Do you want totally predictable magic, in the style of Vancian-casting D&D, where what you readied you cast, and it will be guaranteed to go off?

Or would you rather it be unpredictable, in the style of DCC, where a roll determines whether the spell even works or how well?

Do you want your magic to be zero-cost; or would you rather that it drained hit points or ability scores to cast? Or maybe something else, some deeper cost to make magic effective?

I tend to like reliable. But resistable, counterable, breakable, and interruptable. Which is why I like D&Ds system.

In my own system there was also a risk of channeling too much magic and risking burnout, or overextending yourself and risking passing out. There was also the chance of a spell interrupted having unforeseen effects.

I prefer some options in the system on all those factors so that I can more easily customize to the setting.  It's one of the few things where I want some significant variety as we switch campaigns.

This is one of those questions where setting matters. If we are talking Sword and Sorcery or Horror such as Conan or Call of Cthulhu, it seems fitting that magic has a price, often corruption or the chance to backfire. When you look at something more high fantasy like Lord of the Rings, or most D&D settings then I'm fine with magic just being what it is, this is a thing that happens. Urban Fantasy can really go either way depending on how dominant you want magic in the setting.

Risks are fine as long as they make sense in context and contribute to making the game interesting and fun. For an example of this not being achieved, in one Fantasy HERO supplement, all magic had to be bought with "Requires a Skill Roll" and "Side Effects." But because PSLs weren't allowed, or at least were strongly discouraged, the effective skill rolls were often in the 9- to 6- range, meaning magicians were far more likely to fail than to succeed, which would effectively mean that every magician was so deep in debt to whatever extraplanar entities that either paying off their debts dominated the campaign, or the magician needed a solo side campaign.

Depends on the setting and my mood, but I like the idea of magic draining your hit points (SPI's Deathmaze was the first time I encountered that in a game).

Unpredictable magic can be fun, but that's a tricksy balancing act depending on how far out you want to take it. D&D's Wand of Wonder is something that looks great on the page but I never see actually used. When the bugbears are bugbearing down on you, shooting a stream of butterflies at them sounds hilarious, but doesn't end with the wizard laughing very often.

I like magic to be about as dangerous as a sword in the hands of a similarly-leveled (or experienced, in the case of varying experience for different classes) character. My players seem to be happiest when most everyone can find something useful to do each round.

In general -

I like my magic powerful, rare and dangerous. I want casters to have to really be tempted by the lure of power from it, even while it might destroy them and those around them. I want it to force its users into being disciplined or die for being stupid enough to walk that path without the will required to master it.

But conversely - I want there to be threats in the world that are either due to the nature of Magic and/or can only be dealt with by magic-users. This might be magic-users themselves.

Varies heavily with setting.

In D&D-like games I generally lean towards reliable casting with a mix of at-wills about as good as a basic weapon attack/skill use/mundane tools and resource points for bigger effects (though I prefer something that recharges at X per hour of rest than getting your whole lump back once per day, even if that means you can't fully recharge in just 8 hours if you went deep into your reseves the previous day).

But I also love Mage the Ascension where magic is dangerous and unreliable; especially when you go for the vulgar effects or have been pushing coincidences to the limits... but also very free form.

You could probably put my preferences on a continuum where the the more rigid the effects (ex. D&D style spell lists) the more reliable it should be (the challenge is using only the specific tools you have to solve problems) and the more flexible the magic the riskier it is to use (the challenge is managing the risk of casting vs. the costs of not doing so).

While I've really only played the two ends of that spectrum, I'm sure there's somewhere in the middle I'd probably enjoy too.

My baseline is TFT/GURPS Magic, where there is certainly a roll to cast the spell successfully, which can also affect how well it succeeds, and where the limit on spells cast tends to mainly be about how much mana/fatigue/psychic-energy the caster has available, possibly aided by apprentices, magic items, or surrounding mana level.

The worst TFT crit fail (except for certain spells) is spending all the energy (rather than just 1 on a simple fail) and failing. However after a few years with TFT we started adding crit fail tables and a whole elaborate Magic Item Breakdown system to make magic items not just a "you have a superpower now that you should always use because it has zero risk/cost" thing. GURPS came with a fairly interesting crit spell failure table, which I embellished.

Also, despising the effect of powerful healing spells on gameplay, when I even allow them, I have devised various limiting systems for them, including spell components and steeply-rising risks for using too much healing magic on a patient.

Now I tend to be interested in devising interesting magic systems, and/or playing low-magic games, and usually I do want the results to be somewhat unpredictable and for there to be some risks of consequences that can develop into interesting situations themselves.

Like electricity or fire, dangerous in the hands of the novice, the more you work with it and learn about it the better you are, the more you can do, and the less chance there is of messing up. There is always danger, however, and that danger can range from the mild (slight shock or burn) to the catastrophic (burn down the house or fling you across the room)

It depends a lot on setting. I like a variety of different magic systems.

One common option that I find not well supported, though, is having magicians who come across as truly knowledgeable experts. That is, a wizard isn't just artillery - they have insight beyond that of mortal men, and are mysterious and forbidding.

Random failure doesn't make wizards more mysterious and forbidding, in my experience. It more often makes them seem more buffoonish or perhaps self-destructive. I think the latter is fine for, say, Call of Cthulhu or some other related genres (like other pulp horror).

The problem there is for a magician to have insights and understandings others lack, and for those insights to come off as convincing, the player must also have them, and that takes a huge amount of buy-in on part of the other players when they can just go to their FLGS or online equivalent and get the same insights and understandings. It's been my experience that trying to go that road just does not work. If you're using published material, the other players will ignore your attempts at presenting your character as knowledgeable and perhaps even a bit forbidding, if you're coming up with things on the fly the GM will shut you down, and if the GM is one supplying the understandings then you're resented for being the favourite.

Quote from: Whitewings;1033054The problem there is for a magician to have insights and understandings others lack, and for those insights to come off as convincing, the player must also have them, and that takes a huge amount of buy-in on part of the other players when they can just go to their FLGS or online equivalent and get the same insights and understandings. It's been my experience that trying to go that road just does not work. If you're using published material, the other players will ignore your attempts at presenting your character as knowledgeable and perhaps even a bit forbidding, if you're coming up with things on the fly the GM will shut you down, and if the GM is one supplying the understandings then you're resented for being the favourite.

I'd agree that those are potential pitfalls, but they can be overcome, in my experience. Among published systems, I think Amber games often do well in making characters be competent experts. It's common that the GM will give players secret knowledge based on their background and powers, plus they can have things like personal shadows.


(1) The problem of being seen as the GMs favorite:

This isn't a problem if the extra knowledge is an explicit perk of the game setup, and other players get equivalent other perks. i.e. The knight gets to have in-depth knowledge and sway with other nobles and warriors, and the bard gets to have background dumps on most regions or topics.

(2) Getting shut down by the GM for making stuff up:

That's a deal of just working things out socially with the GM. This is very common in Ars Magica or other rotating-GM games, where the background is collectively generated. It can feature in others, though, like personal shadows in Amber - or having player-created enemies and/or allies that they have detailed. Both the GM and the inventing player have to discuss in advance to set expectations, and check in a bit. It depends on how good the pair are at collaborating, which can vary but isn't random.

I like magic to be exceptionally dangerous, but that danger can be mitigated (or in some instances eliminated) by careful preparation and due diligence. Akin to operating heavy machinery today. If you're not paying attention, if you're not using it safely, then the consequences can be incredibly dire (depending on what's actually trying to be done obviously), and I also don't like using it for things like fireballs, lightning bolts or other "magic projectiles". I like when magic is otherworldly, and using it for something mundane like a projectile is like using fine china and silverware to eat canned soup and hotdogs. You can do it, but it's such a waste.