On Sony CSL Kyoto

March 25, 2020

Disclaimer: Opinions expressed here are solely my own. Please refer to the Sony CSL Kyoto website (which should go live in about a week; the site was scheduled to be up by now, but the company’s internal review process is moving slower than usual due to the global crisis) for official information about the lab.

While the news has understandably been crowded out by the ongoing pandemic, Sony Computer Science Laboratories Inc. has recently announced the launch of a new branch in Kyoto (Sony CSL Kyoto). As one of its founding members, I’m scheduled to start splitting my time between Tokyo and Kyoto starting later this year. At this point, not much about the lab has been made public as we’ve yet to begin operation; we’ll be updating our website with more information once the global crisis subsides (hopefully soon!) and things get back to normal.

With an initial research staff of only three scientists, Sony CSL Kyoto is a small operation. The modest scale does not mean a lack of ambition, however. The launch of the lab is a culmination of heated internal discussions that have been taking place in the past several years, about both the future of the company and the future of research.

Sony Computer Science Laboratories was established in 1988, a bygone era when the Sony Walkman was the best-selling portable music player, the World Wide Web had yet to arrive, and the Japanese economy was the envy of the world. The company (a fully-owned subsidiary of Sony Corp) was founded by Toshitada Doi, a renowned engineer behind many of Sony’s commercial successes, and Mario Tokoro, a computer science professor who gave up his tenured position to lead this new enterprise. More than three decades have passed since then — both founders are already long retired, our hallways are filled with fresh faces with no memory of the early years, and various circumstances surrounding the company have changed beyond recognition. On the occasion of the company’s 30th anniversary in 2018, we began a series of fun, all-hands meetings where each researcher gave a presentation “reimagining” Sony Computer Science Laboratories; if we were newly launching the company now instead of in 1988, what would it look like, and what would be its operating principles? Below I will share some snippets of discussions that had arisen during these internal meetings, which ultimately led to the creation of Sony CSL Kyoto. (Note that none of this is official information; what follows is strongly colored by my own opinions and biases.)

Many of our discussions revolved around how the ways in which new technologies are created and deployed to market have changed drastically in the past 30 years. Following successful examples set by institutions such as Bell Labs and Xerox PARC, Sony Computer Science Laboratories was set up as an independent research outfit within Sony Group that is organizationally separate from teams engaging in product development, sales, marketing, etc. We were to conduct freewheeling research unconstrained by short-term commercial interests, whose outputs would be disseminated mainly in the forms of technical reports, patents, prototype demonstrations, and academic publications. The expectation was that such outputs will quickly percolate throughout Sony Group, leading to tangible products and services — i.e., a “trickle down” theory of innovation. In other words, despite existing within a corporate structure we were expected to play a role similar to those of universities and other academic research institutions: as a fountain of new, original ideas that inspire and influence, rather than directly engage in, commercial product development and other more practical endeavors.

The trickle down model works, albeit at sporadic rates of success, and is likely the only viable model we have for much of basic scientific research. However, as computer science matures as a field and the majority of research efforts move from basic to applied research, we’ve begun to see an increasing number of corporations reject the uncertainty of this model, instead adopting “hybrid” models of innovation where the boundary between research and product development is deliberately blurred. The entire process of technological innovation, from initial conception to commercial deployment, is reimagined as a long-term, continuous pipeline.

Such hybrid models may at first appear to reduce appetite for risk-taking, but under close inspection that does not necessarily seem to be the case. We now live in an age where VC-backed startups routinely pursue high-risk endeavors, from the development of self-driving cars to space exploration, often with compact, tight-knit teams taking charge of the full pipeline from research to product development. At the same time, many academic research labs around the world (perhaps as a result of increased competition and misaligned incentives) have turned themselves into “small-idea factories” that churn out papers, patents, and YouTube videos of quick, often incremental projects, optimizing for publication metrics and media hype than the pursuit of fundamental breakthroughs.

We believe that a new corporate lab for the 21st century should operate on a carefully designed hybrid model; it should push its staff to take on substantial scientific and technological risks, while adopting a holistic approach toward innovation where researchers understand that their responsibilities as creators of new technology do not end with publications or demonstrations. The lab should function as a “safe space” where bold experimentations with uncertain outcomes are not merely tolerated but strongly encouraged (incentive mechanisms must be calibrated to ensure this), with the expectation that successful outcomes of such experimentations should be followed by fully dedicated efforts toward real-world deployment.

Another set of discussions centered on the simple notion that success breeds complacency. Sony Computer Science Laboratories has always had a strong independent streak. Its original staff was an elite squad consisting of seasoned academics, engineers, and other professionals who had firmly carved out their own positions in their respective fields; all had joined the company searching for an environment where they could escape the tyranny of conventional evaluation criteria (e.g., publication metrics) that they viewed as flawed or one-dimensional, and finally be true to their own ideals. Each researcher had, through their years of experience, a clear vision of what exactly they would like to spend their time working on if given complete freedom, which the company afforded.

Ever since then, the company has generally rejected external ideas of what constitutes good research, instead opting to rely on (often intense) internal discussions to gauge the value of each other’s work.

The autonomy granted to the company by Sony Corp, which was extended to its cadre of researchers, made for a fertile ground to produce original research. Although the company has by choice stayed a small and tight-knit organization, it has managed to build up a unique reputation within the computer science research landscape (and more broadly, the scientific research landscape; the range of research pursued within the company has grown over the years). Several of its researchers and alumni can rightfully be called rock stars in their fields. However, history shows that an institution, when removed of the need to listen to others and emboldened by past success, can easily grow out of touch with reality. (Although I won’t go into details here, I can easily name a number of research labs/groups that I feel have fallen into this trap!) Personally, I do not see any evidence of such decay within our company, but the subjective impression of an insider is not exactly a reliable measure.

As a new branch of Sony Computer Science Laboratories, Sony CSL Kyoto will operate under the same tenet of independence. Internal discussions will continue to take precedence over external metrics. However, being a brand new lab in a brand new city, in terms of reputation we feel we are essentially starting from a blank slate. We’ll need to once again prove our worth to society and build up our reputation — and in the process, open ourselves to tons of outside judgment. In a sense, the creation of Sony CSL Kyoto is akin to a self-administered litmus test for our company, intended to see whether we are still as relevant as we believe ourselves to be, or have been unknowingly turning into a shell of our former self. My bet is on the former, but I’m also certain that the experience will force us to make a number of tweaks to both our values and operations, modernizing ourselves to better meet the demands of this age.

Finally, our discussions also often touched on the idea that digital technology, through its monumental ascent over the years, has strayed far from what idealistic engineers and researchers had envisioned in the early days of computing. Though such sentiments may already sound rather trite these days, as a company founded in an era where the future of digital technology was generally viewed with greater optimism it is hard for us to curb the feeling that modern technological advances, despite all their benefits and virtues, have produced a series of unwanted side effects on our individual and collective well-being. Fake news, smartphone addiction, omnipresent surveillance — a number of societal ills have arisen, or been exacerbated, as results of advances in digital technology.

At Sony CSL Kyoto we wish to assemble a diverse, international team of researchers with a range of disciplinary backgrounds, and explore alternative paths of technological development that are truer to human values, cultures, and traditions. We hope to create a home for modern-day idealists, reminiscent of the company’s early days in its optimism, but with a more sober outlook grounded in 21st century reality.

As mentioned earlier Sony CSL Kyoto has yet to begin operation (we’re shooting for an opening date of late May or early June), and as of now we are heavy on aspirations but light on concrete plans. How exactly we can implement all our ideas is still largely up in the air, there will surely be many challenges. But I’m optimistic about the future, and looking forward to my role in this exciting new project.


YUICHIRO TAKEUCHI is a Toronto-born, Tokyo-based computer scientist whose work explores the intersection of digital technology and architecture / urban design. Currently he works as a researcher at Sony Computer Science Laboratories Kyoto, and also directs the nonprofit Wikitopia Institute. He holds a PhD in Informatics from The University of Tokyo, and an MDes from Harvard Graduate School of Design.