Frontier Exclusive Visionary Interview for hardware, software, system related business and and academia


The Best Way to Predict the Future Is to Invent It: It Is Not just Smalltalk

System Design Frontier (SDF): You worked for Xerox (PARC), Atari, Apple, Disney and HP in your entire career in industry, which company was most innovative one you believe when you were on duty at that time?

Alan Kay (AK): A tricky question. Xerox PARC was the most innovative company group that I worked for, largely because Bob Taylor (who had been a Deputy Director of the famous ARPA-IPTO research community in the 60s) in the early 70s set up a research environment that had very good people, the best kind of funding and charter, and could choose its own directions. Xerox as a company got less innovative as the 70s evolved and had great difficulty recognizing the many research results from PARC.

Apple was innovative enough to try to adapt the PARC ideas, but, as Steve Jobs has said publically a few times, did not have enough perspective to understand and recognize the more subtle ones (how object-oriented programming was done there, innovative HW architectures that PARC invented and used, the Ethernet, etc.).

SDF: Personal Computing as opposed to main frame, has offered us, every individual great convenience in our daily life during past 3 decades, what do you believe is the future of Personal Computing?

AK: When Gandhi was asked what he thought about Western Civilization, he said he thought it would be a good idea! I don't think the personal computer revolution has happened yet, in part because only a few surface ideas have been commercialized, and because every new medium has a special new kind of "literacy" that appears after some decades of imitating old media, and we haven't gotten close to that authentic literacy (which, analogously to writing and reading, is about authoring and dealing with "computer stuff" rather than "computer stuff imitating old stuff").

So, let's hope with Gandhi that Western Civilization is possible, and a smaller part of that would be to hope that the real computer revolution will happen and not stay bogged down in its current myopia.

SDF: Object-Oriented Technology, starting from SmallTalk, to C++ and Java, has made software development much more easier, yet it still could not fill the productivity gap that is still widening, what do you believe is the future of Object-Oriented Technology?

AK: Like "Personal Computing", I coined "Object-Oriented" to stand for very particular ways to design things. But both of these terms today are just paint that is put on anything to make it more trendy and seem forward looking. For example, the differences between Smalltalk and both C++ and Java are large and profound, and can't really be discussed in the same category, nor do they have anywhere near the same order of magnitude of productivity improvement. That being said, it is quite fair to say that none of them are well-suited to software tasks in the 21st century, so making comparisons between them doesn't add anything.

One way to look at this is that SW and especially programming languages have not scaled with Moore's Law -- and for various reasons, the most popular and most used languages today are still quite mired in 60s notions of language and programming (so the few real advances in the 70s are not being used today). And today is 30 years after the 70s, so the situation is dire.

It's not so easy to completely reinvent programming, even though that is what is needed. There are about a dozen ideas that are not now used in a deep and fundamental way in programming languages and operating environments that could have great potential to get part way up the mountain. We've just received substantial funding from NSF and other sources to try to make something that would be a Moore's Law jump in software if successful. But my favorite quote in computing is from Don Knuth: "Beware of the above code, I have only proved it correct, but have not yet tried it" In other words, we have to do real implementations in computing to have any real sense of whether ideas might be good or simply mediocre (as most ideas are).

SDF: Being both a computer scientist and a musician, what's in common between programming and composing? How to discover a solution space that could perfectly matches a given problem space assuming there is a solution for the problem?

AK: To me those two questions don't go together. Although there is some problem-solving in the practicing of any art, Art is not primarily about problem solving. And the aesthetics of the two arts are rather different, though both can have great beauty. One meeting ground is similar to the aesthetics of mathematics, the relationships and how things can fit, the surprise of simplicity and profundity in the same forms, etc.

SDF: Could you elaborate to us how you invented Smalltalk and Alto among others in brief? Where did your creativity come from at that time? Is that possible for such creative process be formulated?

AK: These are well documented in the "Early History of Smalltalk" essay that I wrote for the ACM. Please let me know if you can't obtain this.

SDF: In your current role as President of Viewpoints Research Institute, you are working on improving both general education and understanding of complex systems, especially through the use of new inventions in interactive constructive computing, do you think education reform in the US is more important than other reforms such as retirement plan reform, healthcare reform and immigration policy reform? If yes, why is that?

AK: If we think of education (as opposed to "training") as primarily being about helping learners become more aware, gain more perspectives, learn to think better (and to realize when they are not thinking well), etc. then "education" is far more important than any of the other reforms that you mention, since it is required for any reasonable thinking to be done about almost any issue.

SDF: With the rapid advance in hardware engineering, software engineering, system engineering as well as mechanical engineering, thanks to computer-aided technologies such as EDA, CASE, MCAD among others, will it be possible true AI be a reality? In other words, robots might design and (re)produce robots by themselves without any human being's intervention. From ethnical point of view, should such innovation automation be prevented even before it gives birth

AK: I don't think there has been a rapid advance in software engineering, etc. And, computer intelligence like biological intelligence (like life itself) is much more an architectural than an engineering project. That is, engineers can build arches, but thousands of years of engineering were done before arches were invented. It is more likely that fake intelligence will be produced and believed in. And this is quite in line with the fake intelligence of most human beings (and their belief in it). Early on, we should worry more about the ethics of human behavior and the nature of our own artificial intelligences.

Back to Home Page