ON THE CONCEPT OF 'TIME' IN COMPUTER SCIENCE

It has been said: both computer scientists and physicists do not understand time, but at least physicists admit it. The notion of simultaneity is a widely used and yet deeply misunderstood concept in computer science. The word "simultaneous", its derivatives and equivalent phrases are common, yet they carry conceptual baggage known to be incorrect from our most successful physical theories. Modeling "time" as a monotonically increasing function in a single dimensional line in real numbers unnecessarily constrains our understanding of concurrency. A refined view of the notion of time may allow us to build more reliable distributed systems. It may also illuminate improved opportunities to exploit the forthcoming abundance of cores in the computer industry.

This lecture is an inspired speculation. An opportunity to evaluate a potential shift for Computer Science. It presents an alternative model for time, consistent with relativity and quantum mechanics that has explanatory power for interference without the need to appeal to unknowable randomness or an arbitrary wavefunction collapse to explain its probabilistic nature. We will explore together, the theoretical foundations of concurrency in computer science. We ask if sufficient progress has been made and ask how these insights might be made practical, for example: to drive the utilization of cores in the computer industry.

The ideas are a result of careful speculation, to solve real problems in distributed computer systems. However, as with all nascent theoretical results, they still need to be carefully formalized and proven before the claims can be published. However, it might be ripe to begin discussions on how this theory succeeds or fails in addressing the core issues of time, not just in physics, but in directly in computer science, where it might be able to contribute to an understanding of how to increase utilization of cores in the computer industry.

Speaker: Mr. Paul Borrill, Founder/CTO, REPLICUS