This was originally posted on Cohost November 24, 2022 as part of a conversation about the WIZARD PROBLEM started by @mammonmachine. To briefly summarize the discussion, the WIZARD PROBLEM is two-fold. First, "what can a spell (and by extension a wizard) do" is a way less constrained question, both mechanically and imaginatively, than "what can a weapon (and by extension a fighter or similar) do." Second, when confronting an infinite possibility space of what can happen because of roleplaying, the person who can respond to the most possible situations is inevitably more powerful, even if their responses are generally weaker. By the time I posted, others had already mentioned that spell slots/preparation was an attempt to limit this effect, while Slay the Spire and Magic: the Gathering dodged the problem by penalizing inefficient retrieval of the right tools out of a randomized deck.
The struggle I always have with the WIZARD PROBLEM is that it's... very hard to undermine it without ruining the fun. The Wizard can't do something as well as another player, so you push hard into that thing – but then what's the Wizard supposed to do? The Wizard can't last forever, so you stretch out the breaks between uninterrupted rests – but then why are you forcing the rest of the party to have such extended sustain? The Wizard can only prepare for a limited set of circumstances, so you construct scenarios they didn't prepare for – but then why are you denying the Wizard their very purpose for existence?
The WIZARD PROBLEM obviously originates with D&D Wizards, but it's also an issue in other places, e.g. the higher levels of Lancer where a squad can put together tools for just about any situation. As was mentioned above, versatility on its own is very powerful, and as a result the only real solution is to make the few weaknesses of such a character SO debilitating that they become glass cannons. Put that into a setting or story where people want to have long-running characters and plenty of character action, though... There comes a point where the WIZARD PROBLEM forces a GM to either become a true master of the system to balance encounters & rest breaks on a knife edge, or decide: do I care more about threatening encounters or continuing the story?
It's worth recognizing though that the WIZARD PROBLEM is at its strongest when A) the use of magic is risk-free and B) the story is broken into a collection of mechanical barriers, rather than a collection of narrative beats. Wild Magic-style mechanics are supposed to challenge part A by turning spellcasting into a risky endeavor – you never know when something might go wrong with a spell, so why not solve this problem conventionally? Depending on people's past experiences, though, this can quickly become "I will never cast a spell, just in case" or "gods dammit, this stupid magic system kicked in at the worst possible time and ruined everything when it should've been fine." Maybe it works for some groups where they get consistent luck, or the players just like the goofiness of random hazards springing out of their spells, but personally this sort of system on its own is rarely to my taste.
Part B just straight-up cannot be addressed in a system like D&D, because the game is simulationist & mechanics-oriented. All actions are pumped through the framework of d20 pass-fail dice rolls, modifiers, and narrow understandings of cause & effect. However... *puts on my "I don't like D&D and will always advocate for narrative" hat* Powered by the Apocalypse systems (and their derivatives, e.g. Forged in the Dark or Belonging Outside Belonging systems) are all about turning mechanical realities into narrative beats. With those systems, magic might instantly solve any problem, but the limited charges are geared towards keeping you using other abilities, or there's a risk of something else going wrong in the process. With such systems, it's much easier to say "great, you solved the problem-as-written, now let's talk about the new problems you introduced as a result..." After all, if a Wizard is picking a door lock with a spell, there's a lot more that can happen than "the spell succeeds or fails;" narrative-focused systems highlight those secondary factors.
Anyways the bolded line above is probably the biggest takeaway for most folks since I know changing systems (both in the "let's add new rules" sense and the "let's try a whole new framework" sense) is hard. There are more ways to introduce risk to magic than Random Bad Times (e.g. the inefficiencies of Slay the Spire mentioned above), but generally speaking if you're gonna have an "optimal" Wizard-type you're gonna have the WIZARD PROBLEM. End of the day: it's all in the social contract between GM & players about character power, roles, and desired experience.