Thank you! Just to preempt some possible bad words. I think the hard FK hypothesis "is solid". I now realize that is a poor choice of words. I am too used to scientific thinking. First I want to divert attention to me acknowledging that this hypothesis is disputed. I want to emphasize I think this dispute is real, and unresolved. When I labeled the hard hypothesis as "solid", it was to a large extent as a contrast to the hard amateur hypothesis.
What do I mean by "solid" then? I am actually not sure. Introspection indicated the main reason might be that I have yet to see any compelling evidence falsifying this. However it should be added that I have not sought such evidence either, as I have never been in a position to have to take stance in this dispute.
But I am also keenly aware that the hypotesis in itself is hardly falsifiable. A central problem is that "better simulation" is hardly ever well defined. It was pretty well defined for the kriegspeil case tough, however even there setting up a good experiment for trying to falsify the hypotesis would be very hard at best.
I've touched on most of this sprinkled through the response to your final paragraph, but I wanted to say that I, too, am rather used to scientific thinking (my primary training is in physics, especially quantum physics, I only pursued a
minor in philosophy). As someone pretty well steeped in that field--where statistics, probabilities (or, more commonly, "probability amplitudes"), and uncertainty are baked into its very heart, even in my preferred interpretation!--the idea that an "expert" on a scientific discipline would be
consistently superior to, say, Maxwell's equations or quantum field theory is...well, the word "poppycock" comes to mind. It's just not tenable. Scientific thinking, for good and (as we have been slowly learning over the past ~80 years, give or take) occasionally for ill, is
all about the idea that humans are actually rotten garbage at a number of extremely important tasks, such as statistical thinking, data collection, and the specific types of geometric reasoning required for important scientific disciplines. (I will note, though, that uman "probability instinct" exists, and it is actually pretty decent when restricted to the kinds of probabilities our ancestors evolved to need to reason about; unfortunately, those instincts are
very strong and don't suddenly go away when we move outside the assumptions baked into them, and thus they are very apt to lead us astray.)
However I do find myself having a bias toward using hard FK as a working hypotesis. I think one reason is that the obvious counter claim "It is always possible to find a complex enough mechanical system that can provide a better simulation than a human expert" is not falsifiable at all. Hard FK can at least be falsified in theory by for instance by producing a mechanics that is generally accepted to provide better results than an expert. Such mechanics are routinely used in many fields. As such, the absence of any mechanics generally recognised as better than even amateurs in the field of TTRPG, is in itself not a given.
Whereas I find myself with a (pretty significant) bias
against it. Secret knowledge is inherently unexplored and untested. Intuition is a useful tool in many places, but much of what goes on in any kind of arithmetic or statistical structure
cannot rely on intution, because intuition is often simultaneously
very wrong and
impossible to check. When your rulebook is invisible, it's impossible to check it for errors, more or less, and errors creeping in through an invisible rulebook cannot be isolated for testing and refinement.
Rules-systems, by being fixed things, necessarily cannot expand beyond their current definition unless we engage in design work. That is the nature of the beast. But by
being a fixed
public thing, a rule-system
permits criticism. An invisible rulebook not only does not permit criticism, it inherently opposes criticism. You cannot critique what you cannot see.
I would finish with giving one rationale why the hard FK hypotesis might be correct for the scope of TTRPG for all practical purposes, that I think might resonate with you. That is your exelent point related to how the subjective notion of groundedness is essential to the sense of quality of the TTRPG "simulation" (at least for many). A human referee are able to gather feedback from players regarding their subjective sense of groundednes and incorporate that in their rulings. It is hard to for me to see how to practically incorporate something similar into a traditional approach to mechanising simulation at least.
Okay, but that's a significantly different claim. That is, your "hard FK hypothesis", which I'll call "A", more or less as stated, was:
"A: Any expert can produce superior simulation results (abbreviated SRs) compared to any rule-system."
What you're now proposing, call it A`, is: "A`: Any given expert can produce
sufficiently adequate SRs
more often than any given rule-system."
The former is an extremely strong (as in strident) claim. Even without the criticisms invited by the squishy terms "expert" and "superior" (e.g., "what qualifies someone as an expert?" and "how is 'superior' defined?"), this is making a sweeping existential claim. That is, the equivalent contrapositive of A is: "There does not exist any system that can produce superior SRs compared to any given expert." All I need to do is provide a single counterexample: a system that can, in fact, produce superior SRs
some of the time compared to even one single expert.
Your softened A` is in a slightly better position, as it makes weaker claims, but it is still not on "solid" ground as I see it. That is, its contrapositive is: "There does not exist any rule-system that can produce sufficiently adequate SRs more often than any given expert." But the reason the ground is still shaky is because, despite being a weaker claim, it has actually weakened the requirements for its disproof in lockstep, because now, I don't even need a system that is
always superior, nor even always sufficiently adequate! I simply need a system that produces sufficiently adequate SRs more often than even
one single expert, a bar that I should hope we agree is not at all difficult to clear.
Completely separately from that, I don't really take falsifiability arguments all that seriously. Popperian falsifiability was a noble goal, but I don't think it achieved the target it set out to achieve, especially because there are things held to be good science (such as Everett's Many-Worlds interpretation of QM or the Copehagen interpretation thereof, or the theory of evolution) that aren't falsifiable (the latter explicitly so by Popper's own admission/assertion), and because there are things that are agreed to be bad science but which make both falsifiable
and actually falsified claims (e.g. astrology, which makes specific, falsifiable claims). Moreover, the above criticisms that I intentionally left aside--the definition of "expert" and "superior", and perhaps by your own arguments "simulation" as well--specifically open this hypothesis to
non-falsifiability, because one can simply argue that if person X failed to produce results superior to a given rule-set, then either person X wasn't actually an expert, or "superior" was defined improperly, or person X wasn't trying to "simulate" (or was not "simulating" in the right
way), and thus the claim is still true because the contradictory evidence has been excluded by sufficient tweaking of the input assumptions.
Note, however, that none of these disproofs result in the claim you
seem to be thinking I am defending, which I will call the "strong designer hypothesis": "There exists at least one rules-system which always produces superior simulation results than any expert could produce."
I don't hold this hypothesis, and think it almost as bad (not
quite, but almost) as the "strong FK hypothesis". As before, all this would require is finding a single expert, and as experts can adapt much faster than tested, developed rules-systems can, it is functionally guaranteed that, for any
given, fixed rules-system, a singular expert could specifically
train themselves for being superior to that system--meaning, the expert becomes an expert
at that system, discovers every possible way that it
could go wrong, and then engineers a situation where it
must go wrong, so that said expert can then declare victory. Since the system can't adapt but the expert can, it is always possible that
at least one expert could, at least in theory, eventually become better--simply by always applying that system except in the identified edge cases, however infrequent those may be, and then by effort and manipulation
ensuring that one such edge case actually occurs.
Instead, my claim is as follows: "It is always possible to design a rules-system such that any given specific expert,
without further training, would always produce equivalent SRs to that system." Speaking more broadly,
given some minimum level of desired efficacy, I believe it is always possible to design a system which meets or exceeds that level. Call this the "weak designer hypothesis", if you like. If said expert subsequently goes out and refines their knowledge (perhaps, for example, by thoroughly learning said rules-system and finding as many places as possible where it goes wrong), then it is certainly possible for that expert to exceed any fixed rules-system, that seems blindingly obvious to me. But while a
single rules-system can't adapt,
designers can. The designer can pick that expert's brain, using their expertise as the basis for modifications to the original system--and since it must be public knowledge (that's the whole point of being a
system, after all), it can be critiqued by many minds, not just the one expert's mind.
This does result in the conclusion that neither experts nor systems are inherently superior. Instead, it creates a kind of tennis match. In the case of
kriegsspiel, designers lobbed the opening serve, and
eventually experts lobbed back. But it's been many, many, many years since then. We have come to realize, for example, that the
ideal of tactical infinity is not actually something that ever really manifests, nor anything even truly close to it. My own arguments, here and elsewhere, have demonstrated that "invisible rulebooks" can in fact be a liability just as much as an asset. That it is reductive, not to mention dismissive, to write off any rule-system solution as "merely" invoking mechanics without thought or care or creativity (a common, base canard flung out by advocates of "FKR" approaches pretty much every time the topic comes up.)
