Alfred Spector

IT'S MORE THAN JUST SEARCH.

Type Alfred Spector into the familiar text box and, with a click, the results show: “the space shuttle primary computer system” … “Computing our Children’s Future” … “on voice search, hybrid intelligence and beyond” … “The Principle of Consistency and the Conditions for Creativity.”

The last two entries seems to be the most on target, as Spector ’76 (Applied Mathematics), Vice President of Research and Special Initiatives at Google, spends his days testing the potential next big things: live and in real time.

Curious users can find speech processing technologies, auto-translation tools, personal health care record management, advanced search, and much more scatted throughout Google technologies.  

In fact, during his visit to SEAS in February 2009, he asked a group of computer science graduate students to give him the skinny on what Google apps they liked, disliked, and would like to see. No holds barred.

His confidence stems from a combination of empirical testing and raw experience. Before joining the search giant, Spector held a stint at IBM, served as CEO of a start-up begun at Carnegie Mellon University (where he was a faculty member), and earned his Ph.D. at the university that gave rise to Google, Stanford.

In the cloud environment, being “on air” all the time is simply the new reality for delivering computer software and services, says Spector. It is less wait and see, and more design, watch, learn, and iterate.

The answer, after all, may be just one click away.


Some imagine that getting a job at Google is kind of like having the gates of heaven swing open. Do you feel that perception matches reality?

At Google we are really excited, challenged, and made happy by both the incredible mission that we have of organizing the world’s information and the enormous technical leverage great computer science has on the problem.

I do feel now, and I did feel when I first joined Google, that I would be a kid in a candy store, working on incredible problems with incredibly smart people using technological approaches that were really likely to be successful. That combination has really driven the company.

Google does have a fun culture. We try to make it fun, but more importantly, our mission and opportunities make the jobfun.

Google is known for putting its “beta” releases online so users can play around with products still in development and make suggestions/comments. That seems a risky approach to R&D. When does democracy end and dictatorship begin?

Google is a cloud computing company, so we provide services. Services are more amenable to rapid evolution, as we do not have to send out a new CD when we create a new version. It is not surprising that there would be a fluidity of features in our projects—we can change them in our central servers and they quickly flow to lots of people.

So, that’s the technical change; it simply wasn’t possible in the packaged software industry. If you did a daily release to users, you’d bankrupt the company and the users would kill you.

That being said, there is also strategic rationale: We aim to get a first release to a point where we can quickly find if there is creativity and value to what we have done. We know as engineers and product managers that no matter how clever we may try to be, we cannot always anticipate what will be valued and what won’t be.

We are, thus, highly empirical. Rather than trying to analyze to the nth degree, our object is rather to be able to measure and refine. If we put out a new feature and no one uses it, or people use it and then they don’t come back, we know it is not a good feature and we try like the Dickens to improve it.

[Note: Check out Google.com/labs to learn more.]

Google recently announced plans to infuse more money into university collaborations. What are you hoping to get out of these partnerships?

Our objective was first to help universities work on problems that we perceived were very important—and to do that not only by providing funding, but by providing interactions with Googlers who could collaborate.

As a technology provider with many users and much data and processing, we often see problems in ways that university researchers cannot. We also want to further education, so we are providing fellowships as well.

Our grants are primarily related to computer science and engineering, but we also fund programs in humanities and social sciences, in the increasing areas of overlap with computer science.

As a side effect, we hope universities will quicken the pace at which they innovate. In particular, we hope this will help us collectively to focus on longer-term objectives and objectives that are motivated by the large-scale problems that Google envisions.

The way universities manage and create knowledge is changing. What should universities be thinking about in terms of scholarship in the digital age?

The change is already here. First, there is a vast amount of data becoming available. Whether in the hard sciences, the social sciences, or the humanities, research can now be data-driven in a way that was previously impossible. This is brought about by the wide availability of sensors, the vast connectivity provided by networking technologies, and vast processing and storage.

Second, the internet makes it possible to disseminate information instantly to billions of people, rather than, say, one thousand. That’s potentially a million-fold increase in distribution of information with essentially zero economic cost. And the dissemination can be instantaneous. This clearly will have an impact on scientific flows: More people and fewer delays.

By no means does this indict peer review and expert editorial opinion. Both of these work in a networked world. There is clearly a need to validate and substantiate, but techniques will no doubt continue to be modified to work well in the much faster and increasingly online world. I think the opportunities to enhance scientific publication (speed, quality, reach) could materially impact the pace of advances.

Just a few questions to ponder: Should a scientific paper once published be static or can it change as the author’s views change? How do you read the collective scientific output of someone or on some subject? Is it an integrated approach (often, better for most readers) or a historical, item-by-item, analysis, which could be useful for some scholars?

And, then there’s an opportunity for readers to provide instant feedback, so documents become more alive. How does the social network enable you to know what you need to know through recommendation systems? That is, what are other relevant people are reading?

There is no doubt that change is afoot. Universities and their researchers are already embracing it, but there is more to do.

By putting out course materials and other assets, could Harvard easily be co-opted or virtualized elsewhere? Or could some of its essence be lost?

In my view, Harvard, by having its materials widely available, only increases its value to society. By so doing, it also increases its brand value. The more people who use Harvard materials, the more who will realize it is a marker for top scholarship.

Could you duplicate Harvard in a purely virtual or online world? My answer is ‘no’ because Harvard is most significantly the interactions of people, in a variety of settings, where co-location is quite important. I think of my undergraduate experience, for example, in North [now Pforzheimer] House. My interactions with my roommates Geoff Clemm [formerly of Sun and now at IBM] and Trip Hawkins [founder of Electronic Arts] as well as people in other disciplines, like Yo-Yo Ma, vastly enriched my education.

The open courseware initiatives, like the one at MIT, are a valuable approach, I think, to getting more material to a wider audience. There is a lot of change happening in education, but I see every reason to believe that the online technologies should significantly enhance Harvard in the broadest sense.

Do you have any memories during your time at Harvard that you would like to share?

-Google's Profile of Alfred Spector-

Alfred Spector joined Google in November of 2007. Alfred was recently Vice President of Strategy and Technology and CTO of IBM's Software Business.

Prior to that he was Vice President of Services and Software at IBM Research. He was also founder and CEO of Transarc Corporation, a pioneer in distributed transaction processing and wide area file systems, and was an Associate Professor of Computer Science at Carnegie Mellon University.

While at CMU he did fundamental work in a number of areas, including the Andrew File System that changed the face of distributed computing.

Alfred received his Ph.D. in Computer Science from Stanford and his A.B. in Applied Mathematics from Harvard.

He is a member of the National Academy of Engineering, a Fellow of the IEEE and ACM, and the recipient of the 2001 IEEE Computer Society's Tsutomu Kanai Award for work in scalable architectures and distributed systems.

Yes. In the early part of my career I was very much a programmer and spent an enormous amount of time in the [former] Aiken Computation Laboratory and the then-new Science Center. (I recall thinking it was such a strange building, but it’s proven a useful space.) In the summer after my freshman year, I worked with fellow students and a number of faculty, including Professor Bill Bossert. My experience was so intense, that I likened it to a medical internship: Immersive training.

During that period, we developed the Harvard-Radcliffe Student Timesharing System that used a PDP11 to support about 20 students (including Bill Gates) interacting via teletypes. It was one of the first systems outside of Bell Labs based on the Unix Operating system, a system which eventually achieved very wide use in multiple guises. I particularly focused developing the PPL programming language, which was the introductory programming language for undergraduates for quite many years. I still meet people who used the software I wrote.

Of course, there were humorous things. I remember my college roommate who is now a distinguished engineer, streaking through a course in Sanders Theatre.

I found Applied Math 110 taught by Professor Ben Wegbreit Ph.D. ’70, who eventually became an entrepreneur, to be an amazingly superb course that taught machine language programming and LISP, the language used for early AI work. The course changed my life.

I remember learning from Professor McElroy, who taught me modeling and differential equations, and who did some of the earliest work analyzing the effect of CFC’s on the ozone layer. I found this to be a fantastic use of modeling that has influenced my understanding of and sensitivity to all the work that needs to be done on the much harder problem of the global warming.

I remember listening to Yo-Yo Ma many others play cello and piano and North House, which was a music dorm at that time. (I now coach my kids on cello, violin, and piano.)

You moved from being a serious programmer to a high-level manager. How do you balance deep skills with broader ones?

I strongly believe this an individual thing and there is no correct answer. I was very deep into programming at Harvard, but I took Humanities 3Philosophy 5, and The Decline and Fall of the Roman Empire, and I took a spectacular course in moral dilemmas that influences my life in many things. I was able to do an enormous amount of work in my field and still get some breadth in economics and the humanities. I think students should really avail themselves of the breadth of Harvard.

How much depth or how much pure mathematics or how much computer science? The one thing I will say about the field of computer science is that there is the opportunity to do incredible work at the core of the field. But, there’s also an incredible opportunity to do work that is interdisciplinary—to apply computer science to all of the application domains, whether bioinformatics, medical informatics, economics, astronomy, or virtually any other discipline.

And you don’t just have to do one thing your entire career. Even if you start off in one place you have a lot of flexibility. The stereotype of someone that is a computer scientist sitting in a basement programming 60-hours a week (without time for a shower) is incorrect.

What motivates you? What gets you up in the morning?

Whenever I can solve something.

I feel very excited if I can go and solve a problem in almost any domain—whether it is hiring someone (that could be an opportunity that could change what we do) or if it is getting involved in some technology where I can see its uses and make connections in the organization so that the technology is used. Occasionally, I have a good idea myself that can then be utilized.

I find that meeting a challenge is what gets me to want to wake up.