The Tavern

It incentivizes the player to either over-invest in invocations or just not use banish against hard enemies, and overall makes Lugonu weaker than it should be. When you really need banish to get rid of a hard unique or unkillable ghost, the failure % just represents a random and potentially very large piety tax for the action (which also costs MP). Lucy blink is convenient and corrupt can help you get runes quickly, but this mechanic leaves a really annoying gap in the AK background.

I disagree. I think it's good that banishing bigger monsters is generally more expensive, and invocations skill reducing that expense doesn't bother me. Using an MR check is a nice way to do this since it's an existing mechanic that is relatively transparent.

Do you have the same complaint for Elyvilon pacification?

Banish essentially is a hex. Casting a hex will cost you MP whether it succeeds or not. I don't see why Lugonu banish needs to be special-cased. Saying that Lugonu is "weaker than it should be" isn't persuasive. There's no clear answer to the question of how strong a god "should" be. FWIW, I think Lugonu is pretty good.

So you invest in a bunch of invocations and get your banish % up to around 50% vs some nasty unique. Great. Now the piety cost to banish is random and could be either very low or very high. You even might spend a bunch of piety and not manage to banish the monster at all. Now you're running low on piety and still have to deal with the monster. It's a terrible mechanic and you know it.

All piety costs are random. Every ability in the game that has a piety cost has a randomized one. And this is far from the only ability that can "fail" and still cost piety: I mentioned Ely pacification, there's also Makhleb's servants and several abilities that don't exactly "fail" but will frequently fail to produce a useful effect (major destruction, call merchant, any Nemelex ability with bad cards).

I agree that hexes are a poor mechanic, but removing the fail chance from one specific one wouldn't really improve anything imo.
I don't think any of these things are bad. I certainly don't think any ability should be able to reliably get rid of dangerous monsters - if it does, they stop being dangerous because you can just banish them. Damaging abilities can roll low and not kill; Imprison is at least temporary (and honestly broken anyway...). Banishment isn't going to do anything to HP unless it changes completely, and I assume you would not want it to be temporary. So banishment needs to have a way to fail, different from those two. Checking MR seems like the best way to do this to me.

Are these really analogous though? Banish is a hex that costs piety, and a good portion (most?) of the god's usefulness is your ability to use it relatively often. Makhleb abilities don't really give you zero if you know what you're doing (e.g. if you're not using major destruction when it could produce an effect that doesn't have enough range, etc.) and it wouldn't matter as much because no one has piety problems with makhleb. Piety doesn't seem to be an issue with new nemelex either but maybe I don't know because I only use stack five on escape decks now. You can get shops with nothing useful in them from gozag but I don't think this matters that much and you could have just picked food shop I guess.

If failed banish is going to be free, successful banish needs to cost more piety. Mahkleb is already extremely strong and can already spam banish.

Why is it bad? Gods can have tools for dealing with easy monsters, hard monsters and very hard monsters.
For example, Makhleb has heal-on-kills for easy monsters, destruction for hard ones and servants for very hard ones.
Lugonu has distortion damage for easy monsters, banishment for hard ones and corruption for very hard ones.

Free banishment failures will make SpAK and CeAK even easier, just kite until that 1-5% chance triggers.

are you suggesting this is a degenerate gameplay pattern for sp, ce, and vp? because they already do that with regular hexes

Yes, it is degenerative gameplay.

Regular hexes are not as powerful (subject to spell failure and armour limitations, luck with finding the book, spell slots constraint). Anyway, if there is a problem with some species, we should not add yet another problem.

Edit. I have written a reply to already deleted message so here it is as it is related somewhat.
Absolutely safe and unlimited method which requires lots of time and key presses is a perfect example of degenerative gameplay IMHO.
Kiting is absolutely safe so it should not be unlimited. CeHu's kiting with arrows is limited by ammo (at least early game), SpEn's kiting with Confuse or EH is limited by MP, FeAK's kiting with banish is limited by piety. Hex kiting is the worst here as you can just retreat upstairs to restore MP and changing banishment as OP suggests will make banishment even worse as CeAK will be able to banish any non-immune monster despite having Int 1, zero skills in magic and wearing plate armour.

i meant fast characters being strong should be distinct from whether banishment should cost piety

Accepted. Then my question to OP still stands. Why is it bad that banishment does not work vs some hard monsters? Is it bad that Slimify, Cleansing Flame or Minor Destruction does not work vs some hard monsters, should we change those too? Lugonu has an universal solution in form of corruption and is not a weak god so it does not need a buff IMHO.

Is that question for some other OP who complained that banish "does not work vs some hard monsters"? Or are you reading my posts and talking to me?

I think the large randomness in piety cost to banish a monster is not so good. For example, maybe you have invested a lot of XP in invocations in order to get a banish chance of 40% vs some unique with decent MR, yet get bad dice rolls and still must cast banish 5 times or so, or else cast it maybe 2 or 3 times and then stop casting since you don't want to lose corrupt, and accept that you have spent your piety for nothing.

This is literally the only thing I am talking about.

VAF's point is the following:

If there's no Lugonu piety for banish, and you're playing a fast species, then you can kite the monster and keep trying banish till it succeeds. This is true for all hexes, but is even more true for a hex which is "insta-kill", and which has even fewer restrictions on Int, armour choice etc. than normal hexes.

Oops, ignore this

That large randomness is exactly what I like about banishment, and I wish more god abilities had it.
If I use pacification or healing or upheaval or brothers in arms or greater servant 2 or 3 times and it fails to kill a dangerous monster and I have to run away, I also have to accept that I've "spent my piety for nothing" (unless I used it against a monster that doesn't regenerate, I suppose). If I use might or healing or any other resource and fail to kill the monster and have to run away, I have to accept that I've spent that resource "for nothing". If you want to present this as a problem, you'd better have a good case for it, because it would mean the entire consumable and piety systems are broken, and fixing it would involve much more than removing the piety cost from failed banishment.

But more importantly: fundamentally, how hexes work in DCSS is that either the monster dies, or nothing happens but you still incur the cost. Still incurring the cost is a vital part of making the fail rate important. If the cost is only incurred when the hex succeeds, or the cost is insignificant, then to beat monsters with high MR, you don't need to increase your hex power to get a good success rate, you just have to get a non-abysmal success rate like 10% and then spam the hex until it works. You can see this problem already with hex spells: to kill death yaks, enchanters just kite them around and use EH or Confuse until it works, even though the success rate is bad, because the MP cost isn't much of a cost. There are a few monsters that are too dangerous to try this on, such as spriggans, but there are far more monsters where it works perfectly well.
The same does not happen with wand hexes or banishment, because the cost when those fail isn't something you just shrug off like MP or hunger. This makes the fail rate much more meaningful and prevents the aforementioned degenerate gameplay.

You can say the hex system is a terrible design and needs to be overhauled, and perhaps you could even get me to agree with you. But you aren't saying that. You just want one of the hexes where the failure cost matters to be changed into one of the hexes where the failure cost doesn't matter. That will result in more of this tedious, degenerate gameplay, so I am confident that it is a bad idea. A further downside is that this change would make banishment inconsistent with how every other hex in the game works.
While you could keep the failure rate relevant by dividing the ultimate piety cost of banishment by the success rate, that would make it even more inconsistent with other hexes and even more difficult to communicate to the player. It also calls into question what should happen if you use banishment with a 1% success rate and it succeeds; should it instantly excommunicate you? Should banishment without enough piety to cover the cost be forbidden entirely? How do you communicate that to the player? What about when a monster has unknown MR gear so the reported success rate is inaccurate? It's much simpler to let the failure rate matter by leaving the piety cost on failure.

It seems like your argument boils down to "spending piety on failed banishment feels bad as a player", but pretty much everything that hurts your character feels bad as a player. Dying feels bad, miscasting a spell at an important time feels bad, failing to kill a low-HP monster with your attack feels bad, yet these are all vitally important game mechanics.

It is a question to you. It is on the same page. I don't see why "or just not use banish vs hard monsters" is bad.