The first concepts of DP [1] and physics incorporated a rational cosmogony, modeled the concept of temporal evolution well, and modeled certain laws of physics in an ad hoc and unattractive way. The idea was no more than that of viewing fundamental physics as a computational process on an ordinary computer. However, it established two of the most important features of DP, Universality and Digital Cosmogony. Working with that model led to the conviction that computational universality had to be a property of physics. Whatever the most basic laws of physics were, it had to be true that they were computationally universal [2] , otherwise computers as we know them would be unable to exist. The cosmogonical problem of physics disappears in Digital Philosophy. The puzzle is: given the laws of physics (especially conservation laws), it is a stretch of the imagination to conceive of the universe (space-time, its contents and laws) being created out of nothing, consistent with those laws. In Digital Philosophy, the computation that is physics runs on an engine that exists in some place that we call “Other”. There is no reason to suppose that Other suffers the same kinds of restrictive laws present in this universe. Computation is such a general idea that it can exist in worlds drastically different than this one; any number of regular spatial dimensions or almost any kind of space-time structure with almost any kind of connectivity. There is no reason to think that concepts of matter and energy, of conservation laws and symmetries, and of beginnings and endings are applicable to Other. Digital Philosophy even offers us a few common sense conclusions about Other. In Other the RUCA was loaded with the initial conditions and the computation was put into operation. DP has no problem with cosmogony. [3]
Early concepts of DP, dating back to the 1950’s, easily encompassed solutions to three problems in physics; Process, Universality and Cosmogony. These are not the kinds of problems that physics students are made aware of. Every operating computer implements a system characterized by temporal evolution. A computer process is a temporal sequence of states, each a function of the previous state. This is also true of physics. The universality argument is two-edged; universality is required of a DP model of physics, and if there is an exactly correct such model, any and every universal system (computer) would be able to implement that model exactly. Finally, as will be explained, every kind of computer model of physics finesses the cosmogony problem; it ceases to exist.
These were and are important concepts and problems which were not generally recognized in the 1950’s and which are not generally recognized today. Once DP included computational universality it was obvious that theoretically every and any kind of discrete physics could be modeled exactly by any Universal Computational DP model. The only question was whether the model would be direct and simple, or baroque and contrived. Those three early results, on Process, Universality and Cosmogony provided impetus for the further development of Digital Philosophy, despite the fact that the earliest DM models had nothing else to recommend them.
It was a great advance to go to Cellular Automata as a basis for DM models. The author’s first CA model, the XOR rule, was a step forward in being a direct model of a space-time that was locally Cartesian. It replicated patterns and had a wonderful form of superposition, but it was a step backwards in that, at the time, we could not see how such simple rules could model universality, be reversible or be consistent with any other aspects of physics.
After being shown early DM cellular automata models and understanding the concept of Digital Philosophy, Marvin Minsky issued a challenge: “Find a simple CA rule that propagates with asymptotic spherical symmetry as opposed to the typical 4 fold symmetries produced by rules like 2D XOR. This was a brilliant insight, combining a known property of physics with what might be possible within the then primitive understanding of CA models. Minsky’s challenge took years to get done but it was an important step in the evolution of DP. The early CA models of DP faced seemingly insuperable obstacles. No simple CA was known to be computationally universal [4] , and it was common knowledge that all simple universal models of computation were known to be irreversible. Those 2 facts alone were enough to cause almost any rational person to abandon DP as a sensible model for physics.
It was ridiculous to assume that something as arcane as a Turing Machine or a commercial computer could be a substrate for physics. The idea of an irreversible computer modeling the reversible laws of physics was obnoxious. Progress on DP was put on hold in order to solve those two problems: inventing simple universal cellular automata and inventing simple models of reversible computation. Finally, a third problem emerged; we had to do more than discover a model of reversible computing, we had to gain understanding and familiarity with every aspect of reversible computing. After completing the first of these 3 tasks, there appeared, out of the blue, Conway’s remarkable “Game of Life.” It was a CA and it had stable particles that moved! This was another fantastic step that arrived unexpectedly. Gosper was able to demonstrate that the Game of Life was actually a UCA [5] . We then discovered Konrad Zuse, who in the late 1960’s, came up with a similar general concept of Digital Philosophy, and published a book called “Rechnender Raum”. We invited him to come to MIT where he found the ideas in his book appreciated for the first and only time during his life. (According to Zuse.) In 1974 the author invented Conservative Logic, as a physically correct model of reversible computation [6] . Following that invention, the author and his students developed a great and complete understanding of issues related to reversible computation. Along the way, the author invented the Billiard Ball Model of computation which served as the model for the first RUCA developed by Norman Margolus. Finally all of these tasks were accomplished by the mid 70’s by the author who, with his students and colleagues, including Roger Banks, Norman Margolus and Tommaso Toffoli [7] expanded and elaborated these concepts. Charles Bennett had independently discovered a Turing Machine model of reversible computation, but his motivation and methodology were unrelated to Digital Philosophy. Gaining a thorough understanding of reversible processes took a number of years. The development of Digital Physics and Digital Mechanics has been a strange process but we have no shame.
To summarize, Digital Philosophy carries atomism to an extreme in that we assume that everything is based on some very simple discrete process, with space, time and state all being discrete. The workings of DM are like the workings of a hypothetical computer processor: there are bits of information, there are discrete instants of time and there are rules that govern how the state at one instant is time is translated into the state at the next instant of time. Today, our understanding of how to simultaneously incorporate Universality and Reversibility into CA models (RUCAs) is so advanced, that we take it as a matter of course that all our DM models incorporate these 2 principles. 30 years ago there was considerable doubt as to whether or not a RUCA was possible. Most of those who thought about the problem had assumed that reversibility and universality were incompatible concepts.