I mean, for one thing, because they still work.
I mean, first of all, there speaks the voice of inexperience. But secondly, if it really was a 20% chance to end or win a combat on the spot, that would still be pretty darn good.
If I have two actions, one of which is pretty much guaranteed to make concrete actual progress, and one of which has an 80% chance to do
literally nothing, I'm going to take the concrete progress--and in essentially all contexts, that's what you
should take.
Especially because for the vast, vast, vast majority of these spells, it's not "20% chance to end the fight." It's "20% chance to make the fight much easier." But killing/incapacitating
one thing reliably is almost always more progress than having an 80% chance of doing absolutely screw-all.
Depends on what you are going for. If your primary goal is balanced gameplay with uniform spotlight sharing and facilitating challenging gameplay then maybe.
Not uniform, and not spotlight. I absolutely
despise "spotlight" balance. It is the most effective deceptive design goal ever conceived. It hoodwinks designers everywhere into thinking their games work in ways they demonstrably won't once people actually start playing. People throw around the phrase "white room theorizing" with casual abandon, but "spotlight" balance
literally is a pure-theory concept that actively clashes with how real people play games.
In being a cooperative role-playing game, the game should be
equitable (each of the participants contributes approximately the same
amount to the group, allowing for statistical spread) and
functional (each of the participants always has something meaningful to contribute, even if it isn't the most meaningful thing one could contribute.) Uniformity is the worst, least-interesting, least-effective form of equitable game design. It should be avoided unless there is no other alternative, or the place it is being used is so fundamental that uniformity provides some other kind of value to offset its negative qualities. The unified d20 mechanic, for example, is a place where the advantage of consistency is extremely strong, and where the fact that everyone does things the same way is not a problem, because it's used to facilitate other things, not as an end in and of itself.
But if your primary goal is to simulate the magic of story and fairy tale, then absolutely you don't cut down the spell list because your more interested in a setting where wizards do big flashy things and work wonders and miracles than you are in only whether or not you are achieving some easy to see balance between classes.
Except that,
again, all these allegedly "big flashy things and work wonders and miracles" won't happen in play, or will be the ONLY thing that happens in play, because dominant strategy is a thing.
You can still have wizards who work wonders and miracles
if you design the game to actually support that. D&D has, historically, been absolute garbage at actually doing that, because either magic is THE solution and everything else is nearly-pointless window-dressing, or magic is about as effective as a wet fart and you're forced to use mundane solutions.
But we can do better. We
can design games that have wonders
and mundane solutions. It will be hard! Any good game design
should be hard, because if it were easy we'd almost surely have done it by now. (This is why it is impressive to develop new simple mechanics that
are good game design, e.g. 13A's Escalation Die--they almost always represent a new perspective or clever thought not previously considered.)
Fundamentally, as a guy that has ran 3e for a very long time, your PTSD from whatever bad game experiences you've had and the resulting chip on your shoulder about it just doesn't interest me much. If you want to talk about 3e in some sort of constructive manner, then I'm all for it. But if this is just going to be thinly disguised edition warring, then I'm not interested. I don't agree with you - 3e has some warts, but it is at a fundamental level the best designed edition of the game. Enjoy your DW.
At a fundamental level, it literally allows you to produce nigh-infinite loops and do utterly ridiculous things like self-bootstrapping spellcasting. It allows things like the "
locate city bomb," the aggressively hegemonizing ursine swarm (aka Druid with Natural Spell, a PHB-only character!), DMM cheese, and all sorts of other nonsense. Or, y'know, just the Leadership feat, all by itself, the feat almost every 3e DM bans because it's stupidly overpowered.
Yes, the idea of "roll 1d20+modifiers" is functional, and components like BAB are useful (though iterative attacks, not so much), but by the time you've stripped 3e down to the parts that actually work consistently, you've removed effectively everything that makes it a
game. You have to overhaul feats, classes, ACFs, spells, magic items, monsters (
especially templates), races, and even to a certain extent the mundane equipment. Keep in mind, 3e is the edition where a 10' ladder is less expensive than two 10' poles...so you can literally generate infinite wealth purely by buying 10' ladders, stripping off the rungs, and selling the two 10' poles that result.
It was designed to be naturalistic. Unfortunately, in the hands of even a moderately clever player
using the PHB only, it quickly produces anti-naturalistic results. This is not about edition warring, or at least that isn't my intent. I get, very much, that there's a large chunk of players who adore naturalistic design and who feel that no other edition of D&D has successfully pursued it. But 3e is a
bad example of naturalistic design. There are zillions of ways to make a better game while still keeping it naturalistic. I freely recognize that naturalism isn't one of my key concerns (for me, it's
groundedness, which is a very different beast; unnatural things can be quite grounded if they properly establish their precedents.) But even for someone who
does prioritize naturalism, you can (and IMO should) ask for
much better than what 3e offers.