MIGUEL BOYAYANResearcher Claudia Bauzer Medeiros leads quite a busy academic life: a professor at the Computer Science Center of the State University of Campinas (Unicamp) for the last 20 years, her extensive academic output has involved over 30 projects in the field of scientific databases, mainly as applied to biodiversity and agro-environmental planning. “The Unicamp environment houses many enthusiastic researchers. To this day, all I have to do is cross the road where my building is located to find research partners in other areas”, states the professor, who is equally enthusiastic about her scientific initiation students and her post-graduate students. Claudia is also something of an activist. She has chaired the Brazilian Computer Science Society(SBC – Sociedade Brasileira de Computação) and during her term focused on a worrying phenomenon: the worldwide drop in the number of students interested in computer science, especially women. SBC’s work in the search for new vocations, and initiatives such as the Programming Marathon and Programming Olympics, bringing together students from all levels, was internationally recognized last year, when Claudia was granted the Change Agents award by the Anita Borg Institute for Women and Technology and by the North American Computer Science Society.
Under Claudia’s leadership, SBC was also involved in outlining certain major long-term research challenges that the IT field is bound to face in Brazil. “Several development agencies have sought us out to try and define computer science priority areas, but we lacked an articulated response”, she states. Last year, computer science researchers met with experts from other areas indicated by the Brazilian Academy of Sciences. Five main challenges emerged from this debate, ranging from the pursuit of post-silicon era technologies to the development of research capable of ensuring the reliability of the technological network that now engulfs our life (mobile phones, security cameras, automatic automobile systems, elevators, hospitals, etc). Accessibility, which is one of these challenges, was the basis for a public notice released last month by the Microsoft and Fapesp IT Research Institute (see pg. 22 article).
Born in Rio de Janeiro, Claudia Bauzer Medeiros graduated in electrical engineering in 1976, from the Rio de Janeiro Catholic University (PUC/RJ – Pontifícia Universidade Católica do Rio de Janeiro). She later obtained her doctorate in computer science from the University of Waterloo, in Canada, did post-doctoral work in France at the National IT Research Institute (INRIA – Institut National de Recherche en Informatique et en Automatique), and completed further post-doctoral work (livre-docência) at the sTAte University of Campinas (UNICAMP), where she also passed the placement exam for head professor. She is single and has two brothers: one is a mechanical engineering professor at the Federal University of Minas Gerais (UFMG – Universidade Federal de Minas Gerais), while the other teaches mathematics at the Fluminense Federal University (Universidade Federal Fluminense). Below, excerpts from the interview she granted Pesquisa FAPESP.
At the launch of the Microsoft and Fapesp IT Research Institute, you presented a vision of the computer science research infrastructure in São Paulo. How does this São Paulo base compare to the rest of Brazil?
Actually, São Paulo is a privileged state in research terms in all fields. Regarding the number of researchers, research projects, research groups and even post-graduate programs, the state accounts for 25 to 30% of Brazil’s total. I’m referring to scientific production as a whole, not only to computer science. From the data I’ve obtained, São Paulo accounts for about 50% of scientific production where published articles are concerned. With regard to computer science, I obtained two indicators. The first one is from CNPq. Computer science currently has roughly 26 researchers classified at CNPq as researcher 2 or 1. And São Paulo has 26% of them.
Is this a small community?
Worldwide, as well as in Brazil, the computer science community is new compared to other sciences. I have just come back from Chile where I talked to people from other countries about this. My impression that Brazil undoubtedly produces more in the computer sciences field than all the other South American countries put together, or more even than all of the rest of Latin America, is becoming increasingly consolidated, though I have no figures to prove this. The post-graduate students are the second indicator. São Paulo accounts for 26% of all of Brazil’s Ph.D.s and master’s degrees in computer science. One of the problems of these statistics is that there are post-graduate computer science courses connected not only with computer science but also with the different types of engineering and with an area that Capes classifies as multidisciplinary. And I did not take these courses into account. The interesting thing regarding young people in this field is that the State of São Paulo, for instance, is home to Brazil’s oldest undergraduate computer science program at Unicamp. This is also one of the oldest programs of its kind in the world, dating back to 1969. Although the Federal University of Bahia (UFB) course was created at the same time, the earliest Unicamp classes graduated first. Unicamp’s computer science courses played an undeniable role in the definition of curricula throughout the rest of Brazil. Another thing I must stress about the São Paulo state leadership is FAPESP’s work. It is well known that most research support foundations in other states have a very irregular financing scheme, whereas FAPESP guarantees the stability of research in São Paulo, which was certainly instrumental in helping the state’s researchers stand out, including those who do research into computer science.
What is the stock of computer science graduates?
There are computer science courses connected with the different types of engineering, mathematics and other fields, so that it is difficult to quantify the number of graduates. The Ministry of Education’s 2005 census data show that there are some 30,000 graduates a year, a large but still insufficient number in terms of meeting corporate demand. These 30,000 receive diplomas in computer science, computer engineering, systems analysis and systems administration, a series of different course names that have to be painstakingly picked out in the national statistics and that account for about 3.5% of Brazil’s total yearly number of graduating university students. This isn’t much. The various names are connected with another issue, namely, the need to recognize the computer science area as something that permeates all fields of knowledge. This is an ongoing discussion between researchers and research support entities. Professor Carlos Henrique de Brito Cruz, the FAPESP scientific director, recognized this computer science characteristic in a way when he created the Computer Engineering and Science coordination office.
The issue of multidisciplinarity extends to all areas, not only to computer science?
One of Brazil’s chief challenges regardless of the field of research is the need to do joint research. You rarely manage to advance on your own. And computer science is one of the cornerstones for research into other areas.
What are your views on these cross areas?
The results of research into computer science serve everybody. Furthermore, all over the world, the range of research that calls for a computer science researcher to solve a problem is growing. For instance, how to manage the data that research generates, or how to develop sophisticated analytical techniques. Another aspect is cooperation at a distance among teams. Thanks to Internet facilities, you can work with a lot of people. But how do you maximize cooperation? There is a field of computer science research called Computer Supported Cooperative Work that concerns computer-supported collaborative efforts. Most of this research aims at providing support for people from any area in their distance work. These cross areas also benefit computer science research, opening new fronts and generating challenges.
In your presentation, you showed something of the São Paulo physical IT base?
I only mentioned some very important FAPESP programs, such as the IT Program for Advanced Internet Development (TIDIA – Technologia da Informação no Desenvolvimento da Internet Avançada), which is not directly connected with the physical base, but that involves a network portion, for example. Internet first appeared in Brazil primarily in São Paulo, through the ANSP network. Infrastructure is a good word because it reminds me of the need to mention that researchers from other fields feel the need to collaborate with computer science. But what they ask for is infrastructure work, the establishment of networks, or software development, all of which isn’t really research.
There’s a great difference between research and technical support?
To be able to really interact with people from other areas, those in computer science must first persuade them that they, the computer people, have something to contribute. There’s always the vocabulary issue. First you really need to work together to convince the sociologists or the biologists or the environmentalists or the philosophers, whoever, that you understand their vocabulary and that you’re able to cooperate at another level, and not only as a provider of services.
And what is this other level?
It consists in being able somehow, through your research, to add to his research, which doesn’t only involve your skills in installing software in his PC.
What is the added level of cooperation that a computer science researcher must produce, for instance, vis-à-vis a traffic engineer dedicated to the drama of managing traffic in a city like São Paulo?
I’ll give you examples at several levels, in laymen’s terms. First: researchers in the interface area can develop screens that enable the engineer to see different level of what is going on. The interface area is growing a great deal and accessibility is connected with this. Accessibility means computer science research into how to make information accessible to those who use it. At a second level, we have the construction of programs that make simulations easier. The traffic expert may understand optimization, but the specs of new programs and how to build them calls for algorithmic knowledge. You can create more sophisticated algorithms that can accelerate visualization of what is happening in real time, for instance. This is computer science research. Then there’s the network infrastructure part. Where is the traffic engineer getting his signals from? From sensors. One must have computer network experts, who will say: “I’m going to research how the signals can reach the traffic management hub more efficiently, with less noise.” And there’s also my area, databases. How do you guarantee data reliability? If a sensor fails, how do you know that it’s the sensor that’s failing? How do you store data for analyzing the past, to keep the same thing from repeating itself in the future? If, for example, a curve describing the fluctuation of the number of cars going through a certain area is behaving in a certain way, when did similar behavior appear in the past? And if it did occur, what does this say about the future? If that behavior tells me that in ten minutes a humongous, five-hour traffic jam is going to take place, then it’s time for me to make a decision. All of this depends on research into databases.
How many computer science sub-areas are there?
There are different classifications, developed by several international organs that review them permanently because computer science is very dynamic. Two or three years ago a major commission was set up at CNPq for restructuring the fields of knowledge table used to classify research. When they published their results, many groups reacted very strongly (including the computer science people), because they felt inadequately classified or reclassified. The initiative came to a halt. What did computer science do at the time? We had been dislocated from our current slot in the larger field of exact sciences and earth sciences and bundled together with different types of engineering. The Brazilian Computer Science Society (SBC – Sociedade Brasileira de Computação), with the support of all the post-graduate computer science programs in Brazil, developed a statement that was sent to this commission, with the following effect: we are not merely exact, nor are we only engineering, we are both. We also used as examples the US model and the European model, and said: what we want is a major area called computer science.
Without too much theoretical discussion about the epistemology of computer science, right?
SBC currently has 20 Special Commissions. These are groups of researchers dedicated to a field of computer science in sufficient numbers for at least one scientific event a year to be organized, an established tradition in Brazil. SBC organizes 35 congresses a year. So you might ask: “So there are 35 computer science areas?” The answer is no.
And couldn’t you define, let’Computer science in the futures say, just a few priority areas in computer science?
It’s impossible. All attempts at this, even global ones, have failed. Consider, for instance, the criterion for defining as a priority area all those with a social impact. Well, would anybody who is proposing a theorem produce social impact? If this theorem shows that a certain type of cryptography guarantees the security of somebody’s medical data, it has social impact. It’s very hard to say: “Ah, this has no social impact.” It’s difficult because computer science permeates all fields.
You are an electrical engineering graduate. Does computer science still absorb professionals from other fields?
Computer science absorbs a lot of people from other fields, which leads us to another subject, namely, regulating the profession. SBC has one executive office solely geared toward this. A few years ago, we managed to submit to Congress a bill whose chief merit was having in its first article a statement to the effect that “the exercise of computer science professions in Brazil is open, regardless of registration with a professional body.” What does this mean? That one doesn’t need a diploma in computer science to be a good professional. Obviously, this has nothing to do with research, but with the issue of the professional. It’s the company that is responsible for hiring. It can tell good and bad professionals apart, regardless of whether they have a diploma in the field or not. Other bills on the theme regulate the work of programmers, analysts or each computer science professional , associating each diploma with the activities the person is allowed to carry out, restricting the profession enormously. They are based on an exaggerated esprit de corps that is not applicable to computer science, among other reasons because every three to four years new skills appear in this area.
MIGUEL BOYAYANWho advocates this excessive esprit de corps?
Certain professional bodies or the students from poor quality schools, who obviously fear competition, and sometimes even students from good schools. We have to persuade students on an ongoing basis of what is important to know. The diploma is certainly important, but a good quality diploma, not just any diploma. And most of them end up agreeing with this.
What are the SBC bill’s chances of success?
The proposed bill was presented to the House of Representatives by congressman Ronaldo Vasconcellos. However, every time the legislature changes, everything goes back to square one and the bills are filed away, which is what happened to ours. When a new project on the subject is presented, the reporting congressman has to get the bills connected with that subject, put them all together and produce a consensual proposal. When an attempt to regulate the computer science profession next appears, our bill will automatically be pulled out and then we’ll wake up and turn up there again.
SBC organized an event that established five research challenges for the future of computer science in this country. How did you arrive at these challenges?
Many countries have felt the need for long-term planning in the IT area. Bodies similar to CNPq or FAPESP in countries such as South Korea, England and the USA have been organizing events called major challenges since 2000. Commissions have been created and long-term views selected and debated by the involved researchers, who proposed consensual proposals about what these challenges might be. Once this was done, those countries used these challenges to guide their research financing policy in different areas – including computer science. In Brazil, this type of planning is yet to take place. There is even a wish for this, but we have so much to do that it’s hard to know where to start. SBC, however, has recognized this need. Every two or three months, I get a request, as chairperson, to help with a public notice or to state our various priorities . But I can’t do this. So we made a national call for proposals for Future Visions of Computer Science. All this took place between December 2005 and April 2006. 22 visions were selected by a commission of five senior computer science researchers in Brazil. I was one of them. These people met for two and a half days in São Paulo and held debates with guest researchers, from outside the computer science field, who had been indicated by the Brazilian Academy of Science (Academia Brasileira de Ciências). Five major challenges emerged from this meeting. The interesting thing was that we got together experts from many computer science areas; there were young people, old people” As computer science is a new field, there are only a few researchers with 1A rank according to the CNPq classification. At that time, May 2006, there were 12 1A researchers in Brazil, ten of whom attended the event.
And what are the challenges?
Two factors permeate all the challenges. One is multidiscipline. And the second one is to train researchers and manpower so that they are better prepared to work in this multidisciplinary world, where everyone must co-operate. My idea for presenting such challenges results mainly from a hardware vision, all the way to the user. The basis of the first challenge is: what happens when the silicon era comes to an end? Chips are currently made out of silicon. But silicon is reaching the limits of what it can stand because of the demand for high performance. What is happening? The chip gets too hot and melts. What does one do? Instead of using one processor, one uses, let’s say, two of them, on the same chip. Have you heard of “dual core computers”? This means it has two cores rather than just one. One of the solutions is parallelism, with processors or computers working at the same time. But there’s a limit. What solutions are available? One example is biological computer science. The idea is the following: human beings function on a base of trillions of cells that reproduce themselves and when, let’s say, 10% of them fail, the others keep things under control. So why not make computers based on this type of idea? Several specialized computers, like cells? Another example: why don’t we use the DNA model to store data?
How would this work?
A DNA chain stores millions of pieces of information. If you manage to reproduce this type of mechanism, you won’t need large disks any longer. Another impact of the new hardware is how we teach computer science, because computers are going to work differently. We will have to create new programs, new systems and new computer networks. The world is going to change. When is it going to change? We don’t know; we don’t have the faintest idea. But if we don’t start to prepare now, in another ten years we will certainly lag behind those who are already thinking about this. This is the challenge: the transition from silicon to new technologies.
And the second challenge…
The second challenge is connected to data, which is my area, to the management of large volumes of multimedia data. Let’s take, as an example, a recording of this interview. Imagine, for instance, that it already comes out in writing. Assume that somebody wants access to all the interviews that discussed major computer science challenges. Who discussed the subject? What photographs were taken during the interview? This demands that several types of data be interlinked. There are lots of text-based search engines. But there is a growing need for richer content. For example, instead of asking “which pictures did Rembrandt paint?” you might draw an outline and say: “Which pictures have this approach?” Your question isn’t based on text, but on a drawing. Imagine all these satellite images that are generated, or the data from sensors, or, in a hospital, the thousands of medical tests of all the patients that someone might want to study for an epidemiological study. Such studies would be applicable to journalism, teaching and the social sciences. And what makes this challenge different is the fact that it’s multimedia, that it involves video and sound. Additionally, the data is on the web. This again can aid scientific research.
And the third…
It’s the computational modeling of artificial, natural, sociocultural, man-nature interaction phenomena.
Translating your words, are we talking about social phenomena, the crazed crowd during Carnival?
I love this example! Or social interactions on the Internet, within a country, within a community. The Internet demands massive research work regarding the modeling of artificial phenomena. There are millions of computers that join and leave the network all the time. This is also modeling, the modeling of traffic. What is common to all the problems is the idea that you need to find out what variables should be studied, which equations will describe that, and what simulations should be conducted, as well as the data analyses, the samples. In this major challenges workshop I learnt something very interesting: the movie The Lord of The Rings shows a battle between monsters and a group of goody-goodies. How was this section of the movie made? A program was created incorporating some actual interpersonal interactions into each one of the artificial characters. What did this generate? An animation that fascinated social scientists. They had never had a laboratory that showed the interactions of millions of human beings. Modeling generates possibilities for research into social sciences or into psychology.
And the fourth challenge?
That’s the one that helped us put together the FAPESP public notice released by the FAPESP-Microsoft Virtual Institute. This first public notice uses a large segment of text from the fourth challenge. And that made me very happy. It’s an issue of accessibility, of Brazilian citizens’ widespread and participative access to knowledge.
When you made the presentation, FAPESP’s chairman, Carlos Vogt, recalled that the social issue was interwoven with the scientific research project, because one is dealing with access for Brazil’s entire population?
There is the issue of universal access. A second aspect is participative access, something that in a way has to be compared with several government bodies’ vision of digital inclusion.
Does this mean that just outlining solutions for a given group is not enough? And that this group needs to talk so that you can be clearer about what it needs?
It needs to be involved with building the solution. But that’s not all. What’s the use of having a sophisticated program if nobody’s going to use it because it doesn’t fulfill people’s needs? There are hundreds of such disaster stories about the development of software or hardware designing systems without the involvement of those who are going to use it. Billions of dollars are wasted.
Could you give us an example?
There’s a fairly famous one about automating the FBI’s data network. The project was discontinued after the US Congress had increased the budget several times. Every time a report was produced by a group of independent experts pointing out problems, the management was changed and the project went forward. And this was because there was no involvement by its ultimate users. The word “participative” means the possibility of announcing, at each stage, that “Listen, what I want is this”. The fourth challenge is also to understand how Brazilian citizens can contribute to the country’s growth. We’re talking about the need for digital literacy and not merely digital inclusion. Digital literacy doesn’t mean knowing how to click the mouse. It means, when you click, knowing where to click and being able to make use of things. You don’t need to be a computer science expert. This leads into another subject: the issue of teacher training. How can one expect children to achieve digital inclusion if teachers are unable to teach a child to use a computer properly? Of course, giving them a US$100 laptop should help, but this per se won’t solve matters.
The SBC ended up supporting the FAPESP public notice. Was the intent behind this to provide input for policy formulation?
This was the reason we established the major challenges: to provide input for development agencies. They were not designed to be limited to the community, to be just the wishes of a research group. As they were taken up by Microsoft Research, I’m happy to know that the challenges Brazilian researchers prepared are being used in the USA as well. From the FAPESP and the Microsoft points of view, there was a desire for innovative and long-term research, which research should also be socially relevant. The other challenges reflect social interests as well, but the accessibility issue touches directly upon Brazilian needs. Regarding this point, I must acknowledge FAPESP’s support for the major challenges event as well as the support we were given by Professor Brito Cruz, who attended the event’s final meeting.
What are your expectations regarding the impact of this type of initiative?
One of its merits is to show that a development agency can actually work in cooperation with foreign entities – which agencies, keep in mind, also engage in research, because this was done with Microsoft Research – and that this is not “dirty”, though this is a view that a lot of researchers in Brazil continue to hold.
Undoubtedly. There are those who feel that entering agreements with multinationals or with companies is wrong, that government money is clean and that corporate funds are dirty. This rather old-fashioned and unrealistic view that public bodies cannot associate with companies in support of serious research still exists. The institute can be a model for other governmental development entities. Another effect of the public notice is to attract researchers from other fields to cooperate with those in computer science.
What about the fifth challenge?
It’s the development of reliable ubiquitous systems. More and more computer science artifacts are present wherever one goes. Sensors, mobile phones, it’s the Big Brother from 1984 [George Orwell]. And increasingly we need these things to get on with our lives. It’s the elevator with an inbuilt computer. It’s the traffic system. It’s the mobile phone (nowadays it seems I’m the only person who dislikes using one). This is ubiquity. On the other hand, our dependence creates the need for reliability. Our lives depend on having a large number of computer science systems up and running. And also on the scale growth of all of this, whether it’s software or hardware. However, computer science doesn’t yet have sufficiently broad and proven theory and techniques to be able to guarantee this expansion, this ubiquity, this reliability.
You head SBC and are a professor atUnicamp; here at FAPESP, you are one of the three computer science area coordinators. How do you go about keeping all this in order, without using management software to run your life?
That’s very complicated. And no, there’s no management software involved [laughter]. All researchers engaged in several activities will say that their weekends and evenings are a write-off. Actually, we do what we do because we like it, but, on the other hand, there are sacrifices that go hand-in-hand with all of this. I’m very lucky to always do what I like doing and to have wonderful undergraduate and graduate students. Unicamp has this great merit: as it’s a good university, it attracts a good student body and good students attract other good students. Moreover, good students egg professors on. I’m enormously proud of being a Unicamp professor.
The SBC holds the Programming Marathon and the Olympics. What is the impact of these initiatives?
I think SBC is a model because it does a lot of things with only a very few people – who are pulling their hair out, of course. There is the issue of digital inclusion, which the Society is working hard on by means of the Olympics and, on another level, through the Programming Marathon. These are held throughout Brazil, for children from the first year of elementary school to the last year of high school. There are four levels of paper-and-pencil tests. They consist of logic questions and, at the two most advanced levels, for the older kids, they involve a bit of programming. The school registers for this and the whole thing is free of charge. After two selection stages, the best students from all of Brazil come to a university, currently Unicamp, to attend lessons with professors from this field. Once the one-week course is completed, there is a test for the older ones and the four top students are sent to the International Olympics. Brazil has won lots of bronze and silver medals, but we haven’t got a gold one yet.
What is the objective of this?
To attract more people to the computer science area. This is strategic for us and for Brazil. All over the world, the number of students interested in programming is dropping sharply. Here in Brazil, we, from the Brazilian Computer Science Society, were lucky to get the support of the Carlos Chagas Foundation, which sponsors us. Since last year, it has provided us with funds to maintain the Olympics and the Marathon. The Foundation feels that this is an excellent type of digital inclusion in the education of young people.