While reading Paul Rosenbloom’s excellent “On Computing” I feel as though I’ve glimpsed the outlines of something big.
In the book Rosenbloom advances the argument that computing should be counted among the ‘Great Domains’ of science, instead of being considered a mere branch of engineering. During the course of advancing this thesis he introduces two remarkable ideas: (1) a ‘relational architecture’, describing how the Great Domains relate to one another; (2) an accompanying ‘metascience expression language’ which defines the overlap of the Physical (P), Social (S), Life (L), and Computing (C) Domains in terms of two fundamental processes: implementation and interaction.
Though I’m only a few chapters in I’ve already seen that his methods for generating monadic, dyadic, and polyadic tuples of different combinations of the Great Domains could be used to create a near-comprehensive list of every area of research possible within the boundaries limned by our current scientific understanding.
Let me explain: ‘pure computing’ consists of any overlap of computing with itself (C + C), and subsumes such areas as computational complexity and algorithmic efficiency analysis. ‘Mixed computing’ would be the combination of computing with any of the other great domains: computer hardware would be Computing (C) + Physical (P), a simulation of predator/prey population dynamics would be Life (L) + Computing (C), computer security and AI would be Social (S) + Computing (C), genetics/physics simulations would be Physical (P) + Computing (C), brain-computer interaction would be Computing (C) + Social (S) + Physical (P), and so forth.
A simple program could make a list of every possible permutation of C + P + S + L (including standalones like ‘P’ and pure overlaps like ‘P + P’), and you might be able to spot gaps in the current edifice of scientific research — there might be certain kinds of C + L + S research that isn’t being done anywhere, for example. With this in hand you could begin to map all the research being done in, say, Boulder CO onto the resulting structure, with extensive notes on which labs are doing what research and for whom.
(Bear in mind that I still haven’t gotten to the parts where he really elucidates his metascience expression language or the relational architecture, so these ideas are very preliminary.)
This alone would prove enlightening, but its effectiveness would be compounded enormously by the addition of an ‘autodidact’s toolkit’ of primitive concepts which, when learned together, open up the greatest possible regions of the gnostic manifold . In a post generative science guy-who-knows-literally-everything Eric Raymond briefly explores this idea. In a nutshell, the concepts from some sciences can be used in many more endeavors than the concepts from other sciences. As beautiful as it is, concepts from astronomy are mostly only useful in astronomy. Concepts from evolutionary biology, however, have found use in cognitive psychology, economics, memetics, and tons of other places. So maybe a person interested in science could begin their study by mastering a handful of concepts from across the sciences which are generative enough to make almost any other concept more understandable.
Eric and I have been in talks for several years now to design and build a course for exactly this purpose. Someday when my funds and his schedule are in sync we are going to get this done.
This relates to the ideas from section I because a mastery of the autodidact’s toolkit would allow one to dip into an arbitrary point in the gnostic manifold and feel confident that they could learn the material relatively quickly. Imagine being able to look at research being done at a major university and then get up to speed in a month because it’s just variations on concepts 3 – 6 from the toolkit .
But I think we can go even further. Based on discussions of hyperuniformity and the unusual places it appears I began to wonder whether or not there might be special branches of mathematics from systems theory, chaos theory, and possibly information theory which might not act as bridges between some of the concepts from the autodidact’s toolkit. The linked article discusses how a certain kind of pattern crops up in places as far away as the distribution of cones in avian retina and the formations of certain kinds of unusual crystalline solids.
My question is: if you had a map of the gnostic manifold, you’d mastered the autodidact’s toolkit, and you understood the relevant math, might you not have been able to hop into a research gap, spend a month or two looking for hyperuniformity, learn about quasicrystals in 1/3rd of the time anyone else would’ve required, and then glimpsed the pattern ahead of the competition? If so you could’ve had a startup in place to exploit the new knowledge by the time the first research papers were coming out.
Organizing, representing, gathering, and communicating this wealth of knowledge would be much easier with an ‘acquisition infrastructure’. Here I’m imagining a still-theoretical integration of the best mnemonics systems, a supercharged version of Anki, whatever the best knowledge map software is, matlab/mathematica (or open-source alternatives like octave), all running on a supercomputer with insane amounts of both memory and storage.
Furthermore, I want to develop the concept of a ‘drexler audit’, the baby version of which is advanced by Eric Drexler in how to understand everything. The basic idea there is rather than try to understand the details of a given field you instead use a series of object- and meta-level questions to get a firm grasp on what the goals of the field are, what obstacles stand in the way of those goals, and what gaps remain in the knowledge required to move forward.
This absolutely does not count as expert-level knowledge but it does give you the kind of overview which can prove useful in future exploration and investment.
With a map of the gnostic manifold you could choose some fields on which to perform a drexler audit and others to explore deeply with the combination of systems math and the autodidact toolkit. With a breakdown of the who/what/where/why of the research community in a given region you’d be in a position to bring the right minds together to solve whatever tractable problems may exist to give a field a jumpstart. And if you understood the economics of scientific research and the basics of investing the resulting machinery might, with a bit of luck, start coughing up wads of money while doing enormous amounts of good.
(Of course it could also crash and burn, but so could SpaceX — nothing great is accomplished without a healthy dose of risk)
I’ve said all of the above because it points to a tremendous opportunity: an amalgamation of Y-Combinator, Berkshire Hathaway, TED, and Slate Star Codex. If it works out the way I think it might, whoever manages the beast could make Elon Musk look like a lazy, sharecropping half-wit.
The STEMpunk Project helped lay the foundation for the research required. If I can make the necessary contacts and get the funds together, I’d like to flesh this out in the next five years.
 Of course I’m likely underplaying the difficulty here. Brian Ziman, perhaps the most technically accomplished person I know, has pushed back on my optimism at this point. My view is that even if it proves orders of magnitude more difficult to construct I think the Gnostic Manifold is a framework worth fleshing out.