Tuesday, December 06, 2011

China Will Overtake the US in Computing…Maybe, Someday…

[note: the following is a rough draft -- I appreciate comments as I work this into shape and add relevant links to further sources]

December 6, 2011

Today, The New York Times published an article by Barboza and Markoff titled “Power in Numbers: China Aims for High-Tech Primacy.  This article echoes frequently expressed alarmist opinions that China is posed to take over the world.  I have lived in Beijing for the past 2.5 years as a visiting researcher at Microsoft Research Asia, I've taught Computer Sciences classes at Tsinghua University, and it is my opinion that China has major obstacles to overcome before becoming a high-tech powerhouse. The biggest of these is the the way creativity is discouraged in Chinese classrooms. Chinese students who spend time at western universities do pick up these skills. Creativity and the inclination to challenge norms in disruptive rather than incremental ways are at the heart of computing innovations. These traits are all but absent from Chinese universities. A solution I pose is an initiative called World Lab. We need a place for people from various cultures, backgrounds, and countries to come together to take risks in designing new technologies and to train students to become global leaders.

Today's NY Times article by Barboza & Markoff, “Power in Numbers: China Aims for High-Tech Primacy,”  would lead you to believe that the title of this blog entry (“China Will Overtake the US in Computing”) is almost a certainty. I could not help reading this somewhat alarmist article without cringing, as it follows a pattern of reporting on China that I’ve seen from since before I moved to China in 2009 and that I have noticed more frequently over the past two years now that I’m more sensitized to the realities of China’s economic rise. This lack of subtlety and nuance on China is what I’ve come to expect from media outlets such as CNN and I am more surprised to see it from seasoned journalists who are respected for their expertise, Barboza for reporting here in China and Markoff for reporting on computing.

As I prepare to leave next week to return to my position at the University of Washington, I am starting to reflect on what I have learned in my 2½ years in China. My own view is that there is incredible potential in the computing field in China – this is one of the many reasons I chose to pick up my family and move here. At the same time there are many important barriers to China’s eventual rise in computing and these barriers will not fall on their own without efforts at reforming both the educational system and government regulation, let alone certain Chinese cultural norms that are thousands of years old. That is why I’ve subtitled this blog entry “…Maybe, Someday…”. That is, I don’t believe China will rise above the US in computing anytime soon and if it is to do so, several important changes must first take place.

In the rest of this article I’ll try to touch on 1) why am I qualified to even have an informed opinion on China’s rise in computing, 2) what I saw as the misconceptions or omissions in the Barboza & Markoff NY Times article, and 3) what I think China must do to reach its potential in computing and why I think this is a good thing and not something the West should be worried about.

Who am I to Comment on Chinese Computing
As I read the NY Times article I was a bit surprised by some of the folks they had used to comment on the state of Chinese computing. I started to think, “who are the proper experts on this topic?” Later as I pondered this question, I began to think I’m as good an expert as anyone, at least from the academic computer science side, to comment on the rise of China in computing. Why is that? 

I have spent 2½ years living in China and in that time I have: worked at Microsoft Research Asia (MSRA), the top research organization in the country, taught at Tsinghua University, the top computer science department in the country, and organized several major technical research events in China. Before coming to China, I earned my PhD at Carnegie Mellon University (CMU), one of the top computer science departments in the world, earned tenure at Berkeley (another top department), founded a start-up, ran a ubiquitous computing research lab for Intel, and served as a professor at the University of Washington (another top computer science department). More detail on my background is here. I think this experience puts me in good position to make an informed assessment on computing in China. You be the judge. I’m sure I’m not right on everything and these are just my opinions, but after two years I’ve seen quite a bit, talked to many people, and I’m starting to have a good feel for what is going on here in China.

What is Wrong with the Rising View
I believe there is no question that China is quickly rising in all endeavors, whether it is in terms of China’s economics, infrastructure (think ports, highways, freight railway, and high speed rail), education, science, or technology. It is an amazing sight to see firsthand and the energy one feels living here during this important time in history is quite incredible (more than even in Silicon Valley during the 1st Internet boom of the mid to late 90s). Computing is no different from these other areas and China has made huge strides in 20 years, as reported in the NY Times article.

The key questions to ask are 1) where is China with respect to the US and the West in terms of computing today? and 2) where will China be in the future? The impressions that were given by the NY Times article on both of these questions is where I most felt the article lets the reader down. Let’s cover each of these in turn.

Where is China Computing Today
 Academic computer science has been the underlying basis for many of the major commercial strides in computing in the US (e.g., the Internet, the graphical web browser, compression for wireless communications, cloud computing, speech recognition, web indexing and search, gesture and touch-based user interfaces, location-aware computing, etc.).

China has made big strides in academic computer science over the past 20 years in terms of expansion of its programs and making a shift from mainly producing software for state-owned companies to undertaking leading edge computing research and education. In fact, China has passed major milestones in the past 5 years in terms of government support for research and in starting to publish in top computing journals and conferences.

Everything’s Big in China
Five to ten years ago, one would almost never see papers at the top academic computing conferences from China’s researchers, with the exception being papers from Microsoft Research Asia, which was started in Beijing back in 1998 by a group of Chinese and Taiwanese researchers who were trained in the US and worked in the US before returning to Asia. Today, there are many Chinese researchers who are publishing papers at top research venues. But, the number is still quite small given the large number of universities and researchers that are pursuing computing research in China. Computer Science & Technology is the largest undergraduate major in China and some estimates I’ve heard say there are over 1,000 computer science departments in China and over 1,000,000 computer science majors at a time across these departments. This is huge! The government is clearly making massive investments in computing.

Supercomputing isn't so Super?
One of the big accomplishments Chinese computer science has made given these investments over the last 5-10 years has been in Supercomputing: the very large, high speed machines often used for climate modeling, weapons simulation, etc. A couple of years ago China temporarily had the fastest machine in the world with the Tianhe-1A. This coveted spot on the TOP500 supercomputer list has traditionally been held by either US or Japanese supercomputers, though it changes all the time as new faster computers come into service.

Although getting to the top of the list was a major accomplishment for China, the news of China’s conquest of supercomputing really didn’t seem to be big news for almost anyone I know in leading computer science departments. Why is that? I think most leading computer scientists believe that although supercomputers are useful for certain problems, this is a technology of the past that will simply improve incrementally with underlying processor improvements (in fact, most supercomputers today use conventional processors used in desktop computers rather than the special purpose processors used in the past).

The big innovations in supercomputing have been in the programming models, network interconnects, and most recently in cooling/power usage. But, people seem to see much more important innovation going on in the cloud computing clusters that literally combine thousands of commercial processors together in standard racks connected with traditional networks in huge data centers around the world. This is the technology that powers Google, Microsoft, Apple, Amazon, and the many other web computing giants of the world and is then resold inexpensively to every little web site or mobile phone application that needs to do computing in the cloud. This type of architecture supports a far wider range of applications than supercomputing. Cloud computing is a hot topic in both industrial and academic computer science research and American computer scientists are clearly far in the lead in this area of work.

Academic Publications
In my own subfield of Ubiquitous Computing (ubicomp) and Human Computer Interaction (HCI), China is still in its early stages. Ubicomp has been around since 1991 and in those 20 years China has had almost no presence in the field (for example there were no papers from China at the 2010 UbiComp conference). This year I co-organized the conference with my colleagues at Tsinghua University and we held UbiComp 2011 here in Beijing (link). There were over 300 papers submitted and only 50 were accepted for presentation at the conference (a highly competitive 17% acceptance rate). Although this year we saw 38 papers submitted from China (last year there were only 10), only 3 of these papers with primary Chinese authors were accepted (and all of those were from Microsoft Research Asia). There were many US universities that alone had as many or more papers than all of China (e.g., Carnegie Mellon had seven and UW had four!).

This trend is very similar at other top computing conferences: China had almost no representation 5 or 10 years ago and now there is a smattering of papers (e.g., 1-3 papers/year – out of a 30 paper program – the last couple of years at each of the top systems and networking conferences: SIGCOMM, NSDI, and SOSP). Again, the majority of these papers are coming out of Microsoft Research Asia, not the top Chinese universities.

So we see China starting to be represented at major computing conferences, but Chinese researchers are at this stage no more impactful than many other smaller countries (e.g., France). Given the large number of universities and researchers pursuing computing in China, the interesting question is whether this a straight line that is going to continue its meteoric rise of the last few years (similar to China’s economic growth of ~10% for ten years) or is China’s impact in computing research going to start to grow at a much more modest rate (similar to many predictions of its economy growing at still fast yet more modest rates).

Research Creativity: Students, Faculty, & Academic Structure
Creativity, innovation, and “design thinking” have been some of the most overused buzzwords bandied about in the US business press over the last 3-5 years and this has especially accelerated in the few months since the passing of Steve Jobs. In computing research as well as in industry, creativity and innovation are also important topics. These hard to measure attributes are what we all believe lead to “impact”, which is also hard to measure, but is that which we are all after! Counting papers at top conferences or patents does not measure impact, but people (including me above) tend to sometimes fall back on this counting exercise, as it is easy to measure.

Having interacted with many top Chinese students while here in China, at both MSRA (the top place to have an internship for a computer scientist in China) and at Tsinghua (the top CS department), I’ve gotten a chance to observe the level of creativity and innovation in these top students. We’ve also attracted some of the top design students in China to our lab (in addition to hiring top designers from the US and Europe). I’ve also been lucky to interact with the top Chinese research computer scientists (i.e., folks who already have their PhDs) at MSRA and at the universities.

The simple fact is, the level of innovation and creativity in this cohort is much lower than in similar cohorts in the US. And in fact, the ones that are the best on the “creativity” scale almost invariably are folks who received their PhDs in the US/Europe or worked in the US/Europe. This is not to say those who haven’t left China for their education aren’t doing good work – as I mentioned above MSRA is one of the top places in the world for CS research and the researchers there are publishing at the top venues, but many of the most successful of these researchers have spent years under the tutelage of computer scientists who were trained in the West – almost going through a 2nd PhD while working at MSRA.

The simple fact is if you are educated in the Chinese system, from primary school through university, you have a much harder chance of practicing being “creative” than if you were educated elsewhere. This is not a genetic trait (as many Chinese educated in the West have clearly shown), but a trait of the Chinese educational system, which is based on over a thousand years of Chinese culture.

There are many articles (link) on how cultural underpinning of the Chinese educational system does a good job with the basics (e.g., the students in Shanghai beat the entire world on the PISA Test a year ago), but many here in China question whether the pervasive emphasis on memorization, test taking, and a cultural imperative that almost requires copying the teacher (link art article) and the past “masters” leads to a population that cannot think “outside of the box” (link).

Again, this lack of creativity is cultural and obviously there are folks who don’t fit the system and are creative and innovative (the art scene in China is growing by leaps and bounds). For many years, the top students in China have left the Chinese system for graduate school in the US. Although some of these students start out in America as brilliant and hard working students, many do not show much creativity when they start. They have learned not to question the professor, or others in positions of authority, and they are used to being told what to do rather than coming up with ideas on their own. But, many soon rise above this after a few years of practice and have turned into some of the top stars in the field (e.g., my own classmates at Carnegie Mellon, Harry Shum and Qi Lu, are now two of the top executives at Microsoft (links)).

I have personally advised students like this that have gone onto great computing careers, relying on their innovation and creative skills everyday. But this was only after 5-6 years in the “American” higher education system. My colleagues have often told me of similar examples. Now many Chinese are also aware of this key difference in our educational systems. The latest trend among middle class and wealthy Chinese is to send their kids to the US for their undergraduate degrees or even their high school education (some 200,000? were studying in the US this year alone link).

Now this trend by itself would cause one to believe that China will overtake the US in computing as this massive cohort of students return to China after earning their degrees. Although the “sea turtle” trend of returning to China after several years of working in the US continues, it doesn’t appear as common as some would believe. Many Chinese students become very accustomed to what is still an easier life in US cities and often choose to remain in America. In fact, a more important “glue” for these students might be the far more streamlined US corporate life (many Chinese companies are still fairly byzantine in their politics and structure and corruption is still a major problem). In fact, recent reports show that most wealthy Chinese are starting to secure homes and passports in the West, often for the educational opportunities outlined above, but also to avoid environmental degradation, corruption, and find access to healthcare (link report).

Last Spring I attended a major National Science Foundation workshop on computer science research collaboration with China (http://current.cs.ucsb.edu/nsf-uschina11/). Of the 80 attendees, over half were Chinese who were now professors at American universities. In computing research, many Chinese with US PhDs might be staying in the US for the prospect of working at a better university and with better graduate students than they can in China. Will this change soon?

One of the major differences I’ve noted between Chinese universities (and in fact Chinese organizations in general) and American universities is the power structure exposed in the academic hierarchy. American universities are hierarchical in that Full Professors make decisions about Associate and Assistant Professors, and Associate Professors in turn also make decisions (e.g., tenure) about Assistant Professors. But, I’ve also noticed that in the top departments I’ve been in that the more “senior” faculty understand that a lot of the innovation and best work occurs in the groups led by the “young” Assistant Professors and we in fact “protect” them so as to allow them to better develop and get this great work accomplished (e.g., we do not give them a lot of tedious committee work to do and we encourage them to teach advanced courses in their specialized areas rather than large, general undergraduate courses).

In Chinese universities, there is far more power and money concentrated in the hands of the senior faculty. In many universities the Assistant Professors are just that – they assist a senior faculty member and have no true independent agenda of their own. In a fast moving field like computer science, I believe this structure is bound to fail and cannot keep up with the changes in technology that occur so rapidly. Certainly more rapidly than the 10 years or more it will take a hotshot young faculty member to rise to the top of that hierarchy.

Today’s computing technology is nothing like it was 10 years ago! I believe this structural impediment makes it hard for anyone to name a computer science researcher in a Chinese university that they would say is one of the top in the world in their subfield (other than the few famous names, e.g., Andy Yao – a Turing Award winner, who have been “imported” to Chinese universities).

This means that unless the Chinese universities change this system, it will take many years (15-20) before their CS departments could even have a chance of being stocked from top to bottom with world-class computer scientists. And that would assume they start producing the top scientists here in China (which hasn’t happened yet) or start importing them from abroad (only a few have come so far). Again, this is not to say there aren’t good people here already. There are plenty of good people working in Chinese universities. For example, Prof. Yuanchun Shi, my co-chair for UbiComp 2011 from Tsinghua, is doing lots of great research in her group at Tsinghua. These folks are just spread thin and not a single Chinese computer science department has the strength of even a top 25 or maybe even a top 50 computer science department in the United States. This will be hard to change anytime soon without a massive change in hiring practices as well as in how those people are treated when they come on board.

Although academic computer science research in China isn’t yet all it can be and has some major impediments to its continued improvement, I believe the start-up scene is a bit healthier. Although I am not an expert on this, I try to keep up by following the top China tech blogs and writers on twitter (cite niubi, wolfegroupasia, tricia, kaiserkuo, affinitiy china, china law) and pay attention to what is going on at the key start-up events (e.g., TechCrunch Beijing was the most recent such activity).

I’ve also spent time chatting with and reading the works of folks who do study the start-up scene closely, such as Vivek Wadhwa (@wadhwa), professor at Duke and Stanford, who studies high-tech entrepreneurship in Silicon Valley and around the world. Professor Vwada has commented recently on the healthy start-up scene he has encountered while traveling in China (link). Noticing that this culture is starting to come to terms with the need to try and fail and start over again, as has fueled the amazing rise of Silicon Valley’s companies.

The conclusion I’ve come to from watching the Chinese start-up scene is that 1) it is vibrant, 2) some major early movers, especially on the Internet, e.g., Baidu, Alibaba, Sina, have already amassed fairly dominant positions in their niches as happened in the US (though we know as Yahoo has shown most recently that these positions can be lost easily), and 3) the amount of venture funding and number of startups are both increasing rapidly. 

In addition to these traditional spaces for innovation, there are other cool things that happen in China that are an outgrowth of its manufacturing innovation. In particular, the entire Shanzhai market (link), which started with fake name-brand goods, including phones and purses, has quickly moved into making novel new products. Again, they tend to be useful tweaks (e.g., multiple SIM card phones, new shapes, etc.) rather than major innovation. This might be where lots of the creative engineers end up in China as these types of folks may not have conformed with the rigid educational system to get into the elite schools.

There is innovation in the China computing startup world, but the type of innovation that happens in start-ups and in industry tends not to be the innovation that will pay off for the entire computing field in 10 years (e.g., the invention of the internet and many of the other computing advances I noted in the introduction to this article). Start-ups tend to take ideas that have already been floating around for a while and repurpose them to a new problem or incrementally improve on them. China’s start-ups are especially known for this incremental improvement strategy. As noted tech environmental crusader Peggy Liu (@shanghaipeggy) wrote today on Twitter, “China is not good at radical innovation, but it's great at tweakovation.” This quote exactly captures the type of activity happening most often in China’s startup scene.

This criticism for copying and tweaking rather than innovating is probably overblown, but continues to be said in and about the Chinese computing industry. One of the biggest names in China Tech funding, Kai Fu Li, founder of Innovation Ventures and former Google China Head, Microsoft Research Asia head, and all around Chinese high tech success story (from Taiwan), now has the nickname in China of “Start-Copy Li” (check for proper translation) for the propensity of companies in his venture portfolio to simple copy a popular western web site and give it some minor Chinese characteristics. For instance, there were hundreds of Groupon clones in China just a few months ago.

So although start-ups in China might be healthy, if a little less innovative than in the West, I do not think this is a fundamental problem for Chinese computing. The bigger question is can they really make the type of fundamental advances in the future that in the past led the US computing industry to its dominance. And can the Chinese make those advances if they are not first taking place in academic research. I do not believe they can and therefore encourage the Chinese to keep upgrading the educational system and infrastructure – but with more than just increased funding. I believe the structure needs to change (see below).

One argument for China’s future dominance in the fundamental underlying technologies of computing is the large Chinese patent portfolio. The NY Times article pointed out how China has overtaken Europe in number of patents filed and is catching up to the US and Japan. What the article fails to mention is that many, many people believe that many of these Chinese patents are bogus (link Vivek, China La blog) and come out of 1) a quota system that requires organizations to produce a certain number of patent filings per year regardless if they are actually any good and 2) a tendency to copy foreign patents, make minor changes to them, and then use these as trade barriers to western companies trying to do business in China (link China law blog). Leaving this type of information out of the NYTimes article really distorts the patent story. When paired with the lack of strong intellectual property rights protection in China, the patent story leads one to believe that China will not be able to innovate in the future.

How China Can Reach its Computing Potential
My analysis above might leave you with the opinion that I think China’s computing field is going nowhere fast. That is far from the truth. I think China will continue to improve in computing for two major reasons. First, computing in China will improve simply due to China's massive size: (1) in 1.3B people there are going to be a lot of great ones, no matter what barriers you put in their way and (2) the domestic market by itself will be huge and thus a great opportunity! Second, the large investment in technology research funding coming from the government (growth on the order of 10%/year for 10 years) will allow a lot of researchers to carry out many ambitious projects. But, I believe that instead of fearing China, we should see that China reaching its potential in computing could change the world in a very positive way and it is something we should try to help with.

China is Part of the Solution
Why do we want Chinese computing to succeed? I believe that the major problems that the US faces, the rest of the world faces, and China especially faces. China is key to helping solve these problems and by helping China’s research and education system in computing, we have a better chance of creatively solving these problems together. These are problems in:
  •  Sustainability: maintaining the environment, and stopping global warming in particular
  • Education: Improving education for all in both the basics as well as in creativity and innovation
  • Healthcare: Creating a healthcare system that will care for an aging population (North America, Europe, and China all suffer from this) as well as all one that will service all citizens at a reasonable price
All three of these problem areas will have solutions that involve government, policy, and pricing. Yet they also are problems where major technology innovations, especially computing technology innovations, can make a major positive impact. By working together with China on these problems we can help improve the world.

World Lab
In light of this view, I’ve been working the last few months on trying to create a new, multidisciplinary research institute that is jointly housed between a major Chinese university and an American university. This World Lab will become known as the place for risk taking, breaking the mold, inventing the future, and tackling the major problems facing the world. We will apply a new methodology I term “Global Design” to find a balance between design and technology, between human-centered & technology centered approaches, between academia & industry, and between Eastern and Western culture. The World Lab will push the boundaries of what is possible and invent the future today. This institute will help train the students and leaders of tomorrow’s universities and companies to be free thinkers who can create the solutions that society will need to solve these challenging problems.

I believe China’s rise in computing is remarkable, but the future is not assured. As a computer scientist I support helping China improve in computing because I believe it will help the world as well as the population of China. The problems are complex and success is not assured, but together I think we can create a better world.

Disclaimer: The opinions set out in this article are those of James Landay and do not represent the opinions of the University of Washington, Microsoft Corp., Intel Corp, or anyone else (unless they decide to say so – which I’d appreciate).

Acknowledgements: Thanks to Ben Zhao from UCSB (@ravenben) for some of the data on top networking and systems conferences. Thanks to Frank Chen (@frankc), Lydia Chilton, Aaron Quigley (@aquigley), Robert Walker, and Sarita Yardi (@yardi) for helpful comments on this essay.

My Background

Unlike other computing academics who have commented on Chinese computing, I’ve not just dropped into China for a week or two here or there and developed an impression. I’ve actually been living here full time for 2½ years. In that time I’ve helped build a new research group at Microsoft Research Asia(link), taught a course at Tsinghua University(link), co-organized a major international computing conference(link), started a major computing lecture series/symposium on new uses of computing(link), traveled to many different universities to speak, visit, and meet the students and faculty, and attended several meetings of the top computing faculty in China (a few of which also were attended by their US counterparts link: http://www.nsfc.gov.cn/Portal0/InfoModule_479/30695.htm).

I’ve also thrown myself into reading much of the press and blogs on innovation and start-ups in China and I’ve tried to go to events here in Beijing on these topics when I could. I also chat with others about these topics whenever I get a chance. As an expat you can easily meet some of the movers and shakers in this circle even when living in a city of 20M+.

In addition to my time in China, I think I’ve also been lucky to have been at the center of some of the top places in computing over the last 20 years. I obtained my PhD in Computer Science at Carnegie Mellon University (link). CMU is ranked by most as one of the top departments in the world. I was a faculty member and received tenure in CS at UC Berkeley (link), another one of the world’s top departments. Until coming to China, I was a faculty member in Computer Science at the University of Washington (link), another top department. At UW we’ve built one of the top programs in the world in Human-Computer Interaction and Design (link), which is a field that is at the forefront of envisioning and building the future of computing technology.

I also have industrial experience. In addition to the last 2½ years at Microsoft Research Asia, unquestionably the best computing research organization in all of Asia, I was the co-founder and CTO of a silicon valley-based start-up (NetRaker) while on the faculty at Berkeley and I ran a ubiquitous computing research lab for Intel in Seattle for 3 years (link). The researchers at the Intel lab invented many leading edge technologies in that time, including the city-scale, beacon-based location capabilities that were originally found on the iPhone and every single smart phone since (link), activity inference technology that uses sensors to tell what physical activities you are doing in the real world (e.g., running, walking, biking, taking stairs, etc.), which is just starting to show up in products in its most basic form (e.g., the FitBit (link)), and other very cool technologies that hopefully you will hopefully see in products some day in the future.

So, I think I’ve got a pretty significant amount of experience in computing research at the top academic institutions, industrial experience through my time at Intel and Microsoft, and start-up experience through NetRaker, that when combined with my time and study in China puts me in a fairly strong position to comment on where China is in computing and where it might be going.