I feel like PbtA fits with contemporary thinking on information architectures, that avoid complex systems and entanglements across entities formed by rules. Something like a microservices architecture: each move does one job.
I can ask if the current set of rules meets all of my requirements? That is the kind of test my dwarf wizard example represents. If my requirements include a dwarf wizard and the current set of rules doesn't support it, then the set is incomplete as measured against my requirements.
One way to achieve completeness would be to limit my requirements. But TTRPGs like DW are open-ended: I feel able to enumerate requirements endlessly. Thus no finite TTRPG text can be complete by this measure. And describing a procedure for completion does not make it complete. That's what the rough beast slouching in alludes to. GM-fiat can be used as one such procedure.
I am going to say I don't find this analogy compelling at all. Microservices simply provide some capability, which business processes can invoke as-needed. So, yes, a move in DW could be seen as a 'chunk of code', but its not arranged in a service fashion where you MUST invoke some move in order to 'get service'. DW works PERFECTLY FINE without ANY moves. It might not be as interesting a game in that case, but it DOES work. Why? Because DW is a classic example of the design pattern also used in 4e, Exception Based Design. That is, every single instance of operating the core loop of the game is exactly the same, and only when some move's fictional or mechanical trigger conditions are met, does it introduce itself as an exception to the existing play loop. Thus NORMALLY if a player says "I do X" then X happens, unless it is deemed impossible. If a move matches the situation, then instead a more specific exceptional rule is invoked, which is "execute the procedure contained in move X." Note that there's NO restriction, potentially on what move X could be.
This is wny PbtA is so darn flexible and extensible, because it is just a set of exceptions, coupled with some core pure mechanical rules like how harm works, what tags mean, etc. that are customized to each agenda.
Above I wrote
So this is a different test of completeness. Above I gave the example of deciding how much is "a lot" in the Ritual move. In this case, it's incomplete because a game parameter is left undefined. It's not that there isn't a move doing the job of Ritual, it's that the move itself is incomplete.
Mmmmmmm, I still feel like this is just a slightly variant case of the old "figure out what the fiction allows/demands here" question that arises whenever an action is taken at all. No, we don't know what 'a lot' means, but it is clearly tied to the fictional context. Its a rightward arrow from mechanics to fiction, saying "take a lot of the character's gold away." Given the rest of the agenda, and that this is a GM move which the GM agenda covers, this question can be answered in a fairly principled and definite way. What that answer is will be highly situational.
I mean, I get what you are trying to say. I think there are simply some things that can only be discovered, not quantified ahead of time. This is also merely a matter of degree, as if Ritual said 'half the character's gold' you wouldn't call it incomplete, so this is at best an extremely fine line you are drawing. Finally, its very interesting that this is about GOLD. Remember, gold is not really a big focus of Dungeon World. Its a purely fictional kind of thing that PCs may (or may not) care much about. So taking 'a lot of it' seems to me to be more like saying "this should be an inconvenience to the characters later on" more than being some 'fiat thing', as again DW exists as a recipe for narrative, not for casting magical spells.
To me, this shows why I should feel comfortable with incompleteness. I agree with your extensible framework characterisation, and that doesn't make Ritual complete. Rather it is the incompleteness of Ritual that gives it versatility.
And I think it is the exception based design that does that work more than anything else. This here is just a way to state a goal in non-quantitative terms.
I would again say that it shows incompleteness is an
advantage in TTRPG rules so long as
- What is afforded by the rules meets my highest priority requirements, and
- I have a satisfactory procedure for updating the rules to meet any new requirements
I can like the DW procedure and dislike that in Pathfinder. That doesn't make DW complete and Pathfinder incomplete, but it does mean that if using the latter I have no procedure that satisfies me for sustaining completeness against requirements.
I don't know PF well, but in 3.x if there's not a rule for something, there's no exception based design at all, you simply have no rule! There's a sort of a general "make ability/skill checks for things" rule, but it is so open-ended and vague in its application as to be virtually meaningless. The GM can call for one ability check, 2, 5, or 100 before declaring a situation resolved! 5e similarly, though perhaps the guidance on intent is articulated better (I am not much of a student of 3.x and certainly not of PF though).
So, to me there is really a FUNDAMENTAL difference in terms of the 'completeness' of, say, 5e and DW. 5e is undoubtedly incomplete. DW is complete enough that you are likely to never play it to a point of real ambiguity, and there's no 'missing' rules in any substantial sense. 4e BTW achieves the same thing by virtue of SCs, as any situation can be resolved via a finite number of checks, unambiguously. This is why HoML eschews all non-challenge checks, their existence as a standard process of play can create situations outside of the rules.