The Dawn of a New Era – The End of Classical Computer Science

The End of Programming

Most of us are still dinosaurs, waiting for the meteor that will strike.

In the 1980s I learned to program personal computers such as the Commodore VIC-20 or Apple ][e. My professional training, which culminated in a PhD from Berkeley after studying Computer Science at Berkeley College, was based on what I’ll call \”classical\” CS, including programming, algorithms and data structures. The ultimate goal of Classical Computer Science is to reduce a concept to a program that can be written by humans, in languages like Java, C++ or Python. No matter how complicated or sophisticated an idea is in Classical CS, it can be expressed using a program that is human-readable and human-comprehensible.

In the early 1990s, when I was in college, we were still experiencing the AI Winter. AI was also dominated by classic algorithms. Dan Huttenlocher was a pioneer in computer vision and is now the Dean of the MIT School of Computing. My first Cornell research position involved working with him. Dan Huttenlocher’s PhD computer vision class in 1995 was a classic course that did not mention deep learning, neural networks, or anything similar. It was purely classical algorithms such as Canny edge detection and optical flow. Deep learning was still in its infancy and not considered mainstream AI or mainstream CS.