more than critical near-historians of the changes now upon us, perhaps it is worth knowing what that potentially looks like.
Some things are predictable: the deadlines are sharper, hierarchies are sharper. I manage my internal- and external-facing criticisms differently and more carefully than I did as an academic. I also learned early on that “interesting” is something of a dismissal – or rather, my professional colleagues would take “useful” over interesting 100 times out of 100. That’s a marked shift from academic life, where interesting gets you published and my ‘practitioner’ work has been dismissively described as ‘rather material’.
I now work in a more collaborative environment, across a range of disciplines that is largely impossible in academic settings (physicists, software developers, financial analysts, hardware architects, marketers, lawyers). My work is not owned by me; but that also removes many of the inner demons that seem so often to afflict my academic colleagues. Also, the menu of what is possible to get done, in a $170B company, is larger.
As I alluded to earlier, my job nowadays is to conduct research on people, organizations, industries, or ecosystems. And then to make that research technologically tractable to decision makers and engineers. You are not done when you report your findings. I want to give two examples of what that looks like.
The first example is what I spoke about at the outset, the insight that we are moving from a world of “model” to a world of “measure.” Ubiquitous, cheap sensors; readily available analytics; and changes in the natural environment toward more stochastic environmental risks; these are all creating possibilities for companies to move from measures based on proxies (“model”) to those based on more real-time measurement.
We sharpened the argument, and then did an ecosystem analysis to see how broadly we thought this trend might extend. And reported those findings. But that did not make us done! We combined the insight with a set of technologies around distributed ledgers, to see how and where Intel might participate in, or at least not be blindside by, this shift. This requires us to have an implicit model for institutional change.
But that did not make us done! We ID’d an industry and use case where we might validate the business hypothesis.
And finally, we architected a technical solution that doesn’t ‘solve’ the problem per se, but moves Intel towards a space with more options, more flexibility. The models sitting behind this are all sociology: Ted Porter’s quantification, Lis Clemen’s institutional change, Padgett and Ansell’s “robust action.” This is sociology in the wild.
A second example comes out of work that was effectively the reason I was actually hired into Intel – so this pulls us back to 2013, and the work here was already in motion with amazing colleagues: ken anderson, Brandon Barnett, Tony Salvador, John Sherry among others.
We were investigating the possibility of moving us from personal data silos, based on a strong advertising model, towards a more democratized and horizontal-ized personal data economy.
In this case, the work was to identify the opportunity – people are mobile but their data is not – and then to ask whether Intel could catalyze a new ecosystem. The research here was to identify the factors related to transformation, and to systematically determine which of these factors were cause, which were effects, which factors would take care of themselves, which factors had no incentives for anyone to solve them. The four main ones we landed on with were: digital trust, data literacy, digital infrastructure, and data openness.
But we were not done yet. We worked with a series of companies, and the White House, to launch a national day of civic hacking, to see whether there was interest in this personal data economy not just from us, but in the broader tech ecosystem. Were we trying to make our own weather, or tapping into a real shift? As part of this, we helped to launch half a dozen companies.
Finally, we identified and validated these findings against what Intel’s strengths and opportunities might be, including a series of M&A targets that we could pursue.
As I hope you are gathering, this is a world away from academia: identifying a research topic, framing a question, doing the research, publishing the findings. We need our social sciences to train people to do this kind of end-to-end work.
All of these projects are being created by technologists, engineers, business people, designers. And they are being created in absence of a significant presence of social scientists (not to mention ethicists, legal scholars, artists, writers, and poets).
And the truth is, I don’t work with academic social scientists for the most part. Much of this is institutional mismatch – my timelines are shorter, uncertainty higher, I don’t generally publish, the company walks away from projects all the time. Also: I’ve ‘sold out’; my work explicitly presses the interests of a for-profit multi-national corporation; the work is shallow; the research skims the surface of “real” sociology.
Being a sociologist at Intel is like being a dog in space. On the one hand, there is no gravity: you are responsible for making a career that is often undefined, indescribable, largely non- transferrable. On the other hand, you are in space! It’s amazing!
I will end by saying that for the critical work of the social sciences to inform those who are building the technology economy, you need to recognize three things. First, pointing out problems without pointing out solutions is simply a nonstarter. If you are not taking the risk of proposing actionable alternatives, you will most likely be ignored. Problems with not solutions is not a viable path.
Second, connect the research to use cases, design principles, market opportunities, and if possible even specific technology implementations. If you want software people to use your findings, learn how software developers work! Speak the language of agile, turn your findings into user stories, learn how they might incorporate your findings into actual development cycles.
Finally, you need to consider moving into these positions, and to define these positions as you take them. This is perhaps uncomfortable news for many of you, but I am not suggesting that sciences abandon their work to the interests of industry! Some of you should, though. It would be good for you, good for us, good for the social science ecosystem more broadly.
Technology firms speak have a natural inclination to see technology solutions to technology problems. But also technology solutions to social problems. What is often needed are socio- technical solutions: solutions linked to ecosystems, institutional practices, human needs.
Looking ahead, we are seeing digital assistants, massive increase and density of networks, augmented and virtual realities, artificial intelligence, as a cluster of technologies that are going to have tremendous transformative social effect. The pace of change is not slowing; it is getting faster.
These are socio-technical developments. And the socio-technical skills needed to understand, but more importantly, to shape these technologies, have not been as important since the birth of the Internet. It’s time to get started.