Fail rates for magic devices

[Magnate]

Hi all,

I'm resurrecting the discussion started here: http://angband.oook.cz/forum/showthread.php?t=1222 - AFAICT it never did get its own thread.

The basic issue here is that the current fail rate calculation is oddly complex, not very intuitive, and not well suited to displaying easily. (There is also (IMO) way too big a reduction when you're confused.)

So, Takkaria wants to replace it with something simpler to understand, calculate and display. (Btw, we're talking about wands, staves, rods and activatable items.)

Things to consider, with my current views on them, are:

1. Should there ever be 0% or 100% chance of success? The D&D tradition is that there shouldn't, but in some cases (like Teleport Other) we're talking about potential instadeath if we don't allow 100% success. Personally I'm inclined to go for a strictly linear scale rather than an asymptotic one (meaning both 0% and 100% would be possible), because this is how related skills like Saving Throw and Disarming work. But I'm interested in opposing views. Should this rule apply to all devices, or should some have 100% attainable and others not? I don't like the additional complexity of applying it differently, but it might be worth it if there are good reasons.

2. Should an item's activatability be divorced from its native depth? This is almost certainly a yes, as it allows us to make sure that items essential for warriors (like rods of detection) are easier to activate than luxury mage items (like wands of annihilation). So we introduce a "difficulty" stat for activatable devices, which goes from 1 to 100, and we compare this with the character's modified device skill, and apply some formula to turn the difference into a percentage chance of success.

3. Should there be a "critical" activation, which has (say) double the expected effect. This exists in some variants (S comes to mind), and I have no problem with it, but it's a separate piece of coding from the basic failure calculation. An optional extra, as it were.

4. What should the basic chances of success be? We need a few test cases, e.g.:

1st level human warrior activating a wand of magic missile / staff of detect invis / rod of trap detection

50th level gnome mage activating a wand of annihilation / staff of magi or power

(and some others in between - I'll look up the numbers and come back with a list)

5. What should the effects of blindness / confusion / hallucination / cuts / stunning / fear / poison / blessed / heroic be on actual failure rate (or device skill), if any? One advantage of a linear system is that these effects translate more easily from effect on device skill to effect on failure rate. I think at the moment only confusion and stunning affect failure rate, but I'm not sure about that. In any case, we ought to check all the temporary states and revise from first principles. Someone let me know if I've missed one.

Please let me have your views. In homage to Paul Blay, I'd like to ask you to vote for the things you consider most important, either positively or negatively, using a system so complex I'll let you make it up yourselves.

[Pete Mack]

Only confusion affects device failure. Stunning doesn't.

[Magnate]

Originally Posted by Pete Mack

Only confusion affects device failure. Stunning doesn't.

Thanks Pete. That's quite interesting, as stunning affects spellcasting failure, so it's not really consistent. IMO most of the temporary conditions should have some small effect, rather than confusion alone having such a big one.

[bebo]

1. i would go with a linear system, but with the (usual?) caps at 5% and 95% - it adds spice to the gameplay, without really being the cause of that many instadeaths. Maybe give only priests and wizards 0% fail rate, as a way of improving them (they need all the help we can give atm )

2. i'll also go with a yes here - it allows a better fine tuning of magic objects

3. possibly interesting, but i would consider it a fairly minor improvement; as long as it doesn't cause the above improvements to be coded in v3.3 instead of in 3.1.2 i'm ok with it

4. no idea here; of course different classes/races should have a different slope to the curve (and possibly intercept if we chose to vary min-max rates)

5. poisoned/blessed/heroic should have no effect whatsoever; blindness should remain as is (it really gives tactical depth to the game providing an essential factor in the choice between scroll vs staff, in addition to recharging and fire attack sustainability); confusion/stunning/hallucination/berserk i don't know - maybe give a MEASURABLE, CONSISTENT and CLEARLY DISPLAYABLE (does this word exist) malus?

just my two cents

[Nick]

1. I agree 100% should be achievable (and 0% doesn't really matter).

2. Device independent of depth is a definite yes.

3. Critical activation is a maybe - perhaps class dependent and done a bit like the way bolts turn into beams for mages.

[will_asher]

1) I agree we need 100% for stuff like teleport other, but for most things I think 100% should be very hard to achieve. I guess that would require applying it differently for different items.

2) yes.

3) It'd be cool if bolt wands/rods sometimes had a beam effect like the spells.

4) 1st level human warrior with a wand of magic missile: no better than 50% Warriors are supposed to be bad a using items, and they'll get better quickly as they gain levels.
50th level gnome mage should probably have at least 90% success rate for just about everything.

5) It would make sense that confusion should have the same likeliness of causing failure to activate as it does to make you move in a random direction when you walk. You're getting off easy with having the skill halved from confusion. I also think it should apply differently to wands/rods and staffs. For wands & rods instead of increasing failure rate, just have the random direction effect when aiming, but the current effect of confusion on staffs is correct IMO. At the same time, I think staffs and rods of curing should be an exception because they should be practical to use to cure confusion among other things.
Stunning and Terror ("desperate to escape") should probably have minor effects, but I don't think any of those others should affect it.

[Marble Dice]

1) D20 Success/Failure: I would consider just using the min spell fail rate as determined by class and INT, ie only mage/priest can get 0%, and otherwise you can reach 5% min failure at 18/20 INT. If that would make necessary devices impractical for warriors, you could just use 0 or 5 min fail, depending on class.

2) Difficulty != Depth: Definitely.

3) Critical Activation: I don't care much either way. I like getting the occasional beam from a rod, but currently that's just part of the specific spell effects on the rods, isn't it?

4) Base chances of success: In a linear system, perhaps each device could have both a difficulty (slope) and a base failure rate (y-intercept). The modified failure rate could decrease in a linear progression, relative to the device's difficulty, as the player's device skill increased. This way you could have devices that you can use early, but are unreliable, and they stay that way for a long time. On the other hand you could still have devices that are very difficult to use until you accumulate enough skill, and then they rapidly become relatively reliable. As for actual rates, I could see a low level warrior using the basics having around a 40% fail rate, and a high level mage using some of the powerful luxuries in the 5-15% range. I don't think you'd want to put too high of a failure rate on the damage devices without considering the effect that would have on their (frequently lacking) damage output.

5) Effect of status: I'd say let confusion just aim random, and possibly give stun a penalty to device skill or success rate.

[Atarlost]

1: 0% fail is a must. 100% fail is okay most of the time, but for non-time sensative activations like Cubragol's bolt branding it becomes a problem. Any such devices (rods of restoration, rods of recall, and rods of perception also come to mind) should have their difficulty tweaked if they're completely unusable by half-troll warriors at native depth.

2: absolutely

3: maybe for attack devices. One option is to simply not use a charge on a critical sucess.

4: I'll go with 50% fail for a level 1 human warrior with the easiest wand and 10% fail for the level 50 mage with the hardest. Even a warrior should never go above 50% for a basic utilitiy when it comes into depth at an idealized CL=DL/2 dive rate.

5: Blindness should have no effect. Any effect from confusion or stunning should be small because a confused player can't read a teleport scroll. I'm going to say bless and possibly heroism should give a boost and it should be big enough to cancel out confusion. None of the other status effects should do anything.

[bebo]

After more thought i'd say that hallucination should probably work as blindness; confusion should (as already suggested) give the random aim with wands/rods and lower success rate with staves; stunning, terror and berserk should increase failure rate by something around 20% maybe?.

[Magnate]

Thanks for the replies. It's looking like a small majority in favour of allowing 100% success, so I'll see how this works out once I've done a spreadsheet with the various skill levels of the class/race combinations. There seems to be a sensible estimate of 50% for a level 1 warrior using the lowest difficulty device (maybe slightly higher for gnomes, lower for half-trolls), and somewhere around 90% for a fully-maxed-out gnome mage using the toughest device.

Lots of interesting views on the temporary states - I'm definitely in favour of making the game more interesting by having most of them have a small effect one way or the other, but I'll come up with more definite proposals when I've done the spreadsheet. I think the solution to rods of curing (and other confusion-removers) is just to make them really easy to activate, rather than special-case the confusion effect. Since we're divorcing depth and difficulty they can be really easy to activate and still deep (and rare).

I'm definitely leaving critical activations aside for now - they can always be added later.