Skill training system

[galehar]

One of the most common complain I've read here is about victory dancing. It has been said that victory dancing is one of most tedious task in the game and that there should be a system that allows players to control how their skills are trained.
We are aware that victory dancing is something that needs to be addressed, but when you suggest changes to the skill training system, keep it mind that the system needs to meet the following goals:

• Natural training: Fiddling in the skill screen shouldn't be mandatory. If left alone, skills used should be trained automatically. This is a feature of the current system that we want to keep, even if it isn't perfect (some skills are much easier to train than others).
• Simple interface: We want to keep the interface as simple as possible. There won't be knobs on the skill menu to control exactly how many percents of XP is going to each skill. We don't want to turn the game into Dungeon Crawl Simulator (tm).

I've designed a system that remove the XP pool and the victory dancing, give the player more control over his skills while also meeting the above goals. But this is a long term goal. It will need a lot of coding, and even more work for tuning and balancing the system. But maybe we can start by implementing the "studied skill" feature in the current system. Here is how it would work :

• In the skill screen, the studied skill is marked with an *. You can study only one skill at any time.
• Whenever a skill is exercised, the study skill is also exercised by the same amount, even if the first one isn't trained due to being disabled.
• The effect of studying fighting and spellcasting diminish as the skills raise.

Terminology: Exercise means using a skill. Training means gaining skill points. In the current system, skills are trained when they are exercised while there is XP in the pool.

In practice, it means that setting a skills as "studied" ensure that about half your XP will go to it. If you also disable the skills you use, almost all your XP will go to the studied skill.

[dpeg]

Thanks galehar!

There is a related issue: spoiler knowledge is necessary to figure out what skills to dance. (See the recent thread about which skills are worth to raise to the maximum.) Ashenzari highlighted the problem, but is not at all the culprit. Another longterm goal should be to make sure that all skills stay useful in the long run.

Personally, I find the studied skill solution crude, but it will probably be effective. So I support it.

[jpeg]

I don't understand. You can't mean what I think you mean, namely that you can study Spellcasting without needing to actually cast any spells, or Fighting without ever coming near any monster...

[dpeg]

I don't understand. You can't mean what I think you mean, namely that you can study Spellcasting without needing to actually cast any spells, or Fighting without ever coming near any monster...

This is why it's crude. It is also what victory dancing is about...

[KoboldLord]

I don't think it is even conceptually possible to combine 'practice' with 'non-grinding'. Real-life practicing is a form of grinding that takes place in the real world, and you cannot emulate the repeated use of an activity in order to gain proficiency in that activity without it feeling like grinding. Any system that attempts to emulate real-life practicing will continue to be described as 'grindy' by nearly everyone who describes the current system that way, because real-life practicing itself is something they would describe as 'grindy'.

If you want to remove the grind, you basically have to gut the system and switch to something completely user-controlled, for instance the #enhance command from the Nethack variants. It's questionable whether that's a worthwhile trade.

[galehar]

I don't understand. You can't mean what I think you mean, namely that you can study Spellcasting without needing to actually cast any spells, or Fighting without ever coming near any monster...

Well, you still have to gain the skill by bashing some plants or reading a few scrolls! But after that, requiring spellcasters to keep meleeing popcorn monsters when they have XP in the pool isn't that great. Anyone who wants to raise spellcasting is casting spells anyway. And when they need more spellcasting for the MP/slots/hunger, they disable magic schools and victory dance. Players already have a lot of control over their skills. It's just that the interface for that (victory dancing) is annoying.

The alternative way to get rid of victory dancing is to exercise skills only when used in an efficient way, thus removing player control. But I'm pretty sure it would only make things worse.

[jpeg]

Well, you still have to gain the skill by bashing some plants or reading a few scrolls! But after that, requiring spellcasters to keep meleeing popcorn monsters when they have XP in the pool isn't that great. Anyone who wants to raise spellcasting is casting spells anyway. And when they need more spellcasting for the MP/slots/hunger, they disable magic schools and victory dance. Players already have a lot of control over their skills. It's just that the interface for that (victory dancing) is annoying.

So what's wrong with the other suggestion of adding a *4 multiplier toggle instead, so the skills gets trained 4 times as often?

Or how about a compromise? Put skills into groups and train the studied skill if any skill in the given group gets skilled, e.g. group all magic skills together, fighting together with melee weapons and armour, throwing together with ranged weapons and dodging, and stealth together with stabbing and traps & doors. Sorta like crosstraining, only much more general and only applying to the studied skill. Not sure what to do about evocations/invocations, but that way at least it'd make sense.

[Danei]

you can study Spellcasting without needing to actually cast any spells, or Fighting without ever coming near any monster...

I absolutely hate this feature of the proposal because it makes skill training seem that much more gamey, but I'm willing to wait and see it in action (if indeed it goes into action) before I condemn the whole thing.

[dpeg]

So what's wrong with the other suggestion of adding a *4 multiplier toggle instead, so the skills gets trained 4 times as often?

4 is probably not relevant enough at high skills. Recall that the difference (I should rather say quotient) between selecting a skill or not used to be 4. We have changed this recently to be a factor increasing with skill.

Or how about a compromise? Put skills into groups and train the studied skill if any skill in the given group gets skilled, e.g. group all magic skills together, fighting together with melee weapons and armour, throwing together with ranged weapons and dodging, and stealth together with stabbing and traps & doors. Sorta like crosstraining, only much more general and only applying to the studied skill. Not sure what to do about evocations/invocations, but that way at least it'd make sense.

That's a novel idea. I like it, but I need to think about it more.

[jpeg]

I've thought about this some more, and I've got a more radical suggestion: The only reason spellcasters need to train Fighting is because of the hp bonus, right? So get rid of that!

Wait, hear me out! I don't know the actual boosts this provides, but it wouldn't be unreasonable to simply up the hp gains on level gains instead.

[dpeg]

I've thought about this some more, and I've got a more radical suggestion: The only reason spellcasters need to train Fighting is because of the hp bonus, right? So get rid of that!

The basic idea of the Fighting skill is that in the late game, the melee fighter will have significantly more HP, but no access to fancy spells. Your proposal would level fighters and magicians... without giving anything to the fighter. I don't think that's good.

If we allow anyone to use study in order to boost Fighting, then that is indeed problematic. The usual, sane reply would be this: XP is a valuable, finite resource; a magician will not want to train Fighting too much, because that takes away too much xp from crucial skills. On the other hand, fighters usually have a much more restricted skill set, so they can afford to train Fighting.

This is probably wishful thinking right now: there is too much xp in the game, and we're not even talking about infinite branches. (It's not a problem that extended endgame characters get up many skills, however.) In any case, your point is a good reason against adding "skill study" as proposed. We need to think more. Perhaps we have to make sure that all skills are viable until the end in the first step.

[ryak]

I, for one, support the tri-nary skill toggle solution.

Another interesting thing to consider is where to add the third option. It can be below, above, or between the two current options.

I like the idea of having the current two plus an option that completely turns off skill gains in that skill. One example of a situation this would improve is starting off with a merfolk and wanting to completely disable Armour skillups in that leather armour you start with.

[KoboldLord]

Part of that problem, though, is that Fighting and Spellcasting are God Skills, and obviously so. With the hp and mp boosts on top of all their other benefits, they are extremely valuable even for the most inappropriate possible characters. Even a late-game Trog worshipper could use a little Spellcasting in the lategame to power evocation items that have an mp cost and the Guardian Spirit effect, so a few levels of Spellcasting from a non-penance source are going to be more helpful than a similar amount of xp into the already-high Evocation or Axe skills.

Perhaps Fighting and Spellcasting could be removed as skills and the functionality could be replaced with a sort of inferred skill? Instead of training Spellcasting as a discrete skill, for instance, your Spellcasting score is the average (or the average + a constant or a function of your level) of all your spellcasting school skills. Spellcasting maxes out when you've maxed out ALL your magic skills, and you cannot simply grind Spellcasting with spamming Magic Dart alone. Fighting, similarly, might be the average of all weapon skills, plus Armor, Dodge, and Shield. Or something like that.

[danr]

Some very important points here, and good ideas.

I think the single biggest problem skills for victory dancing are armour and fighting. If you fix those, "victory dancing" would still exist but be much less prevalent.

As noted, with fighting, the main issue is really that spellcasters need some levels of it to gain a bit more HP. I actually really like the suggestion of decoupling HP from fighting skill. I can't put my finger on it, but I think there is something wrong from a design philosophy perspective when optimal play for spellcasters includes training fighting skill, not in order to fight, but for a side-effect of more max HP. Certainly it is extremely "grindy" to train fighting as a spellcaster.

Instead of tying HP just to XL, another way to do it could be to tie it somehow to stats - some combination of Dex and Str. Str represents your physical robustness, and Dex in a way also reflects physical health. Fighters would tend to gain more HP because they'll put more points into Str + Dex (with more benefit from Str), but spellcasters will also be able to increase their max HP by raising those stats as well.

This would tie HP to things that are already good gameplay and eliminate the need to grind away at a skill you don't need otherwise.

OR, you could have spellcasting also provide some HP bonus. Maybe because as you become a wizened wizard your body becomes embued with some magical resilience or something.

I don't see this as giving casters something without giving fighters something also, it is just eliminating an irritating element of gameplay for casters. The XP needed for a few levels of fighting is trivial, the effect on gameplay though is massive.

Armour: there are ideas elsewhere for refining how armour is trained, and it's not nearly as bad as fighting for spellcasters.

I like the study system too though. The "related skill groups" idea is okay, but here's another consideration: characters might "study" while they rest. This is totally consistent with real life. One can improve one's skill in, for example, martial arts, without actually beating people up. A "study" skill would absorb XP simply while resting and there was XP in the pool, at a level-appropriate rate.

This in a sense does not eliminate victory dancing, it just removes the tedium of it (which I think really is the main negative aspect of it).

The idea that players should have to actively train a skill IS the source of the grindiness of victory dancing. And let's be honest, which of these is more realistic:

- that you can willingly refuse to become better at a skill even though you are practicing it constantly, or
- that you can get better at something by practicing it, even without a monster present.

In fact, the game totally allows the second option right now, it's just that it requires the player to do it manually. That manual aspect is the tedious aspect, and that is the problem with it - it forces players to bore themselves. Let them set a skill to "study" or "practice" while they rest and the problem is eliminated.

[danr]

Further to my above idea for allowing a skill to be practiced or studied while "resting". This could come at a cost. MP and HP recovery could be slowed while "practicing" to reflect the fact that the player is actively thinking or exercising.

However, MP and HP can currently regenerate while in the thick of battle, and I don't think this actually slows recovery.

At a minimum though it could mean that you are not "searching". So if you press 5 while you have study skill active, you are not paying attention to your surroundings and you will not notice any traps or hidden doors.

You also could not practice while moving.

There could also be an increased hunger cost, rather than a delay in HP / MP regeneration. This would reflect the hunger cost of spellcasting or physical training.

[MrMisterMonkey]

(copied from another of my forum posts)

as for victory dancing in general, I agree that it's ridiculous and should die; I'd be fine with a hybrid lemuel/galehar system (get rid of skill pool from lemuel's, get rid of dancing to tweak training rates in galehar's -- downside: victory dancing still a tiny bit possible), preferably with some additional manual override, or some way to configure how to (automatically) allot skill points (since paying directly for skills all the time would be an annoyance -- downside: interface)

in case it's unclear, the allotment configuration needn't be with percentages; the ternary solution should be sufficient

[szanth]

Part of that problem, though, is that Fighting and Spellcasting are God Skills, and obviously so. With the hp and mp boosts on top of all their other benefits, they are extremely valuable even for the most inappropriate possible characters. Even a late-game Trog worshipper could use a little Spellcasting in the lategame to power evocation items that have an mp cost and the Guardian Spirit effect, so a few levels of Spellcasting from a non-penance source are going to be more helpful than a similar amount of xp into the already-high Evocation or Axe skills.

Perhaps Fighting and Spellcasting could be removed as skills and the functionality could be replaced with a sort of inferred skill? Instead of training Spellcasting as a discrete skill, for instance, your Spellcasting score is the average (or the average + a constant or a function of your level) of all your spellcasting school skills. Spellcasting maxes out when you've maxed out ALL your magic skills, and you cannot simply grind Spellcasting with spamming Magic Dart alone. Fighting, similarly, might be the average of all weapon skills, plus Armor, Dodge, and Shield. Or something like that.

The average? So if I have 10 in short blades, I'll have a higher fighting skill than someone with 10 in short blades and 1 in maces?

[ryak]

You probably have to do something more complicated than the average. You could always do something like use the highest weapon skill plus a small bonus for each non-highest skill over some threshhold value to account for skill in multiple weapons.

[szanth]

I don't think the root of the issue is being addressed; if they want to, the player will find a way to train up an ubercharacter. But why is this a bad thing? It's the player's choice to grind or victory dance. I think the job of the devs is to make it less necessary for someone to do so before they can win. My solution would be relatively simple; all exp goes into the pool, and you can manually spend exp by the tens, hundreds or thousands in each individual skill. Eliminates the need to victory dance, and allows spellcasters to train fighting if they so choose; likewise fighters who need an MP boost can do the same for spellcasting.

The more options and choices the player is allowed to make, the less they can say it's the devs that hinder their ability to have fun.

[acvar]

Now this is a good discussion. Here are my 2 cents worth.

galehar: Your intermediate system would most likely work, but I have to ask why make it so complicated? In the end all it does is give players the ability to train any one skill without practicing it. Why tie it to another skill. Just throw 50% or all incoming experience into the "studied" skill and be done with it. It requires less code, has an easier description in the guide, and a cleaner interface.

KoboldLord: You seem to have a very firm grasp of the situation. Games should not be like real life. Real life is grindy and boring. We play games to get away from that. It makes no since from a reality stand point to gain expertise in an activity that you do not participate in, but it makes a great deal of since in a game. I do not believe that you "need" to give a completely user controlled system to remove grind. You only need to give players the option of taking more control over experience allocation.

jpeg: I agree totally. Get rid of fighting and spellcasting. They are problem children. I think the best way to replace them is to give a HP bonus based on the highest "martial" skill, and determine spell levels and MP off of highest "magic" skill.

In my experience the problem is two fold. Some skills like Fighting and spellcasting are clearly better while other skills like short blades, poison magic, and enchantments have clear levels where training them further is useless. This causes some skills to be more highly desired then others and dancing to raise those skills while denying the experience to the less desirable. Making skills more equal would help solve this particular problem, but would not solve everything because another issue is the fact that certain skills although very important simply do not have as many opportunities to train. Anybody that has played a crusader or a melee focused transmuter knows what I mean. Casting buffs are critical for their success, but for every one buff you cast you literally take dozens of swipes with your weapon. This is where galehar's full proposal runs into all kinds of trouble (which he admits). I just think the solution needs to be simple and straight forward. The reason to remove victory dancing is that it is tedious. Creating a more complex system to try and fix it loses sight of the real goal.

[szanth]

I was gonna let it slide the first time you made that mistake, but by the second one I had to respond: "Sense".

[galehar]

I think the x4 multiplier and grouping skills are solutions that only makes victory dance easier. I'd like something more radical. My proposal allows the player to say: "I need this skill. I activate study. If I also want my other skills to train, I keep them enabled. If not, I disable them. Now, I can play however I want, I don't even have to look at the XP pool, I know my "study" skill is being trained.".

I also like the "train when resting" proposed by Danr. It could allow use to put some restrictions on it. If you want to train a magic school, you need a book with a spell of that school. We could even add a new practise command. When you practise magic, your MP won't regenerate past 60% for example. And it will cost hunger. That way, we stay closer to the current "victory dance" system and just fix the interface.

Some of you said that practising fighting and spellcasting might be too strong past the early game. That's why I suggested that "The effect of studying fighting and spellcasting diminish as the skills raise". So you can easily gain the first levels of fighting, but it progressively become less and less effective. And you might want to use the "study" skill on spellcasting, or on a magic school skill.

I also find the idea of tying the HP bonus to STR and DEX interesting. But how to deal with bonus from mutations, chei, randarts,... ?

[evilmike]

I've thought about this some more, and I've got a more radical suggestion: The only reason spellcasters need to train Fighting is because of the hp bonus, right? So get rid of that!

The basic idea of the Fighting skill is that in the late game, the melee fighter will have significantly more HP, but no access to fancy spells. Your proposal would level fighters and magicians... without giving anything to the fighter. I don't think that's good.

I would have to agree with this. It makes a lot of sense that a pure caster would have less HP than a fighter.

On the other hand, it's still silly that a caster has to train fighting at all. And it does tend to be pretty important: even 5 points in fighting (which translates to 27 hp at level 27) can be the difference between being killed in one shot from full HP, to being badly hurt but still alive, if you're playing something like a deep elf or spriggan.

Galehar and danr already posted this, but I was about to suggest the same thing: instead of fighting skill giving an HP boost, why not make it a function of STR instead? INT is obviously a very useful stat, and DEX at least gives you a decent boost to dodging, but STR is quite lacking. How to balance that would definitely be difficult though.

I also think that just tying it to STR makes more sense than making it a combination of STR and DEX.

[danr]

My thought was to just tie HP to "natural" stats - what you've raised it to with level increases and mutations. Might, berserk, gloves of strength, Chei STR bonuses would not affect HP because those would be too wild.

[KoboldLord]

The average? So if I have 10 in short blades, I'll have a higher fighting skill than someone with 10 in short blades and 1 in maces?

No. The first character would have 10 in Short Blades and 0 in Maces. And 0 in Polearms, Long Blades, Staves, etc.

If you want the effect of 27 Fighting in my back-of-the-menu sketch, you'd have to raise ALL combat skills to 27. You'd get a single hit point every few weapon skill levels on average, I suppose. It would be a pretty severe nerf for many characters compared to the current system, but the benefit could be increased to compensate if need be. It would reward combatants who train multiple weapons, at last, as opposed to the current system that locks you into a single weapon type nearly from chargen.

I also like the "train when resting" proposed by Danr. It could allow use to put some restrictions on it. If you want to train a magic school, you need a book with a spell of that school. We could even add a new practise command. When you practise magic, your MP won't regenerate past 60% for example. And it will cost hunger. That way, we stay closer to the current "victory dance" system and just fix the interface.

If using the practice command on magic cripples my mp in that way and rapidly drains my hunger, I'd stick with victory dancing even if it was several times more intrusive and obnoxious than it currently is. Doesn't matter if I can train in a slightly more pleasant manner if I get killed by the next pack of blink frogs because I'm down almost half my power output.

[szanth]

If using the practice command on magic cripples my mp in that way and rapidly drains my hunger, I'd stick with victory dancing even if it was several times more intrusive and obnoxious than it currently is. Doesn't matter if I can train in a slightly more pleasant manner if I get killed by the next pack of blink frogs because I'm down almost half my power output.

Man's got a point.

[danr]

Victory dancing could be elimininated in favour of "practicing" by making training ineffective when there's no tension, at least for skills that are combat oriented. T&D, to me, would make sense that you have to practice on actual traps. In general you should have an item or situation where your character could meaningfully practice - a long sword, an evocable ring, an invocable God, etc.

So skills would still train as now, when you are fighting and blasting and invocating your way through the dungeon. Disarming traps would be an example of this kind of training.

Then there would be "practice" (or "study" or "hone" or whatever) when you have XP in your pool, and you have a spellbook you've been wanting to get a crack at, so you spend some time probing it's mysteries and once you've spent enough XP, you gain a skill level.

Then you just set a rate of how quickly practicing works (some formula based on skill level and XL), the hunger / MP cost if any, and then you just need a command that asks "What skill do you want to practice" and "for how many turns?" and Bob's your uncle.

So it removes the tedium of victory dancing, and imposes some cost. I think the cost should not be too high though. We shouldn't punish players for avoiding the tedium of victory dancing.

[evilmike]

What about doing something with manuals? These are an already existing alternative to victory dancing.

Most skills can be victory danced. All manuals really do is save time by letting you dump exp into a skill. The only exceptions are skills that are hard to train, eg fighting. These are well within the minority, however. Most skills can be easily victory danced or, in the case of weapon skills, train quickly when toggled on.

Unfortunately, manuals are incredibly rare. But should they be? What if they were made more common? (For balance reasons, extremely valuable ones like manuals of fighting should keep their current rarity.)

Of course, this idea is not even close to a fix to the issues surrounding victory dancing. It's just a way that an existing feature might be able to address some of the issues, for now. Arguably a bit of a kludge, but it's certainly easier than rewriting the whole system.

[MrMisterMonkey]

danr, tension is pretty gameable, so, as I understand it, you could still victory dance if you, say, slowed yourself in front of an ogre or two.

evilmike, for what it's worth, the biggest feature of manuals is they let you dump exp into a skill before you hit the cap in ziggurat-like situations

[evilmike]

evilmike, for what it's worth, the biggest feature of manuals is they let you dump exp into a skill before you hit the cap in ziggurat-like situations

Yeah I know, that's practically all I've ever used them for. "Oh look, I have 20k exp, I guess I might as well use this manual of slings..."

But, the reason for doing this is partially because manuals are so rare. At the rate they drop, it's very unlikely that you'll have a manual for a skill you actually would use. Thus, they tend to sit around collecting dust except for rare cases where your exp will go to waste anyway. Also, their rarity somewhat encourages only using them when you are at or near the exp cap. And, it doesn't help that you're most likely to find manuals in places like tomb:3 or ziggurats.

[danr]

danr, tension is pretty gameable, so, as I understand it, you could still victory dance if you, say, slowed yourself in front of an ogre or two.

Players will do it if it's worthwhile. That's why I would not put too great a food / MP cost on "practicing".

I think what we need to let go of is the idea that skills should only be trainable in actual combat or through actual use. That is not the problem, that is just one approach of trying to eliminate the problem of tedious grindy victory dancing.

Yes, I agree, players should spend their time playing the actual game, and not trying to game it. The solution does not have to be to eliminate the opportunity to victory dance (which is probably impossible). All that's called for is to eliminate the tedium of it, or to remove the need for it entirely.

Or is there another problem with victory dancing apart from it being boring gameplay that I am unaware of?

The main critique of an often suggested approach (to let players just "spend" their XP however they want) is again that it would result in them spending more time in the skill menu which, it is argued, would bore them. So let's just remember that tedious activity is really the only thing we need to be concerned about.

I would rather just eliminate manuals and let people train the skills they want. I personally would not find it boring to go to the skill menu and see which skills I can level up. Yes, it's not hard core gaming action, but remember, this is a roguelike, not World of Warcraft. A big part of the game is about managing your inventory, your spells, your skills.

Also, I really don't like the concept of Ashenzari. If skills were easier to train as is being discussed here, there'd be no need for a God who lets you reshuffle all your skills. "Practicing" as discussed here is WAY less meta-gamey than Ashenzari, and for me I know it would remove the most frustrating aspect of the game (like trying to get into a skill that your not good enough at yet to use in real situations - it really limits the player's freedom to play in ways that there's no good reason to bar them from. So you started as a Wizard, found a wicked broadaxe on D:3 and would like to go hybrid? Why force you to husband your XP pool jealously until D:9 when you finally get enough training with popcorn monsters to melee more serious stuff?

Again, most of the time, you'd just train as now - leave your important skills on and just crawl away. But if you want to break into a new skillset, I see no reason why that should take concerted effort and victory dancing to get those skills to a useable level. Micromanaging the XP pool by victory dancing is meta-gamey and really needs to go, and it is a direct result of how hard the game makes it to train skills that you are not very good at yet.

[MrMisterMonkey]

Reskilling would still have usage with fancy training, as it allows for turning old useless skills into useful skills, playstyle conversion, etc. I wrote a bit on it here.

[galehar]

Or is there another problem with victory dancing apart from it being boring gameplay that I am unaware of?

I think there is. Sometimes, the XP pool can influence tactical decision. In the middle of a fight, after having killed a monster, you might be tempted to cast a spell if you can afford the MP and the turn, just to "cash" the XP.
The "practise when resting" solution doesn't address that. It's a sort of automatic and convenient victory dance, but you still have to keep an eye on the XP pool. What I like about my initial proposal is that you can completely ignore it.

My thought was to just tie HP to "natural" stats - what you've raised it to with level increases and mutations. Might, berserk, gloves of strength, Chei STR bonuses would not affect HP because those would be too wild.

What about stat drain? Anyway, I think we shouldn't try to fix everything at the same time. Let's talk about the fighting HP formula in another thread. Here, we'll just assume that most casters need about 5 levels of fighting. So I suggest that practising fighting is capped to level 5, after that, you need to get blood on your blade! Another idea: fighting can't be raised higher than your highest level skill. We can also use both limitations.

I'm not sure about spellcasting. Maybe we can just forbid studying it. Or when studied, it just train slightly faster, and only when using magic skills.

Appart from those 2, I don't think any other skill need to be special cased.

[dpeg]

I don't have the time to reply in depth to the individual postings. Please let me just mention some general principles:

Xp and skills in general:

• The basic assumption is that xp is a valuable resource. You can only spend each point of xp once, as our politicians tell us. Something is wrong if everyone wants to have Fighting 27. That's a lot of xp that (with perfect design) should've been much better spent elsewhere for many characters.
• Conclusion 1: we need to make sure that skill investment remains useful till the end, for all skills. (This is hard work, on a case by case basis.)
• Conclusion 2: need to ensure that xp is limited enough so that there's an actual choice of which skills to get up.
• What do we want from skills? First, characters should play differently according to skill set. This is the case. Next, we want neither "generalise" nor "specialise" to be no-brainers. I think that's the case too: early on, you want to specialise (everyone who has tried to get up too many magical skills at once, finding himself unable to kill stuff in the mid-game will know what I mean), but there is a clear danger of over-specialisation (everyone who has found himself unable to deal with too many opponents in the late game knows this). These could probably be improved upon, but should definitely be kept.

Training:

• Apart from this (and especially with these solved), we can afford skills to train fast. There is no need to artificially slow down training if xp is important enough. A problematic skill is Armour, which trains noticeably too slow later on.
• Training should be automatic. I think a good example is Stealth: you train it when walking around sleeping monsters. I found my stabbers to gain Stealth fast enough just by minding their business.
  A bad example is Armour again: neither "only train when hit" nor "only train when an attack missed" are good -- both encourage the rat dance. Instead, Armour should train regardless of hit or miss if the attack was potentially strong enough. If we ramp up actual Armour training, then this excludes the rat dance, but would make Armour training by way of normal fighting completely natural. (This is again case by case analysis.)

HP and Fighting

• HP is good for everyone; Fighting provides HP until the very end, so everyone likes to have some Fighting. So far, that's okay.
• The problem is that Fighting is very difficult to train for non-fighters. Since they also want HP, they'll find themselves grinding the Fighting skill up, (Been there, done that.)
• One proposal is to notice that fighting goes up whenever you actually fight, i.e. train one of the melee skills (including Unarmed Combat). Since Fighting does little (even nothing) beyond giving HP, we could scrap the skill and instead compute the HP boost by total amount of xp spent into all melee skills. (I don't think that Fighting 27 should correspond to all melee skills 27, by the way.) This seems clean, but has two obvious drawbacks: casters will now want to grind up some melee skill (the way I phrased it, the aptitude does not matter). And we lose the potential of the additional Fighting aptitude: it would be conceivable to have a species which is very good at getting HP up (great Fighting skill) but only mediocre at actual fighting, or, perhaps more interesting, the other way around.
• The other proposal is to remove HP gain from skills and use something else instead. The Str stat is not so bad. It would help a lot with making the stat choices interesting. We don't want maximal HP to fluctuate all the time, so total Health would only be affected by base Str (possibly including mutations) but not by items, potions (decay, might).
  It might be an option to have each point of Str give a certain boost on HP, and to use half that boost for a point of Dex.
  The drawback of this approach is that we lose granularity: effectively, we are replacing the smooth curve of 27 Fighting skill ticks by the coarser 9 choices of a stat. However, that might be okay.

Ashenzari

• It's okay to not like the god and/or the reskilling mechanic. But this has nothing to do with the skill system or victory dancing. If you see it that way, Ashenzari is nothing but a lazy but expensive (xp cost) shortcut to dancing. (The xp boost will go away, it is a mistake.)
• However, the real point of reskilling is tactical flexibility: whereas victory dancing is concerned with funneling xp into a certain skill, reskilling is about scrapping one skill in order to build another one. This is a kind of strategic flexibility nobody else has. (And that's why I think the ability is fitting the god and interesting.)
• In order to keep this thread sane and somewhat focused, please refrain from discussing Ashenzari. That god has nothing to do with the skill training system.

[szanth]

My philosophy on exp is kinda like a J.G.Wentworth commercial. It's -your- experience, use it when -you- need it.

[galehar]

[*] Conclusion 1: we need to make sure that skill investment remains useful till the end, for all skills. (This is hard work, on a case by case basis.)

I agree. Worst offenders are probably weapon skills, but fiddling with the damage formula scares me

[*] Training should be automatic. I think a good example is Stealth: you train it when walking around sleeping monsters. I found my stabbers to gain Stealth fast enough just by minding their business.

Maybe this is because stealth cost only half as much XP as other skills. This is an old hack that should be fixed, but I don't know how.

A bad example is Armour again: neither "only train when hit" nor "only train when an attack missed" are good -- both encourage the rat dance. Instead, Armour should train regardless of hit or miss if the attack was potentially strong enough. If we ramp up actual Armour training, then this excludes the rat dance, but would make Armour training by way of normal fighting completely natural. (This is again case by case analysis.)

Sure, armour skill training could be better. Are you saying that we should fix training instead of adding a "study skill" feature? What about T&D and magic school skills?

HP and Fighting

As said above, I agree that the HP bonus from fighting needs to be addressed, but this is a different topic. By trying to fix everything at once, we end up doing nothing.

[galehar]

Also, I think we should put some conditions for setting a skill as "study":

Fighting: skill level < min(5, highest weapon skill). You also need to carry a weapon for which your skill is higher than fighting. Example: you've got fighting:1, short blade:3, long blade:5. You can study skill up to 3 while carrying a dagger, but need to carry a long sword to study it up to 5.
weapon skill: carry a weapon of the type studied (and appropriate missiles for ranged).
armour: wear some heavy armour (EVP*3 > armour skill)
dogdging/stealth: wear some light armour
shield: wear a shield
stabbing: carry a dagger or a quick blade (other stab appropriate weapon?)
T&D: nothing
magic school: have a spell of the trained schooled memorised with a casting success of at least good (or very good?). So you can't just learn a high level spell and study the school, saving the amnesia scrolls normal training would have cost you.

[acvar]

This thread is starting to devolve into:

Hey we could do this.
Nope that would not work because of fighting skill

Well how about this?
Nope that would not work because of spellcasting and fighting.

Then why not this?
Nope fighting and spellcasting.

etc. etc.

At some point you need to put the blame where it truly lies and address the real problem. Getting rid of fighting and spellcasting will not solve the problem of victory dancing, but you will not be able to solve the problem of victory dancing without first getting rid of fighting and spellcasting.

[dpeg]

Also, I think we should put some conditions for setting a skill as "study":

Fighting: skill level < min(5, highest weapon skill). You also need to carry a weapon for which your skill is higher than fighting. Example: you've got fighting:1, short blade:3, long blade:5. You can study skill up to 3 while carrying a dagger, but need to carry a long sword to study it up to 5.
weapon skill: carry a weapon of the type studied (and appropriate missiles for ranged).
armour: wear some heavy armour (EVP*3 > armour skill)
dogdging/stealth: wear some light armour
shield: wear a shield
stabbing: carry a dagger or a quick blade (other stab appropriate weapon?)
T&D: nothing
magic school: have a spell of the trained schooled memorised with a casting success of at least good (or very good?). So you can't just learn a high level spell and study the school, saving the amnesia scrolls normal training would have cost you.

I believe that "study" will only things more complicated in the future, without addressing the core problems. This is why I took the time to write the treatise (instead of replying to individual mails). We will want both Conclusions, so we should set out to get there. (After that, we may find we still need to add "study". But proper order is fix the inevitable first.)
Your proposal is ingenious, but demanding flavourful conditions (a la "carry Foo") is effectively distraction.

Are you saying that we should fix training instead of adding a "study skill" feature? What about T&D and magic school skills?

Absolutely, yes. That's what I am saying (and I tried to movitate why). The plans to repair Armour and Dodging are not new, we could do this any time. I am happy we did Stealth at some point. As I said, there's no way around a case by case analysis. T&D is currently hopeless, there is no reason to train it very high. The radical solution would have been to scrap traps in general, we didn't do that. So the new plan is to add trap planting (from wands, which is an ingenious idea by b0rsuk). Assuming that an expensive wand, planted with T&D 27 will have marvellous destructive potential, that would save the skill.
Regarding melee and magic skills, someone needs to do the dirty work and crunch numbers.

[danr]

Very interesting and fruitful discussion. My thoughts keep evolving:

- I agree, ideally training should be automatic. The ability to study or practice is only proposed as a way to avoid how unnecessarily difficult it is to train certain skills currently.

- Because of that, I'm leaning back to the notion of setting a "study" skill in the skill menu that would be accelerated somehow, as opposed to actively "practicing" as we've been discussing, because that's a bit more automatic. Even the idea of just setting another multiplier would go a long way.

- However, I also agree that if certain problematic skills can be fixed, the need for this extra technique is reduced.

Re: problematic skills:
- Armour: wouldn't it be simple to just let it train faster? It could be like stealth, where simply walking around under a certain condition (wearing armour) trains the skill. Just going about your ordinary business will make you more comfortable in it.
- Alternatively, we could get rid of armour skill. Increasing AC would simply be a matter of finding better armour, and enchanting it more. Getting better at magic with it would be a matter of increasing your Dex.

- Fighting: is HP really actually the main benefit of it? If so, I agree it's potentially redundant. HP could be boosted in other ways (based on stats, or some other skills (even armour?)). If it stays, there should be a hack (like study) for spellcasters to boost their HP via fighting. The need for casters to get a few levels of fighting is by far the worst manifestation of the issues around training.

- Spellcasting: again, another meta skill. It's effects could be provided in some other way (total of other magical skills, etc.). However, I currently have no frustrations training spellcasting with characters that need it, because any caster naturally uses it often. Just for simplicity of design it could be melded into the other magical skills.

- T&D - doesn't really bother me. Sure there may be issues with it, but they are not harmful to gameplay. A trapping ability would be cool though. Finally give my spriggan a use for those throwing nets.

- Invocations: I mention this because it is way too hard to train it up with certain gods to the point where their higher level abilities become useful. Here a shortcut might be nice. Perhaps it could be practiced by praying?

- Stealth: I think this works well currently. My XL25 HaBe has stealth of 15. If I'd really been focussing on stealth I should be able to have much higher stealth by now, I think it trains at a good speed.

Finally, I just want to point out one thing. One objection to the study or practice approach is that it could potentially compel players to spend a lot of time in the skill menu. With the current XP pool system, you also have to do this a lot because the only way to accelerate a skill is by deactivating a bunch of other skills. So the interface would be greatly streamlined if all you had to do to focus on one skill was to select that one skill, rather than to deselect a whole bunch of other skills. It's just more direct, obvious, intuitive to set a "focus", "study" or accelerated skill (or even a few of them, as this would reduce the need to switch back and forth between multiple skills that you are wanting to focus on, while not crippling all the others.)

[danr]

After all that, and in light of the comment about the need to not try to fix everything, here's a proposal:

For now, just create an additional speed of skill training - so you would have three choices for each skill:
- Disabled (current formula)
- Enabled (current formula)
- Focused (current Enabled formula * a multiplier)

Or, they could be a bit more flavourful:
- Disabled
- Active
- Studious

If needed, the multiplier could vary - eg. give a high modifier up to skill 5 to reduce grinding for those characters who just need a few levels and then want to ignore this skill for the rest of the game. Or maybe fighting and spellcasting would have different multipliers for "Dedicated".

This is simple to implement. After a while we could reassess and see if there is still a problem with grindy training activities.

[wwf]

I think the idea of making manuals more common/affordable would make victory dancing less of a problem. More manuals in book stores would make also benefit melee characters who otherwise would have almost no interest in a book store.

Manuals have a limited number of uses, and the XP points are the limiting factor in gaining skill, not the access to manuals. Flooding the game with manuals would not unbalance the game. It would reduce the need for victory dancing.

Another way to make manuals more common is to include a relevant manual in the inventory of all starting characters.

[wwf]

I would like to think of manuals as a training manual used in the school that is responsible for your background skills. You will start with one.

The manual would not wear out, but would only be able to train your skill to a specific level (say 10). After that, you need first hand experience to gain additional skill.

This may also open up a few different strategies for survival of early characters without affecting the balance in the mid and end games.

[seth]

I always thought it was weird that you learn nothing while fighting an Ancient Lich, but then immediately afterward skill up your fighting by bashing rats. Shouldn't you only learn while you're fighting the tough enemies?

What if you spent your XP as you fought the enemy? You only get the XP if you successfully make the kill, of course, but what skills get the XP is determined by what skills you used as you were fighting. If you sneak up on a guy, some points are allotted to stealth. If you are Enchanting or Transmuting will be trained based on how often you use (and not just cast) a spell--if you have Blade Hands on, Transmutations trains each time you strike a baddie. Not sure how Traps should work, but I don't think it should be a big concern since it's a very minor element at this point.

Maybe for most cases (i.e. not one hit kills), there should be something like a 50% limit on how much of the XP you can train on one skill. This would go down to 15% or whatever for when you have the skill turned off.

With this system, there would be no point to spend any time bashing rats at lvl 27. Hopefully, it wouldn't encourage training skills on Ancient Liches--it wouldn't be worth the risk to keep zapping Corona to get your Spellcasting up. You would have to train on baddies that are moderately tough so that XP spent is worthwhile, so you won't be able to lean on keys.

[KoboldLord]

The trouble with that sort of system, Seth, is that you just encourage "pre-dancing" instead. If you want to train enchantments or transmutations, you just have to spam a spell of that type at the beginning of combat regardless of whether it's remotely appropriate. Every. Single. Combat. In many ways, this is worse than the current system, which only occasionally demands that you victory dance after a particularly big kill.

Having the game engine distinguish between skills that are used productively and skills that are used frivolously is probably impossible, not least because there's plenty of room for honest difference of opinion. For instance, is casting Repel Missiles only valid if something actually shoots at you, or can you re-cast it regularly just in case a yaktaur band wanders through? I usually try to keep it up at all times if I have it, but to the game engine it must look pointless.

[danr]

I think the idea of making manuals more common/affordable would make victory dancing less of a problem. More manuals in book stores would make also benefit melee characters who otherwise would have almost no interest in a book store.

Manuals have a limited number of uses, and the XP points are the limiting factor in gaining skill, not the access to manuals. Flooding the game with manuals would not unbalance the game. It would reduce the need for victory dancing.

Another way to make manuals more common is to include a relevant manual in the inventory of all starting characters.

More manuals could help. There would have to be a lot more though to guarantee a decent chance of finding a helpful one.

Starting with a relevant manual would only help so much. Typically what I'd really like a manual for is for a non-starting skill (e.g. fighting with spellcasters) because those are the skills that are hard & somewhat dangerous to train - you have to use them a lot to train them but they are pretty weak.

IMCO (in my current opinion) training fighting with spellcasters is just not worth it to me - within the game, it means a lot of time spent trying to kill things in a really inefficient way which carries a higher risk of death which I think probably offsets the benefit of long-term HP gain, and at a meta-game level it means I'm spending a lot of time playing a weak fighter as opposed to the powerful wizard that I was hoping to play. That's just me though, a lot of players have no problem with that. However, that's the main reason I'm currently playing mostly Hunters, Berserkers, Fighters, Chaos Knights and Assassins.

[acvar]

But why go to all the trouble of adding more manuals when the same can be achieved by just putting a train mode in the skill interface. Just hit "!" and then the letter of the skill you want to practice. No need to scour the dungeon hoping to find the manual you desperately need yet you still get the same effect. The fact that manuals exist, and that people want more of them is evidence that there is a very basic problem in the system, and that problem is at least in part a philosophical problem. Some developers appear to believe that they should have more control over character development then the person playing the characters does.

I keep hearing about how the interface needs to be transparent. Why? The current system is worse. We don't need a perfect system just one that works better then victory dancing. If the choice is a slightly more cumbersome and intensive interface in the skills menu, and victory dancing I sure as hell will take the former.