Some conjectures on the nature of magic, the invention of new spells, etc.:
Whatever makes magic work is basically mechanical but smart enough to be sensitive to human intentions.
It has to do something broadly comparable to simulating the universe.
It is equipped with defensive measures of one or more of the following kinds. (a) Some things—e.g., certain kinds of escapades involving Time-Turners—would require excessive, possibly infinite, amounts of computation, and it won’t allow that. (b) Some things—again, including Time-Turner abuse—actually lead to contradictions. Maybe those are caught in advance and prevented, or maybe those simulations just end when the contradiction is reached. (c) There are deliberate countermeasures against anything that looks like it’s trying to marshal too much power, in case it’s an attempt at hacking the underlying computational substrate (or whatever it is that makes magic work).
That is one reason why inventing new spells is dangerous. You might accidentally produce a contradiction or an excessively expensive simulation, or you might do something that looks like an attempt at becoming a god, in which case you—or your laboratory—or your country—or your universe—is terminated with extreme prejudice.
It is also why the most powerful magics—the rituals—require permanent sacrifices: something that involves a permanent sacrifice is harder to apply recursively in some weird way that turns it into a Become God spell. (It’s like the way that the class of functions you can compute using only loops whose number of iterations is bounded in advance is smaller and easier to prove things about than the class of all functions you can compute.)
Whatever-makes-magic-work wasn’t programmed in advance with any particular spells, potions, etc.; it’s a general-purpose magic substrate. All the specific things have been brought into being by human beings—it’s a gigantic game of Nomic. That is why, e.g., all the spells have names that were obviously thought up by humans.
The words, gestures, etc., are likewise basically under the control of the inventors. That is why they can be stupid things like “Wingardium Leviosa”, or require you to say “Oogely boogely” with just the right timing. (Perhaps the precision required is up to the inventor, or perhaps whatever-makes-magic-work decides how much precision to require on the basis of how powerful the magic is.)
Bringing a new magic into being isn’t fundamentally difficult [EDITED to add: that would explain why there is such a thing as “the usual evasions” when someone asks precocious questions about spell creation]; perhaps it just requires forming a certain sort of intention and then demonstrating what actions (etc.) are required to bring about the desired effect. The difficulty and danger come (a) from the protective measures already mentioned, (b) from the risk that when you tell whatever-makes-magic-happen what you want you might fail to be specific enough and get something that meets your stated spec but is disastrous, and (c) from restrictions like one I mention below.
Maybe the heritability of access to magic is itself a thing that was decided by (a much earlier generation of) humans, perhaps as a deliberate measure to limit the number and increase the cohesion of those designing new magics.
How would this sort of general-purpose Everything Machine likely be set up, so as not to produce a universe of total chaos as everyone starts inventing their own new spells? (Even assuming no one is able to do anything too destructive or internally inconsistent, on account of the sort of restrictions I already mentioned.) Probably by having a requirement that new spells be consistent with existing ones, in some rather broad sense of “consistent”. That would be another reason why inventing new magic is difficult. It might also give another reason for things like the Interdict of Merlin: an opportunity for old magic to be “garbage-collected”, freeing up regions of possible-magic-space for new generations to explore.
I don’t assign a super-high probability to any of these, but I find each more plausible than any other similarly-specific alternative I’ve thought of.
Some conjectures on the nature of magic, the invention of new spells, etc.:
Whatever makes magic work is basically mechanical but smart enough to be sensitive to human intentions.
It has to do something broadly comparable to simulating the universe.
It is equipped with defensive measures of one or more of the following kinds. (a) Some things—e.g., certain kinds of escapades involving Time-Turners—would require excessive, possibly infinite, amounts of computation, and it won’t allow that. (b) Some things—again, including Time-Turner abuse—actually lead to contradictions. Maybe those are caught in advance and prevented, or maybe those simulations just end when the contradiction is reached. (c) There are deliberate countermeasures against anything that looks like it’s trying to marshal too much power, in case it’s an attempt at hacking the underlying computational substrate (or whatever it is that makes magic work).
That is one reason why inventing new spells is dangerous. You might accidentally produce a contradiction or an excessively expensive simulation, or you might do something that looks like an attempt at becoming a god, in which case you—or your laboratory—or your country—or your universe—is terminated with extreme prejudice.
It is also why the most powerful magics—the rituals—require permanent sacrifices: something that involves a permanent sacrifice is harder to apply recursively in some weird way that turns it into a Become God spell. (It’s like the way that the class of functions you can compute using only loops whose number of iterations is bounded in advance is smaller and easier to prove things about than the class of all functions you can compute.)
Whatever-makes-magic-work wasn’t programmed in advance with any particular spells, potions, etc.; it’s a general-purpose magic substrate. All the specific things have been brought into being by human beings—it’s a gigantic game of Nomic. That is why, e.g., all the spells have names that were obviously thought up by humans.
The words, gestures, etc., are likewise basically under the control of the inventors. That is why they can be stupid things like “Wingardium Leviosa”, or require you to say “Oogely boogely” with just the right timing. (Perhaps the precision required is up to the inventor, or perhaps whatever-makes-magic-work decides how much precision to require on the basis of how powerful the magic is.)
Bringing a new magic into being isn’t fundamentally difficult [EDITED to add: that would explain why there is such a thing as “the usual evasions” when someone asks precocious questions about spell creation]; perhaps it just requires forming a certain sort of intention and then demonstrating what actions (etc.) are required to bring about the desired effect. The difficulty and danger come (a) from the protective measures already mentioned, (b) from the risk that when you tell whatever-makes-magic-happen what you want you might fail to be specific enough and get something that meets your stated spec but is disastrous, and (c) from restrictions like one I mention below.
Maybe the heritability of access to magic is itself a thing that was decided by (a much earlier generation of) humans, perhaps as a deliberate measure to limit the number and increase the cohesion of those designing new magics.
How would this sort of general-purpose Everything Machine likely be set up, so as not to produce a universe of total chaos as everyone starts inventing their own new spells? (Even assuming no one is able to do anything too destructive or internally inconsistent, on account of the sort of restrictions I already mentioned.) Probably by having a requirement that new spells be consistent with existing ones, in some rather broad sense of “consistent”. That would be another reason why inventing new magic is difficult. It might also give another reason for things like the Interdict of Merlin: an opportunity for old magic to be “garbage-collected”, freeing up regions of possible-magic-space for new generations to explore.
I don’t assign a super-high probability to any of these, but I find each more plausible than any other similarly-specific alternative I’ve thought of.