Xerox' Hervé Gallaire:

February 9/16, 2005
As Chief Technology Officer of Xerox Corporation and President of the Xerox Innovation Group, Hervé Gallaire is responsible for Xerox's research and technology organizations and $800 million annual investment in research and development. He oversees a thousand-member team that includes the company' inventors and intellectual property experts.

Before joining Xerox in 1992, Gallaire headed the department of mathematics and computer science at the French university l'École Nationale Supérieure de l'Aéronautique et de l'Espace; he has also managed development divisions at Bull SA and GSI. His research activities focused on relational and deductive databases, logic programming and constraint-based programming. He has written 4 books in these fields.

Gallaire holds a master of science degree and a doctorate in electrical engineering and computer science from the University of California at Berkeley and a master's degree in mechanical engineering from the French university École Nationale Supérieure des Arts et Métiers. He is a board member of Cosytec SA.

Technology Research News Editor Eric Smalley carried out an email conversation with Gallaire during the week of January 31st, 2005:

TRN: What got you interested in science and technology?

Gallaire: Very simply, during my university time in France I noticed that I liked the classes that had some theory to present and then apply. Automatic control was then my favorite, and one day we had a 4 hour presentation about computers and software -- Fortran, given by an IBM senior person, very enthusiastic -- remember this was 1966 in France, where it was really the beginning of computing.

That got me hooked, literally. That's all we had but it was enough that I decided to apply to an American university; I got into Berkeley where I developed a passion for not just computers, but information theory and other matters, and got my Master and PhD very quickly. Frankly it's only after my PhD that I started to think about applications -- which led me to abandon automata theory for databases and [artificial intelligence], and architecture, etc. Very straightforward.

TRN: What are the important or significant trends you see in science and technology research?

Gallaire: It may surprise you that as a "software" person I am seeing materials science as still going strong and a major key to our future. It is clear that computing, communications, and much more, owe to materials science (and physics) the tremendous progress that has been achieved for the last 50 years and that is going to continue with exciting new materials and new applications.

I am thinking about polymer-based electronics, for example, and nanotechnology in general. I believe also that digitization of most things that we see around us is going to continue to transform everything we do, in our professional lives of course, but also in our personal lives -- at home, during our leisure, in the design and manufacturing of every object, whether they are large items like cars or small items like packaging of retail stuff.

To make that vision happen a lot needs to be invented in terms of networking, security, privacy, and ease-of use. The extreme competition between all the new communication technologies will drive a good chunk of this progress. Last, but not least, I see bio-engineering taking an ever increasing portion of the science and technology development.

TRN: What types of networking, security, privacy and ease-of-use challenges are standing in the way of that vision?

Gallaire: Let me first start by saying that in a world of many networked devices, in the home, in the enterprise and in the mobile world, wireless connections are going to rule. In fact wireless is the key to networking in a world of ubiquitous devices.

So, if you imagine that we push more and more digitization into smaller and smaller elements that perform as part of their function some communication, the number of networks we may create will be exponentially large compared to today.

In many cases these networks will be ad hoc and short-lived as devices come in close proximity. A lot of information will need to be processed in the network, not in some central node, by having these edge nodes collaborate through new protocols. As they gather information (e.g. localization information, activity logs, etc.) privacy of data will be a challenge, and so will be authentication.

Today, we deal with this largely with two approaches: encryption and lack of transparence. Neither of these will be appropriate in a ubiquitous networked world. Adding more and more encryption strength to keep ahead in the race against computer power would lead to devices with prohibitive cost. Also, keeping the functioning of devices proprietary for security purposes goes against ease-of-use for ease-of-integration. We will need very new approaches here.

TRN: What are the most important technologies for organizations like yours to pursue, and why?

Gallaire: I wish the list was short, but it is not since we have so many different needs in our systems. Let me pick a few:

- Systems engineering, to design better products, which provide users only what is necessary for your business goals at the right price point while enabling a platform approach.

- Systems controls to enable us to replace more mechanical controls (e.g. tight tolerances) with sensors and software. You may not know, but our iGen3 flagship product has over 80 microprocessors, over 100 motors and close to 200 sensors, and we'll put more in.

- The technology of polymer materials because they are key to many components in our systems.

- Software -- writing better, faster with less cost and higher quality; sometimes replacing writing code by model-based programming as we have started to do in limited subsystems.

- Imaging technology, to improve images, better compress them, better render them, providing consistency of those renderings is a big deal; so is color science and technology of course. For example, we are exploiting our deepening knowledge of the human visual perception system to improve how we render color images.

- A last, more specific example would be the solid ink technology which we intend to play out over time into a more complete stream of products.

TRN: Tell me more about utility the of model-based programming.

Gallaire: This is programming that is sometimes also called constraint-based programming.

If you have a model of an object or of a system (e.g. a sensors and motors that move, for example, paper along a track on which the sensors are...) and rules (models) that describe how the objects behave (e.g. how the paper moves, governed by which physical equations relating to the state of the sensors and motors), taking also into account the state of the other objects in the system at a given point in time, then instead of writing a scheduler to move paper from an input point to an output point, you could have instead the models be executed by a problem solver that takes into account the description of the input and uses the model and the rules to dynamically generate the orders to the motors and sensors.

No one writes a program to create the path; only the model is necessary, and a program (problem solver) that interprets the model. For a different problem the model changes, but the interpreter is the same (depending on the generality of the problem solver), hence realizing large cost and time savings.

Obviously not all programming can be done this way, but a large class of embedded software can.

TRN: What details of our knowledge of the human visual perception system can be exploited to improve color image renderings? I'm hoping you can really paint a picture of what's possible, so to speak.

Gallaire: You may have seen those optical illusions where two identical colored shapes appear very different because the colors surrounding them are very different? Another way to demonstrate this effect is to pick colors that are different but look the same because of the differences in the surrounding colors.

We exploit these kinds of effects to perform what is called color gamut mapping. All hardcopy devices have some limitation in what colors they can produce. So, when a user specifies colors outside of the printable gamut, we have to convert the colors to something within the gamut. The most interesting way to do this is to choose colors that are perceived similarly to the desired color within the environment of the surrounding colors.

Another example along these lines is reflected in image segmentation technology. Images are broken up into layers that contain text fonts, color planes, backgrounds etc. The human visual system does a sort of segmentation by using different receptors for high frequency shapes like dark fonts and colors (Rods and Cones). There is a lot to be learned by understanding the systems that nature has provided us.

TRN: I'm hoping you can explain solid ink technology a little further.

Gallaire: Xerox has commercialized a printing technology that we call solid ink. It's a direct marking technology, like inkjet technologies, but instead of using an aqueous or a solvent based ink, it jets a material that starts off as a solid, finishes up as solid on the substrate and in between is liquefied for jetting purposes. This technology combines the advantages of direct marking with the image robustness, such as waterfastness, of xerographic images. Our latest product, the Phaser 8400, prints 24ppm in full color and sells for less than $1K. We have plans to cover more markets with this technology.

TRN: Tell me about the trends in information retrieval and document management. What are the pluses and minuses of these technologies as they exist today? What did you see as the most urgent needs in these areas?

Gallaire: We all agree we are drowning in information. The Xerox Global Services CTO, Bob Bauer, calls it the "digital landfill" -- the image is correct. There is a lot that is available today, thanks to the search engines popularized by the highly successful Google, Yahoo, AskJeeves, soon Microsoft and the many smaller players that are defining innovation.

Yet, the glass is, in my mind, less than half-full for all the good features these systems have, and I do use them and appreciate how they are bring new ideas to the market. I would characterize what we need to improve on, invent and innovate along 4 dimensions:

1. More powerful query tools, beyond keywords

- Natural language, 2D-3D languages, notes, etc.
- Object typing, taxonomies, context-based [tagging]
- Localization
- Activity-based
- Paper as an input

2. Improved search

- Personalized
- Profile-based
- Structured and unstructured data -- tagging... desktop search, enterprise content integration and device integration...
- Deep web
- Multiple media: audio, video, graphic, text
- Semantic web

3. Improved presentation of responses

- Categorized
- Classified
- Mining

4. Integration to work processes

- [Application programming interfaces] open and usable

TRN: Tell me about the trends in human-computer interaction. What are the pluses and minuses of these technologies as they exist today? What do you see as the most urgent needs in these areas?

Gallaire: I am not as clear about what should happen here, because it is really open space. Are we talking about interaction in the mobile world? With desktop applications? With remote services on the web?

If I think what could be common among all of these I would think that easy personalization of the interaction is a key, but simplicity remains the dominant ultimate goal. By the way, this is why we have the vision that our devices are more than devices; they are portals to the services they access.

I also see voice as becoming more important, but slowly. I see paper, by the way, as a way to improve human-computer interaction where it makes sense, and where technology that bridges between paper and electronic document is available. We may wish that everything was digital, but that is and will not be the case for a long time; so bridging across the boundary is still an exciting trend, and it is important to make that experience richer.

TRN: Why are portals better than devices?

Gallaire: Devices (e.g. multifunction device) accomplish their functions (print, fax, scan, copy); but they can and should be more. They can act as portals to a set of services that are on the device or on the net, or on the web. For example I can use the device as a tool to create a workflow and execute that workflow. Simple ones would be scan-to-anywhere, scan and email-to-anywhere, accessing address books, or even scan-to-application where, for example, the application could analyze the document, extract information and do whatever it needs to; that is a powerful paradigm, enhancing the mere scanning capability of the device.

TRN: What are your outside interests, and how do they inform how you think about science and technology?

Gallaire: I have a major interest in sports. I have practiced it -- rugby a lot, running cross country, marathons and mountain races to a good amateur level -- but I am interested in all types of sports.

Why? Two things I guess -- competition, the uncertainty of the near future, every second is intense, no let-down. The second is that sports taught me about team spirit and what that really meant. Even running, by essence a personal effort, is about team work at the modest level I reached -- training together, learning to follow, or to lead, not losing anyone on the course until it must happen.

I see technology development through the same lenses -- and no doubt others -- compete against others or yourself, make the adrenalin flow, be the best you can be, and then of course being a team player, leader or not. Because alone you can't do much that will be significant -- it may be brilliant, it may be enabling, it may be spectacular, but you need so much more to create real value through your immediate team to your value-chain partners.

I am also very interested in politics, and societal issues, but so far, I have been passive.

TRN: How do the technologies you're working on relate to business, culture, and social life?

Gallaire: Xerography, that we invented, has had a huge impact on social and business life. Being able to disseminate accurate information quickly, cheaply, was made possible by the copier that Chester Carlson invented and made possible by Joe Wilson's vision. This is an extraordinary contribution, even if we take it for granted now.

Obviously Xerox is much more than about copying paper documents. But the ultimate goal is the same -- making... life easier and more productive. This is why document management is critical, supporting collaborative work processes as does our Docushare product.

Lastly, documents are the blood of business transactions, aren't they? So, whether they are paper, electronic, or move between paper and electronic, they enable business processes. They are fundamental. Obviously we are working now on how to create a better integration between documents and processes by unleashing the power of the information contained, and too often, secluded in the document.

This requires being able to get at the content of the document, search, mine, infer, and link to the right business process. And it will also enable the emergence of technology in support of the service businesses -- analyzing, mining, optimizing, etc.

TRN: What are the important social questions related to today's cutting-edge technologies?

Gallaire: I see many important questions.

For example, making sure the digital divide does not increase and in fact diminishes over time, [and] providing access to the benefit of cutting edge technologies to as many as possible. In some cases it means devising alternate distribution forms and factors.

Another would be the risk of over-complexity imposed on our lives if these new technologies become so overwhelming that we lose their benefits. For example, being swamped in useless information; feeling insecure because we know so much about specific risks (medical, environmental) that we become paranoid even though probabilistically the risk is very low. Happiness and technology have not been demonstrated to be correlated; does more technology in the future means less happiness?

Privacy, security are obvious additions to this list.

Another big one [is] I do not think we understand well enough the long-lasting impact on the environment of technologies, or rather of what goes into technology-based products and their disposition.

TRN: Can you give me an example of alternate distribution forms and factors?

Gallaire: I was really thinking about projects to find ways to distribute cheaply drugs in developing countries; to create money lending programs for very small sums to encourage entrepreneurship in developing economies as now done in India.

I was also thinking about projects like AMD's project to create an appliance that is a full-fledged PC, with communication capability, rugged, to sell into developing countries for $250 a piece.

The beauty is that these are not philanthropic efforts -- they are real businesses that should make money and therefore ensure they will be developing fast and make a larger impact than pure philanthropic ones (which are still required).

TRN: In terms of technology and anything affected by technology, what will be different about our world in five years? In 10? In 50?

Gallaire: Let me start with 5 to 10 years because that is easier.

I see the evolution of telecommunications changing the way we access our world, more connected, fully mobile, personalized to a large extent, with lots of new devices at home, in the office, on the road. Sensor networks will be there in the 10 years timeframe if not earlier.

I also think the IT business model will have changed to become a hosted model -- either inside the enterprise, where mainframes made of rack-mounted servers [and] using individual multi-core chips will provide everything large mainframes provided.

They'll connect "dumb", easy to administrate "PCs"... Applications will be accessed as web services offered either in-house or in many cases hosted by a service... At home... people will use storage and services offered by third party suppliers to pay by the drink.

I cannot comment about bio-engineering evolution, but certainly I hold hopes that major diseases will be better prevented, if not cured, by early detection made possible again by technology.

As for 50 years from now, here it's purely speculative. But, if we extrapolate the ever increasing trends of technological advancement, the world will look little [like] it does today. Nanotechnology, which is in its infancy today, will likely have produced solutions to many diseases and aging problems.

TRN: What do you imagine you or your successor will be working on in 10 years? In 20 years?

Gallaire: I expect my successor to work on the technologies I mentioned as being important for the future, but of course my successor will have to provide leadership in a different environment, leading very likely to a new vision.

TRN: What's the most important piece of advice you can give to a child who shows interest in science and technology?

Gallaire: This is a tough question since my kids never listened to my advice. Well, they both have extensive scientific and technical training, so I should not complain too much.

What I would say today would be that science and technology are exciting because they offer the prospect to understand your environment, to do something no one else has done before. They challenge you like only sports can do, but they are more useful to society than sports (I am a big sports fan, and I know this statement can be challenged somewhat, I still believe it is true).

TRN: What's the most important piece of advice you can give to a college student who shows interest in science and technology?

Gallaire: I would ask them to always try to relate science understanding to systems understanding. A theorem has a meaning that goes beyond the statement of the theorem -- learn to verbalize that meaning, learn to relate that meaning to real problems where it might apply -- and all of a sudden you have a toolbox that helps you shape the way you look at answers to problems.

Finally I would tell them that reality is never like what we learn in class. Problems are not as well defined, and they have to learn how to cope with it; cope with uncertainty and incompleteness -- not just in science or technology, but in life. Actually, that would well apply to a child, I think.

I would also tell them that many of the interesting, and important ultimately, discoveries occur at the confluence -- the boundaries between many fields. For example medicine is advancing thanks also to mathematics and computation. So perhaps some advice to remain curious, not to specialize too early. Minors are important.

TRN: What are your thoughts on the state of conventional wisdom on science and technology?

Gallaire: I am not wise enough to know what you are referring to. There is a lot of controversy -- fear of science and technology -- these days -- more so than when I grew up. If that is conventional wisdom, we have to work hard to change that perception. For instance, at the Academie des technologies, this is our prime goal -- educate at large.

If conventional wisdom is that science and technology is dry and not as human as other disciplines, I would completely disagree, because in fact there is a lot of beauty in many theories when you can verbalize them, and because there is definitely a lot of excitement and awe to witness accomplishments that we may take for granted. Frankly, our entertainment system is so dominated by superficiality -- to be kind -- that I understand this state of affairs. It must change.

If conventional wisdom is that money is not in science and technology, well, there is enough in science and technology to not worry about that to the point to prefer becoming a lawyer or [going] to a business school or into sales.

TRN: What could be done to improve the pursuit of science and technology research in terms of business trends, politics, and/or social trends?

Gallaire: There are good initiatives that have been published recently. I will refer to the report of the Council on Competitiveness presenting the National Innovation Initiative. Not that I agree with everything they propose, but I cannot do a better job than they have without extensive work. The reference is

TRN: What question would you like to be asked in an interview like this? What is the answer to that question?

Gallaire: There are many questions I have been asked. One that may not have been asked is about how I live technology? I am a big technology fan, especially everything that keeps me connected. I really do not feel that as being stressful, quite the opposite; it frees me because it helps me manage urgency as I want it, without stress about it, with very few exceptions.

TRN: What books that have some connection to science or technology have impressed you in some way, and why?

Gallaire: There are a number of books that have had a lot of impact on my thinking that connect to science and technology. Chronologically, I would name:

The Art of Computer Programming (3 volumes) by Don Knuth -- the beauty of reasoning about algorithms, understanding complexity at its best.

Gödel, Escher, Bach by Douglas R. Hofstadter -- purity, range of thoughts, inspired by mathematics.

Redesigning Humans by Gregory Stock -- so insightful about how bio-engineering might change the world we all know.

The Innovator's Dilemma by Clayton Christensen -- not about science or technology, but I liked again the clarity, perhaps the simplicity of thinking, refined in the books that followed.


Discuss this interview or anything else to do with science and technology on the TRN Forum.

TRN needs your help. Please click here for details.


Page One

Atomic scopes eye living tissue

Brainwave interface goes 2D

Software ties marks to digital text

View from the High Ground:
Xerox's Hervé Gallaire

All-silicon chip laser demoed
Nanotubes crank out hydrogen
Nanotubes on plastic speed circuits
Lens design promises tight spots
Silicon nanocrystal transistor shines
Nano triangles concentrate light
Musical additions sound alarm


Research News Roundup
Research Watch blog

View from the High Ground Q&A
How It Works

New: TRN's Jobs Center

RSS Feeds:
News  | Blog  | Books 

Ad links:
Information on RFID, VoIP, VPN, ZigBee and more

Buy an ad link


Ad links: Clear History

Buy an ad link

Home     Archive     Resources    Feeds     Offline Publications     Glossary
TRN Finder     Research Dir.    Events Dir.      Researchers     Bookshelf
   Contribute      Under Development     T-shirts etc.     Classifieds
Forum    Comments    Feedback     About TRN

© Copyright Technology Research News, LLC 2000-2006. All rights reserved.