In Memoriam -- Ray Reiter (1939--2002)

Daniela Drosu, the president of CSBSGS, sent me a message saying that 
"the graduate students of our department ought to show their respect and 
appreciation for professor Reiter" and she added "[she does] not know 
anybody more appropriate to write a message in the very sad event of his 
departure from this world". Though I might challenge this title of being 
the most appropriate grad student to write such a message on behalf of 
graduate students, I certainly feel sure and safe enough to write this 
small in memoriam on behalf of Ray's graduate students. My message will 
of course have a personal touch as it would have been the case for anyone 
of my fellow graduate students who could have written this message. 

During the academic year 1995-96, I was preparing my Master's degree 
at the university of Bonn, Germany, under the supervision of Gerhard 
Lakemeyer. Gerhard did his Ph.D. at Uoft with Hector Levesque. After UofT, 
he came to our university where he helped disseminate some of the ideas 
that have been developed by the Knowledge Representation Group. These  
ideas were centered around the revival of the old Situation Calculus 
introduced by John MacCarthy decades ago to reason about action in 
logic-based Artificial Intelligence. Towards the end of 1995, Gerhard 
gave me a copy of a paper published the previous year by Yves Lespérance, 
Hector J. Levesque, Fangzhen Lin, Daniel Marcu, Richard B. Scherl, and Ray  
on "A logical approach to high-level robot programming". I found the paper 
ground-breaking. In fact, the entire robotics group in Bonn found the paper 
ground-breaking, because we did not believe that classical logic could be used 
at that extent to cope with the intricate problems encountered in robotics. 
A fellow graduate student of mine, Dirk Schulz went on to program an 
interface between the low-level robotics software that was being developed 
in Bonn and GOLOG, the situation calculus based programming language 
presented in the aforementioned paper. The result of all this was a tremendous 
success of the subsequent releases of the software that now included 
a high-level task planing module written in GOLOG. Papers based on 
these subsequent softwares won best paper awards at both the German Artificial 
Intelligence Conference, and the American Association for Artificial 
Intelligence Conference.
   
I first met Professor Ray Reiter in Bonn towards the end of 1996. Ray was in 
town for a conference and Gerhard Lakemeyer invited him for a talk in our 
department, followed by a seminar in the robotics group. Ray was accompanied 
by the late Javier Pinto, who at the time had graduated two years earlier from 
UofT under his supervision. Ray Reiter gave a stimulating talk on natural 
actions, a topic on which he published a paper some months before his talk  
with Javier Pinto. He engaged in a fascinating discussion with some of the 
people in the audience. His way of explaining things on a foundational level 
moved me and convinced me of the immense intellectual value of the man. Of 
course, at that level of graduate education, I had already encountered such 
buzzwords as "default logic" or "Closed World Assumption" and I could trace 
them back to the name of Ray Reiter. I even read these two pieces of 
classical literature of non-monotonic reasoning and theoretical databases, 
respectively. However, it was fascinating to see the inventor of these 
concepts, and the talk convinced me that I should talk to him about 
possibilities of working towards a Ph.D. under his supervision. So after Ray's 
talk I met him. He was sitting on the front steps of the building to enjoy 
some nice weather and smoke a cigarette before giving his afternoon seminar 
to the cognitive robotics group. I walked to him and asked whether I could 
bother him about a request. I was of course intimidated by his greatness. 
He invited me to join him to enjoy the good weather and chat. What followed 
was a great lesson of humility that I will never forget. Usually, you don't 
address a typical German professor (of course with a few obvious exceptions) 
like I did it that day with Ray: even if he is just on visit, you better ask 
for a formal meeting, then if he agrees, you'd better concentrate on what you 
exactly have to tell him, and then walk out by the time frame he allocated 
for you. Ray, however, took more than an hour to listen at my very shaky 
English while I was trying to explain to him that I was impressed by his 
presentation, that I liked his classical work on closed world assumption, and 
that I would like to work with him in Toronto. He took all his time 
explaining to me what he was doing and which research topics were important 
to him. He also mentioned that what I referred to as his classical work was 
just a piece of history and we should not be looking at it. With the word 
"history", he really meant "past things without much interest now". I could 
hardly believe that such a tremendous and famous man would say that 
of pieces of work that are now classical material in Computer Science. So my 
only conclusion was: that was a man with profound humility and modesty on  
what he has achieved.        
         
My wife and I came in Toronto in January 1997. Ray had asked Michael 
Soutchansi and Eugenia Ternovskaia to help me in my first steps through the 
labyrinth of administrative matters at UofT and in settling down. Since he 
was away on sabbatical at the time somewhere in Europe, He assigned me to 
Hector Levesque as his temporary replacement. Ray came back the following 
term and what followed was a great intellectual experience for me. I must say 
right away that Ray was intellectually very challenging. He was not the kind 
of supervisor that would set up regular meetings where a student would come 
in, receive instructions on what to work on, walk out, go on to follow these 
instructions, and then come back to receive further instructions at the next 
scheduled meetings. To see Ray for the first time for talking about a possible
thesis or project topic, you'd better have at least one or two solid topics 
prepared in a well articulated way, with a convincing example that shows that 
your ideas are credible. Then you should give him a lecture for roughly half 
an hour. After that, he would stand up, go to his white board and show you 
all the flows in your approach. He would almost easily come up with an 
alternative way of viewing the same concepts, but a way that is more elegant, 
theoretically sound -- without such a soundness, he would not listen at 
you --, and, above all, simple. Ray loved simple theories that you could 
play with only in your head, without going back to complicated written 
formulas. He very often said that such theories are those with the most 
fruitful and intuitive consequences. 

Writing a paper with Ray was quite an experience. He would reread the same 
paper ten times before giving his OK. At each iteration, he writes down what 
he thinks are rules of good scientific writing. He numbered them: Rule 1, 
Rule 2, etc. He almost never wrote down more than a rule at a time. Doing so, 
he surely wanted to insist on something that was very important to him: 
always do one thing at a time. As an example of this motto, he seldom allowed 
a paper with more than one topic, insisting on the fact that a good paper 
must be centered on one single topic. 
   
Patiently, I tried to convince Ray that instead of doing robotics, I would 
like to go back to some of his early work on databases, a classical 
reconstruction of relational databases in logic that he provided in a book 
edited by John Mylopoulos and others a decade ago. Instead of viewing 
relational databases as model of some first order theory, Ray proposed to 
view it as the theory itself. I wanted to extend that view to deal with more 
world knowledge captured in form of database transactions and active rules. 
I wanted to apply the situation calculus he helped revive with Hector and 
others to find a unified semantics of active database and database 
transaction models. Ray agreed, and I did my Ph.D. thesis on that topic. 
Throughout the whole process of my thesis, he wanted me to come up with 
directions of my own. It was his way of teaching students how to work 
independently. He would of course tell me immediately not to follow a 
way he thinks is a dead end. He would say it very quickly. 

A Ph.D. process is psychologically demanding and stressful. Ray has been 
very supportive on many aspects of my student life. He has been very 
supportive of my studies both financially and emotionally. He would always 
ask how things are in Congo. I happened to know of his hobby of travelling 
through rain forests. I suggested him that Congo had in fact the second 
largest rain forest in the world after Brazil. He really got interested in 
African rain forests and asked whether visiting Congo would be a good thing. 
I told him that because of the political turmoil there, that would not be 
a good idea for now, but he could instead go to a calmer country like Gabon. 
So he went to Gabon and enjoyed it very much. The year thereafter, he 
planned to go to Cameroun. He carefully planned everything asking a lot 
of questions about the country. So when I heard from Hector that he suddently 
cancelled his trip and instead had to go to hospital for treatment, I 
realized that it must have been something very serious to force him to cancel 
a trip to a rain forest !       
 
Ray Reiter was an truly extraordinary man. A man of great and sharp 
intellect. After being told that I was a student of Ray Reiter, many people 
I met in conferences always expressed their admiration about the sharp 
intellect of Ray. I have been very fortunate to have this man as my Ph.D. 
supervisor. It is sad that further generation of graduate students will not 
benefit from his tough and marvelous school of thought. He left the scene 
just a few weeks after I officially entered the faculty at the University of 
Ottawa, and a few weeks before I defend my thesis. Though I am immensely sad 
of his passing, just 4 weeks after my mother unexpectedly died in Congo, I 
feel on behalf of his graduate students that there is more space for 
celebrating his exceptional and remarkable intellectual life than mourning 
him. We ought to go out there and try to have good intellectual careers 
to make him proud of us. 

Ray, que la terre te soit legere !

Iluju Kiringa