We’ve Run Through 4 Cultural Ranks, Are We Ready for 5? [cultural Evolution, Progress]

Cultural evolution is one of the continuing topics at New Savanna; it’s something I’ve been interested in since graduate school. And, going back to graduate school, I’ve developed a particular account of the long-term course of cultural evolution in terms of something David Hays and I have called cultural ranks. In writing and publishing about cultural ranks, Hays and I have talked of four ranks, as follows:
Each rank is articulated and consolidated by an overall architecture grounded in a specific cognitive invention, speech, writing, calculation, and computation, respectively.
In our conversations back in Buffalo in the 1970s and later in New York City, we sometimes speculated about a fifth rank that would have been just emerging. We may even have had a name for the cognitive invention; ‘metagramming’ I believe we called it. But we never could figure out just what metagramming might be, though we did a bit of research about it – see Hays and Benzon, Metagram Software – A New Perspective on the Art of Computation [1]. But even there we only talked of four stages, not five (pp. 15 ff.), though Hays did mention the possibility of another level (p. 17).
I still don’t know what metagramming is, but I do want to talk about a fifth level, or rank, of cultural evolution, because I’m pretty sure that’s where we’re headed, either that or the collapse of the world order into who knows what for who knows how long.
We’d argued that computation governs the architecture of rank 4 cognition [2]. In talking of computation we conflated three things: 1) the abstract theory of computation, e.g. Alan Turing’s work, 2) the physical devices of all kinds that carried out computation, but particularly digital computers, and 3) the software needed to run digital computers. The theory of computation is what it is – and Steven Wolfram, it seems, has managed to make it into a theory of everything – but I note that Turing’s conceptual model, of an abstract tape-reading device, along with some insights from von Neumann, showed us how to implement open-ended general computing in a physical device, the stored-program electronic digital computer. The most spectacular advances in computing have been in hardware. A current day smart phone has more computing power than a gymnasium-sized vacuum tube computer from the 1950s.
To be sure, there have been advances in software as well. We know how to implement many virtual devices in software and have developed a plethora of programming languages at various levels of abstraction, from machine code to high-level scripting languages. But programming is still problematic and software inevitably seems to be buggy. We’ve got to do better.
And in the next era we will. Just how we’ll do it, I don’t know. But we will. Here’s how I see things going: We have the origins of rank 4 conceptualization in statistical thermodynamics and evolutionary biology in the late 19th century. Other scientific and technical advances followed and we went up the rank 4 growth curve through the end of World War II and into the 1950s. That’s when digital computers were first developed. I’m now placing the whole development of computing and information technology from the 1950s to the present on the rank 4 plateau. This is the consolidation of rank 4 thinking, including cultural, economic, and political structures and institutions. The move up that curve saw economic growth. Correlatively, the economic stagnation that has so bothered Tyler Cowen and other economists set in during the consolidation or plateau phase.
When do we start moving up the rank 5 growth curve? Perhaps we’ve already begun, perhaps we’re at the toe. We’ll only know in retrospect.
Notice the position of artificial intelligence in the diagram. It encompasses the consolidation phase and goes partway into the rank 5 growth curve. Whatever it is that allows us to move up the curve, that I’m just calling ‘Factor Ω’. It’s whatever Hays was looking for when he talked of metagramming. Artificial intelligence itself is not some one thing, like speech, writing, calculation, or even computing. It’s a bunch of computing techniques, including these very peculiar techniques that don’t do what the programmer specifies, but rather learn for themselves how to function in a given domain. The programmer, of course, has to design and implement the learning algorithm and provide the AI engine on which it can learn.
But it is the AI engine itself that’s doing the learning. If we can figure out how to crank that up to 11, we’ll see magic. Well, not really. But the duck will become a rabbit, and we’ll be doing new and even more interesting things with computing engines that will have become semi-autonomous beings of a new kind. They won’t be super-intelligent computers, whether benevolent or malevolent; by the time we get to that level we’ll have forgotten all that Singularity nonsense. We’ll still be human beings, albeit with powerful new conceptual tools, and we’ll have these new computational devices with which to deploy and explore these tools.
References
[1] David G. Hays and William L. Benzon, Metagram Software – A New Perspective on the Art of Computation, Rome Air Development Center Technical Report, RADC-TR-81-118, 1981, https://www.academia.edu/34341841/Metagram_Software_A_New_Perspective_on_the_Art_of_Computation.
[2] William Benzon and David Hays, The Evolution of Cognition, Journal of Social and Biological Structures. 13(4): 297-320, 1990, https://www.academia.edu/243486/The_Evolution_of_Cognition.