Randomness and non-combat skills

[Zeyelth]

Did a quick search, but couldn't find this particular topic.

 

One thing which never really translated well from the D&D based games was the randomness involved in non-combat skills checks. That is, skill checks for conversations, or finding hidden doors, etc.

 

The problem often manifested when your stats were good enough for you to succeed, but low enough to frequently fail. Since a huge difference between a Pen and Paper RPG and a CRPG is the ability to save and reload, this often made for an unfortunate meta-game, where you memorise a dialogue tree, and save/reload until you get it right, and you pass the right skill checks. Is it cheating? Yeah, kind of. But most games I've played completely lacks an incentive for accepting a bad roll of the dice. There are ways around it, but while a bit of randomness in combat makes it less predictable and more fun, I think it's a mistake in non-combat situations.

 

I'm not sure what a good alternative would be, but one idea is to base it on your skills. No randomness, and then shift the requirements over time, depending on things you do. For example, conversations with an NPC at one point in the game may lower or increase the requirements to pass a future check. That way, it's harder to just reload a previous save, and you can use a strategy guide predictably if you really want to.

[rjshae]

They could generate a random value for each party member and a random key for each skill check, then use the pair to determine the result. (XOR?) The same character would always get the same result for the same test, regardless of save-restores. But the results would change on each subsequent play through, giving you a completely different game.

[Zeyelth]

That could work, if carefully balanced. Otherwise you may end up with a situation where you get the worst roll possible in a really important situation, which may require you to massively level up a skill just to pass that check.

 

On the other hand, it does add a frustrating aspect to character building which is very hard to plan for; there's no real feedback, and things would just seem to randomly fail, without any reasonable justification ("hey, I just talked down this powerful demon, why can't I convince this peasant to go away?")

[rjshae]

That could work, if carefully balanced. Otherwise you may end up with a situation where you get the worst roll possible in a really important situation, which may require you to massively level up a skill just to pass that check.

 

On the other hand, it does add a frustrating aspect to character building which is very hard to plan for; there's no real feedback, and things would just seem to randomly fail, without any reasonable justification ("hey, I just talked down this powerful demon, why can't I convince this peasant to go away?")

 

Erm... tough nuggies? Thanks for playing, you can stop save spamming now?

[PK htiw klaw eriF]

I thought that randomness with non-combat abilities was a good thing. After all, it is possible to screw up when picking a lock, trying to bluff, or forging a weapon. I would like something like d20, Gurps, or Palladium style checks, where you have a certain chance to succeed, but can actually fail at something.

[Hassat Hunter]

I would be fine with fixed checks in dialogue. Random rolls in non-dialogue/non-combat rolls.

 

And success on, say, wilderness lore, might also depend on how much of the map you discovered, how much info you gathered about a location. Dialogue depends on reputation, previous things you said or did etc.

 

I seriously doubt you *ever* need to pass a test and if you fail it's game over. OE aren't amateurs. It may have bad consequences, but I would just say... suck it up. Can't win all checks in the game. Need to make choices, sacrifices...

[Zeyelth]

Erm... tough nuggies? Thanks for playing, you can stop save spamming now?

Sure, I could be apply some strict self-control and not reload saves when I get a bad roll, but that's missing the point. The point is that the mechanic isn't well suited for computer games, and could be replaced with something better. It was originally designed back when all you had were a handful of dice, and in some cases a deck of cards in place of random events.

 

I would be fine with fixed checks in dialogue. Random rolls in non-dialogue/non-combat rolls.

 

And success on, say, wilderness lore, might also depend on how much of the map you discovered, how much info you gathered about a location. Dialogue depends on reputation, previous things you said or did etc.

 

I seriously doubt you *ever* need to pass a test and if you fail it's game over. OE aren't amateurs. It may have bad consequences, but I would just say... suck it up. Can't win all checks in the game. Need to make choices, sacrifices...

Yeah, I would prefer some other checks which tie in to something more tangible, like past conversations, your reputation, knowledge (visibility of map), etc...

 

Not saying that OE will somehow write a situation where you fail a check and have to start over (that's just bad design), but the mechanic often introduces randomness in situation where there should be none.

 

As an example, randomness on persuasion checks. What exactly does the randomness represent here? A sudden memory of something funny which breaks your pokerface? It would be more natural if your choice of words made a difference. Having a low int. or charisma could restrict you to a subset of replies, or make some options less viable (charm may not be very effective if you just soiled yourself), but I don't see why the *success* of a given option should be determined by chance.

 

Picking locks? Yeah, randomness could definitely represent the quality of the lockpicks, making some more likely to break than others. I think Skyrim implemented it for durability, while making stats affect difficulty, which tied in to your own skills. Not saying that PE should use that mechanic, but it's something to consider.

 

tl:dr; I want to be punished for making bad choices, not because the random number generator decided to stay on the low end of the pool.

Edited October 15, 2012 by Zeyelth

[rjshae]

Erm... tough nuggies? Thanks for playing, you can stop save spamming now?

Sure, I could be apply some strict self-control and not reload saves when I get a bad roll, but that's missing the point. The point is that the mechanic isn't well suited for computer games, and could be replaced with something better. It was originally designed back when all you had were a handful of dice, and in some cases a deck of cards in place of random events.

 

The alternative is what? Player-skill-based minigames? Always succeeding when you pick some option in a dialogue? If you're at skill rank X you fail, at X+1 you succeed? Life is a crap shoot and you usually don't always know all the variables. Random seems fine to me.

Edited October 15, 2012 by rjshae

[AGX-17]

Erm... tough nuggies? Thanks for playing, you can stop save spamming now?

Sure, I could be apply some strict self-control and not reload saves when I get a bad roll, but that's missing the point. The point is that the mechanic isn't well suited for computer games, and could be replaced with something better. It was originally designed back when all you had were a handful of dice, and in some cases a deck of cards in place of random events.

 

I would be fine with fixed checks in dialogue. Random rolls in non-dialogue/non-combat rolls.

 

And success on, say, wilderness lore, might also depend on how much of the map you discovered, how much info you gathered about a location. Dialogue depends on reputation, previous things you said or did etc.

 

I seriously doubt you *ever* need to pass a test and if you fail it's game over. OE aren't amateurs. It may have bad consequences, but I would just say... suck it up. Can't win all checks in the game. Need to make choices, sacrifices...

Yeah, I would prefer some other checks which tie in to something more tangible, like past conversations, your reputation, knowledge (visibility of map), etc...

 

Not saying that OE will somehow write a situation where you fail a check and have to start over (that's just bad design), but the mechanic often introduces randomness in situation where there should be none.

 

As an example, randomness on persuasion checks. What exactly does the randomness represent here? A sudden memory of something funny which breaks your pokerface? It would be more natural if your choice of words made a difference. Having a low int. or charisma could restrict you to a subset of replies, or make some options less viable (charm may not be very effective if you just soiled yourself), but I don't see why the *success* of a given option should be determined by chance.

 

Picking locks? Yeah, randomness could definitely represent the quality of the lockpicks, making some more likely to break than others. I think Skyrim implemented it for durability, while making stats affect difficulty, which tied in to your own skills. Not saying that PE should use that mechanic, but it's something to consider.

 

tl:dr; I want to be punished for making bad choices, not because the random number generator decided to stay on the low end of the pool.

 

Speech and persuasion shouldn't have much, if any, randomness involved. It's a little absurd that you can be given the "correct" line but fail because of the vicissitudes of the random number generator. I'd like to see persuasion more about watching and understanding the character you're trying to persuade. Maybe chat them up occasionally without bringing up "THAT" to get a feel, and as you talk to them and exhaust other options your persuasion line gets more refined and more successful as a result. Rather than a binary succeed/fail there could be varying degrees of success or failure.

 

 

Erm... tough nuggies? Thanks for playing, you can stop save spamming now?

Sure, I could be apply some strict self-control and not reload saves when I get a bad roll, but that's missing the point. The point is that the mechanic isn't well suited for computer games, and could be replaced with something better. It was originally designed back when all you had were a handful of dice, and in some cases a deck of cards in place of random events.

 

The alternative is what? Player-skill-based minigames? Always succeeding when you pick some option in a dialogue? If you're at skill rank X you fail, at X+1 you succeed? Life is a crap shoot and you usually don't always know all the variables. Random seems fine to me.

 

By that standard, you just have to hit on that hot girl at the bar repeatedly over and over and eventually she'll marry you. And there are no such things as stalkers or restraining orders. The real world is not simple "luck," genetics, context and any other number of variables are involved. Did you get enough sleep last night? Did you take a shower this morning? Are you overweight? Are you athletic? Does her biology cause her to find your facial structure attractive or ugly? Are you assertive, aggressive, shy or fidgety? Are you a good conversationalist? Are you drunk? Is she drunk? Do you have an ugly wingman to make you look better in comparison? The list goes on...

Edited October 15, 2012 by AGX-17

[Zeyelth]

Not going to bother with quotes, but following up on what AGX-17 said: Randomness in conversations is a cheap way of "simulating" all those little tiny details which normally affect other's perception of you, and the result is not very good. It's not that randomness itself is bad, it's that the game doesn't offer a plausible reason for the failure; there's no feedback (and no, seeing "persuasion check failed" printed on screen is not proper feedback).

 

In real life, you get subtle clues, and you can infer the reason from that; you didn't seem to care (lack of sleep), didn't seem very enthusiastic, etc.

 

With classical CRPGs, we have been conditioned to accept that everything is determined by a roll of a die, with some modifiers applied. Again, this is a legacy mechanic from way back when the only alternative was having a storyteller (DM) say "because I said so". Games which rely on pure randomness for non-combat situations are often based on one of the many pen and paper rule systems.

 

Given the choice between completely static conversation trees and random-influenced conversation trees, I'd choose the former, even if I don't think it's very good either.

 

That said, there are quite a few ways to avoid involving randomness. Some which have even been implemented in highly successful games. For example, Mass Effect use skill points to limit your conversation options (charm/intimidate), but the available dialogue options vary depending on other factors too; taking sides open some dialogue options and closes other. Some actions may alter, or even completely remove some conversations. Sure, there's a lot of other problems with Mass Effect, but in most cases there's an actual reason as to why you can't steer a conversation in a given direction.

 

When it comes to computer games, there's no real reason NOT to store all those tiny details which the randomness was intended to replace. Furthermore, you don't have to limit skill checks to a simple pass/fail, you can add other requirements too; maybe your trap-finding skills aren't that great, so you can't see the traps unless you get close enough. Bumping that skill would simply mean that you can see them from afar. In this case, randomness could also be involved, but it shouldn't affect the skill check, it should be applied to make the visibility range variable (simulating flickering lights and other things which could help or distract you).

[jivex5k]

You shouldn't try to stop people from reloading saves, or make decisions about game mechanics with this intent.

People will cheat/mod/save whatever, it's their game let them! If someone gets more joy out of loading and loading it until it's what they want then so be it, why should we stop them?

It doesn't hurt people who don't want to play that way, its a choice.

 

Case in point - X-Com

You can play ironman mode yourself if you want without turning it on if you have some discipline.

Or you can save after each move to ensure a perfect campaign where no one dies.

Or somewhere in between.

 

But the game isn't designed to limit your reloading to get the best outcome, it says: "Yeah man go ahead, have fun". It doesn't hurt anyone who doesn't want this.

 

The OP says: "But most games I've played completely lacks an incentive for accepting a bad roll of the dice. There are ways around it, but while a bit of randomness in combat makes it less predictable and more fun, I think it's a mistake in non-combat situations."

 

The incentive for sticking with a bad roll is it makes your victories more meaningful, there shouldn't be some other in-game reward or incentive. Don't take out randomness in non-combat scenarios to limit the amount of saving/reloading.

Some people like to stick with their bad decisions even if they could reload.

 

Basically: People like to cheat, if you don't like to cheat then don't, but don't stop others from having fun in a single-player game.

Edited October 16, 2012 by jivex5k

[Larkaloke]

Short of complete simulation, which is really not feasible, I believe that randomness is the best way.

 

It's a matter of opinion or belief whether or not luck plays a part in real-world interactions and skills, but it is an observable fact that things do not always go precisely the same way every time you try them. One could juggle four balls in the same pattern at three different times in the day, their agility not having changed, and their skill presumably not having changed in that short an amount of time. One of those times they might do it flawlessly, one of those times they might drop the balls. The randomness accounts for hundreds of little factors that it would be extremely tedious to take into account in a game -- many of which are things that it is somewhere between very difficult and effectively impossible to discern.

 

Also, beyond that and personally, I find it more interesting to have a chance of failure at things that my character might be expected to succeed at, and a chance of success at things they might not be expected to succeed at -- to a point, of course.

 

The problem often manifested when your stats were good enough for you to succeed, but low enough to frequently fail. Since a huge difference between a Pen and Paper RPG and a CRPG is the ability to save and reload, this often made for an unfortunate meta-game, where you memorise a dialogue tree, and save/reload until you get it right, and you pass the right skill checks. Is it cheating? Yeah, kind of. But most games I've played completely lacks an incentive for accepting a bad roll of the dice. There are ways around it, but while a bit of randomness in combat makes it less predictable and more fun, I think it's a mistake in non-combat situations.

 

The failure is in itself the incentive. I accept the poor roll because I do not want my characters to always succeed, and often, it is just as interesting to fail as it is to succeed. If I for some reason do not feel like accepting failure, I can reload and try again. I don't enjoy cheating in that fashion, but if I did, why should it bother somebody else? It surely does not bother me that other people do it. The way I play the game has no effect on the way they play the game, and vice versa.

[Tusck]

Anything that will prevent min/maxing is good in my book. Make non-combat skills important.

[AGX-17]

Not going to bother with quotes, but following up on what AGX-17 said: Randomness in conversations is a cheap way of "simulating" all those little tiny details which normally affect other's perception of you, and the result is not very good. It's not that randomness itself is bad, it's that the game doesn't offer a plausible reason for the failure; there's no feedback (and no, seeing "persuasion check failed" printed on screen is not proper feedback).

 

In real life, you get subtle clues, and you can infer the reason from that; you didn't seem to care (lack of sleep), didn't seem very enthusiastic, etc.

 

With classical CRPGs, we have been conditioned to accept that everything is determined by a roll of a die, with some modifiers applied. Again, this is a legacy mechanic from way back when the only alternative was having a storyteller (DM) say "because I said so". Games which rely on pure randomness for non-combat situations are often based on one of the many pen and paper rule systems.

 

Given the choice between completely static conversation trees and random-influenced conversation trees, I'd choose the former, even if I don't think it's very good either.

 

That said, there are quite a few ways to avoid involving randomness. Some which have even been implemented in highly successful games. For example, Mass Effect use skill points to limit your conversation options (charm/intimidate), but the available dialogue options vary depending on other factors too; taking sides open some dialogue options and closes other. Some actions may alter, or even completely remove some conversations. Sure, there's a lot of other problems with Mass Effect, but in most cases there's an actual reason as to why you can't steer a conversation in a given direction.

 

When it comes to computer games, there's no real reason NOT to store all those tiny details which the randomness was intended to replace. Furthermore, you don't have to limit skill checks to a simple pass/fail, you can add other requirements too; maybe your trap-finding skills aren't that great, so you can't see the traps unless you get close enough. Bumping that skill would simply mean that you can see them from afar. In this case, randomness could also be involved, but it shouldn't affect the skill check, it should be applied to make the visibility range variable (simulating flickering lights and other things which could help or distract you).

 

Mass Effect is a poor model because it railroads you into either a good or evil (half the renegade options are just evil, and ME2 just backs that up by giving Shepard a "sith" look to go with Renegade play,) path without allowing for any gray area. You MUST take all Paragon options to maximize paragon points to get Paragon options in later missions, if you don't have enough points those options are locked out. It's a terrible system and should never be emulated. It was a poor system in ME1, and they kept it despite all the complaints in both 2 and 3. No subtlety and no appreciation for context or dynamism of personality there. You could be a lawfully good guy who's dealing with someone so abhorrent that you know it's for the greater good to just kill them without trial or reading their rights, you can't do that sort of thing in ME unless you take a certain number of Renegade options (many of which are full-blown evil, which flies in the face of roleplaying a good character.)

 

If you alternate between them, you can't choose either when the time comes that they ultimately matter in big decisions. There's no purple zone on the dialogue wheel in ME.

 

 

Anything that will prevent min/maxing is good in my book. Make non-combat skills important.

 

This. SO this. min/maxing and dump stats are a sin that must be cleansed. The worst part is guys like Bethesda who promote or build min/maxing into their systems (and deliberately inject it into existing systems like SPECIAL, a greater crime.) It destroys the experience, it attracts the wrong kinds of players and it's just ludicrous to suggest one man/woman can be the best at everything under the sun. Or over.

 

One of the things I love about Obsidian is their dedication to nonviolent solutions to conflicts. There are always chickenhawks who use the hyperbolic argument that "oh but what if the enemy is a mindless zombie horde where's your precious diplomacy?" but if your antagonists are a mindless zombie horde your game isn't worth playing in the first place.

Edited October 16, 2012 by AGX-17

[Gizmo]

I came here tonight looking to start this very topic; I'm glad that I found this one before making a new one.

 

I must have asked this question 15 times on the Chat during the live stream when they were answering questions.

 

My hope is for it to closely resemble Fallout 1; I greatly prefer a weighted percentile check to a threshold ~and do not want minigames (like lock picking) if it can be helped... Although, if it was highly influenced by PC stats and skill then it could be fun.

 

The Weighted Percentile system can encompass all of the implied elements of chance leading up to the event; where as skill thresholds do not.

Consider Speech checks: With a threshold system, the PC succeeds if they have sufficient points in the skill... despite anything at all that might make the challenge unusually difficult at that particular moment in time. An NPC could be on the fence about them [normally], but that night they could be distracted by something else; or who knows ~have a headache and not want to talk... yet the PC always succeeds under that kind of system. With Weighted Percent, the challenge would come with an element of risk to it ~something I sorely missed in FO3/NV. The PC could even have consummate skill and just make a mistake, (this happens in life). However the weighted aspect would make this outcome very unlikely in normal circumstances.

 

With Weighted Percentile, the PC has an innate/ or learned ability at a task, and can use it to overcome difficulties (like the above)... (In this case it could be that they were so good that they talked someone out of something they wanted to do while that person was in no mood to be agreeable; and even looking to refuse [iE. a percentile penalty].

 

A Weighted Percentile system sets the bar for success (like to pick a lock) and the PC tries their best within the bounds of their personal ability, and they can fail; or they can succeed ~or they can really fail (jamming the lock); or really succeed (open the lock without scratches, and in a short time). Along the lines of the lock example... Percentile also allows the occasional fluke of luck ~it could be possible to pick open the lock as a complete novice (with a 3% chance of success) ~because it could happen; it's possible; it's very unlikely, but not impossible.

*However: A hard lock with a penalty that exceeds the PC's percentile skill at picking locks would be impossible. But on the flip-side... anyone can open a consumer-grade lock if they have a lot of time to keep trying ~but they will certainly scratch it up; showing that it was tampered with; (I wish a RPG would come around that gave guards a slight chance of noticing details like that).

 

About Save-Scumming: People will do that anyway; engineering around that is kind of almost the same sort of thing as DRM ~the intention is to discourage a behavior, yet the ones doing the behavior will do so regardless; while the rest are equally affected by the attempted 'cure'. Who cares if they save-scum? Who cares if they reload every 2 minutes? They are playing the game, and it's keeping them busy ~that's what games do.

 

What would be grand (for starters), is to not telegraph skill checks in the dialogs, and to just show the results of the-behind-the-scene calculations at work. Reveal dialog options only if the PC is capable of ~or lucky enough to get those choices. With other skills, they could have envelope of varied success or failure ~with extremes for both; and occasionally have the opposite result be the most interesting.

 

Lastly... It could be really cool to have the game be 'cognizant' of the player's actions throughout the the program (including the menus ). I'm a real fan of the idea that a game could perhaps pre-calculate a long list 'rolls' and match it to the save-game; that the rolls would survive game reloads and always be the same unless it was truly a new attempt at something; (if someone really wants to hex-edit ram or the game files ~let them ).

Edited October 18, 2012 by Gizmo

[Jarrakul]

I like the idea of a player-skill-based minigame, actually. And why not? Player skill matters in combat, why not elsewhere? Think Deus Ex: Human Revolution. That game had the best persuasion minigame I have ever seen. It was the best thing about the game, in fact. I would kill for another game with a persuasion system like DE:HR's. So add one in Eternity. More ranks in the Persuade skill might give you more information about the target, eliminate wrong answers, start you off closer to convincing the other guy, etc. The point here is that it'd take player thought and input. Sure, you could still reload and try again, but you can do that for combat too. The point is to make it more interesting than "roll, reload, roll again." And a persuasion system like DE:HR's would do that.

[UncleBourbon]

I'm a fan of partial randomness, probably from my background in later edition D&D. What I mean is that you have a skill level, but also a dice roll. So say you've devoted a couple of years to carpentry - enough to make a meager living. Well, building a birdhouse should be a simple matter - assuming it is a simple birdhouse. In several of the table top games I played in, the DM was very creative on things like this. (Granted, crafting is a good bit different from persuasion and such, but I'll get back to it in a second). So a simple birdhouse is a simple matter and as such the DM generally would say "well, I'll let you roll to build it, and have a ten or so chance of failure, or double the time and get it done certainly," but maybe the apprentice carpenter wants to try to make a very elaborate birdhouse complete with a bird feeder and a bird bath. Well, that increases the difficulty, so it adds a significant element of failure again.

Getting back to persuasion, I think certain levels of charisma/skills should unlock certain options/trees, but that it should likely be a matter of gradation. That is, there are a series of possible outcomes from trying to persuade someone. An example which comes to mind is trying to convince a merchant to give some of his wares to the local guards to help defend the town from an oncoming invasion. If you're a real smooth talker, you may have access to a really daring option like trying to get him to not only hand over the gear, but perhaps his bodyguard and/or his smuggled-in narcotics to be used as poison agents. Otherwise, the standard tree might have a complete failure response along the lines of "no way and I going to just hand this stuff over - I'll never see it again, and I'll not only be stuck in this battleground, but be broke" which is possible only with a really low roll and skill/bonus total. A soft failure would be the merchant offering significant discounts/kit deals to the guardsmen, and an intriuiged but unconvinced might be him telling you to go convince an old, retired veteran NPC to rejoin the guard or something and he'd help. Maybe a soft success of him giving his surplus stock over, and a strong success of him handing over his best pieces (with receipts and promises of repayment, of course). I would also like to see things like a hidden, running fame/renown score being taken into account and perhaps an order of things making the more optimal outcomes more likely. And example here would be getting the aforementioned veteran back into the guard would help persuade the merchant, but if you manage to persuade the merchant first you could more easily convince the veteran. Likewise, I think there should be negatives such as an old mage who had bad run-ins with the veteran and frequently overcharged by the merchant making success more difficult and vice-versa.

Of course, this makes VERY complicated dialogue trees and checks. Having a DM on hand who is in real-time generating the responses is a far easier solution, but also a far bit more difficult to replicate. Basically, what I suggest means something like two or three options in approaches (good/lawful/peaceful and the opposite, or perhaps just friendly/not so) and multiple levels of success on each of those... So in game? I'd like this sort of system for companion quests, if they exist, and maybe major storyline NPCs. For example, I'd like companion influence losses to be partially avoidable through smooth talking or exhibiting of sufficient skills. But most lesser NPCs I think should be easily persuaded via fame/money and any activities you've partaken of which immediately impacts them. For instance, if you walk into a small village and get into an argument with a guy walking out of a bar, and end up killing him, his brother might refuse to do business with you - even if you convince him the guy was the aggressor, he should say something like "Yeah, he was a lousy, stupid drunk, but he was family all the same, and I don't deal with the killers of my kin." Of course, this depends on the supposed family relationship and such...

Edited October 18, 2012 by UncleBourbon

[Zeyelth]

Lots of opinions, and an actual discussion on the pros and cons of various mechanics. Awesome! This is why I created the thread in the first place.

 

You shouldn't try to stop people from reloading saves, or make decisions about game mechanics with this intent.

 

People will cheat/mod/save whatever, it's their game let them! If someone gets more joy out of loading and loading it until it's what they want then so be it, why should we stop them?

It doesn't hurt people who don't want to play that way, its a choice.

 

This thread is not an argument against people who save/reload to optimise their playthrough; I'm fine with people cheating in single player games. I'm actually in favour of mods and tools which manipulate game data. I love games with an intact debug console.

 

Cheating is only a problem if it negatively impacts the experience for others (i.e. unbalances multiplayer games, devalues achievements, etc.) Ultimately, games are entertainment, and if optimising stats via meta-gaming increases the entertainment value for some, then more power to them. That said, I still believe that it's a flawed mechanic, and that it can be improved upon to the point where you no longer feel that save/reloading is necessary, even if the outcome is less than ideal.

 

The incentive for sticking with a bad roll is it makes your victories more meaningful, there shouldn't be some other in-game reward or incentive. Don't take out randomness in non-combat scenarios to limit the amount of saving/reloading.

Some people like to stick with their bad decisions even if they could reload.

 

I actually try to avoid reloading when a bad roll improves the flow of the story. Likewise, I can sometimes reload when an outcome felt unrealistically good. I don't particularly care if people cheat, but I don't want to feel that reloading is necessary for the story to make sense; it's a workaround for a broken mechanic, or at best, an unbalanced design.

 

And by incentive, I don't mean some powerful item, skill bonuses, or anything of the sort. I'm talking about making the player feel satisfied by the outcome, even if it's a failure. Something which makes the player think: Well, that sucked, but I totally deserved it. The incentive would be more along the lines of a good narrative flow.

 

Part of my problem with the "current" design is that you are rewarded for succeeding, or punished for failing, however you want to look at it. In other words, a reward system is applied to a mechanic which you can't really control. The result is that the game actually gives you an incentive to cheat, not the other way around.

 

Short of complete simulation, which is really not feasible, I believe that randomness is the best way.

 

This is a rather binary statement, and is an attitude I find to be rather counter-productive. No offence, but it reads like "We can't simulate the world perfectly, so there's no reason to even try."

 

A better solution likely lies somewhere in between, with variables you can control (accumulated actions and events throughout the game), combined with some randomness to give it flavour, without taking over completely. I'd like to go over some detailed design suggestions, but it's getting late, and this post is already getting far too long.

 

It's a matter of opinion or belief whether or not luck plays a part in real-world interactions and skills, but it is an observable fact that things do not always go precisely the same way every time you try them. One could juggle four balls in the same pattern at three different times in the day, their agility not having changed, and their skill presumably not having changed in that short an amount of time. One of those times they might do it flawlessly, one of those times they might drop the balls. The randomness accounts for hundreds of little factors that it would be extremely tedious to take into account in a game -- many of which are things that it is somewhere between very difficult and effectively impossible to discern.

 

 

Indeed, we don't have to implement the butterfly effect, but these type of seemingly random events does not translate well to games. At least not if they are replaced by a random number generator.

 

Two things:

1. If a skilled person fumbles, he or she will likely know why. Maybe a sudden gust caused a loss of balance, or sleep deprivation caused a loss of focus. We don't get this with the way skill-based dialogue options are currently implemented; there's no feedback. No reason.
    
   
2. A skilled person will be able to recover in the face of failure. If a smooth talker says something inappropriate, he or she could quickly turn this around by making a joke about it, drawing attention away from it. Likewise, a juggler could counter an accidental stumble by throwing the balls higher, giving him/her time to catch the one s/he's about to drop, etc.
   

Both of these points are possible in pen and paper RPGs, because the DM can give detailed explanations, and adapt to allow players to recover from fatal mistakes if they are creative enough (and if it's appropriate).

 

Also, beyond that and personally, I find it more interesting to have a chance of failure at things that my character might be expected to succeed at, and a chance of success at things they might not be expected to succeed at -- to a point, of course.

 

Agreed, with emphasis on "to a point". It shouldn't feel as if the failure is out of character, or if it is, the game should provide a reason for it. Maybe the character is drunk, or is otherwise in a poor/awesome mental state. Though again, things like that could actually be used to replace randomness. Given enough of them, the outcome will still be variable, but you can also figure out why, after the facts. Especially if hints are dropped in the response to your attempt.

 

If my hypothetical bard, with a high success rate of persuading people, had one too many pints, and suddenly got rejected with a response of "Go home bard, you're drunk." I wouldn't even want to reload, because it's actually quite appropriate.

 

Mass Effect is a poor model because it railroads you into either a good or evil (half the renegade options are just evil, and ME2 just backs that up by giving Shepard a "sith" look to go with Renegade play,) path without allowing for any gray area. You MUST take all Paragon options to maximize paragon points to get Paragon options in later missions, if you don't have enough points those options are locked out. It's a terrible system and should never be emulated. It was a poor system in ME1, and they kept it despite all the complaints in both 2 and 3. No subtlety and no appreciation for context or dynamism of personality there. You could be a lawfully good guy who's dealing with someone so abhorrent that you know it's for the greater good to just kill them without trial or reading their rights, you can't do that sort of thing in ME unless you take a certain number of Renegade options (many of which are full-blown evil, which flies in the face of roleplaying a good character.)

 

If you alternate between them, you can't choose either when the time comes that they ultimately matter in big decisions. There's no purple zone on the dialogue wheel in ME.

 

Yep, that's one of the biggest flaws with Mass Effect's way of doing things. What I did want to bring up with my example though, was how your responses are remembered, and will be referenced later on, sometimes even in later games. Without going too deep into spoiler territory, one example which comes to mind is the Rannoch mission, where choices from all three games will affect the available conversation options at a rather critical moment. Sure, the Renegade/Paragon points as an implementation for reputation could be better, and we could argue about some of the details of this mission and resulting conversation, but the key point is that success or failure is largely based on your own actions; ultimately you only have yourself to blame.

[Mr. Magniloquent]

The more skill checks, the better. Skills differentiate the classes deeply, and make the unique assets of a class much more immersive and prevalent. When only combat skills matter, class almost become irrelevant so long as you can kill something.

 

I've had to use skill and ability checks walk a normal pace through a blizzard, step across a stream without slipping, etc. These make your choice of class and build far more significant, and really make every task a team effort.

[Larkaloke]

Short of complete simulation, which is really not feasible, I believe that randomness is the best way.

 

This is a rather binary statement, and is an attitude I find to be rather counter-productive. No offence, but it reads like "We can't simulate the world perfectly, so there's no reason to even try."

 

A better solution likely lies somewhere in between, with variables you can control (accumulated actions and events throughout the game), combined with some randomness to give it flavour, without taking over completely. I'd like to go over some detailed design suggestions, but it's getting late, and this post is already getting far too long.

 

It would seem my words were ill-chosen, then, because that isn't what I meant. I believe it would be too complicated and time-consuming to implement a simulation of various factors that might cause a failed skill check to occur without any random factor, but that I think that randomness serves the same purpose so long as there are limits on it.

 

I'm not saying that any one of the various ways that random skills have been handled in the past are themselves the best thing, just that something with a base of randomness is. I do think that major environmental factors should be taken into account in a non-random fashion (things such as the surface you are attempting to balance on being icy, the character being drunk, the lock being slightly jammed, at cetera), but that all of the minor, difficult to perceive ones be respresented by a roll of the dice or similar.

 

It's a matter of opinion or belief whether or not luck plays a part in real-world interactions and skills, but it is an observable fact that things do not always go precisely the same way every time you try them. One could juggle four balls in the same pattern at three different times in the day, their agility not having changed, and their skill presumably not having changed in that short an amount of time. One of those times they might do it flawlessly, one of those times they might drop the balls. The randomness accounts for hundreds of little factors that it would be extremely tedious to take into account in a game -- many of which are things that it is somewhere between very difficult and effectively impossible to discern.

 

 

Indeed, we don't have to implement the butterfly effect, but these type of seemingly random events does not translate well to games. At least not if they are replaced by a random number generator.

 

Two things:

1. If a skilled person fumbles, he or she will likely know why. Maybe a sudden gust caused a loss of balance, or sleep deprivation caused a loss of focus. We don't get this with the way skill-based dialogue options are currently implemented; there's no feedback. No reason.
    
   
2. A skilled person will be able to recover in the face of failure. If a smooth talker says something inappropriate, he or she could quickly turn this around by making a joke about it, drawing attention away from it. Likewise, a juggler could counter an accidental stumble by throwing the balls higher, giving him/her time to catch the one s/he's about to drop, etc.
   

Both of these points are possible in pen and paper RPGs, because the DM can give detailed explanations, and adapt to allow players to recover from fatal mistakes if they are creative enough (and if it's appropriate).

 

I agree with those two things, but I disagree about how well they translate. I think that, with a few minor tweaks, it would work fine. Perhaps implementing several different levels of success and failure would help, so you could have complete success but also the barely-recovered-from sort of success (perhaps more similar to how it's handled in Rolemaster than it is in D&D).

 

Also, beyond that and personally, I find it more interesting to have a chance of failure at things that my character might be expected to succeed at, and a chance of success at things they might not be expected to succeed at -- to a point, of course.

 

Agreed, with emphasis on "to a point". It shouldn't feel as if the failure is out of character, or if it is, the game should provide a reason for it. Maybe the character is drunk, or is otherwise in a poor/awesome mental state. Though again, things like that could actually be used to replace randomness. Given enough of them, the outcome will still be variable, but you can also figure out why, after the facts. Especially if hints are dropped in the response to your attempt.

 

If my hypothetical bard, with a high success rate of persuading people, had one too many pints, and suddenly got rejected with a response of "Go home bard, you're drunk." I wouldn't even want to reload, because it's actually quite appropriate.

 

Sure, I think that adding things like that on would be great. I'd prefer it as a more additional sort of thing, but I think it would be a good thing. Specifically with the persuasion, a lot of that would also depend on the NPC that the character is trying to persuade -- things such as their mood at the time, how much they're ever inclined to do favours for people (or what have you), and all that. You could enter such data in for every single NPC no matter how minor, but that seems unnecessarily tedious to me, so I think it's best that that sort of thing be covered by the randomness.

 

Although, I suppose, you could also assign those values randomly to minor NPCs, which might come out to about the same thing.

 

Basically, I agree that randomness is something of a cheap substitute for simulation -- I'm just not sure of whether or not it would be worth devoting the time and effort to come up with a less cheap way. That would, I think, depend on how much time and effort was involved, and how much of a noticeable difference would occur in the end result. It could very well end up still looking random if you didn't know the system, and I'm not sure I'd want all of the factors that went into the failure to be listed every time a skill check was made... although, again, that could also end up being interesting; hard to say.

[mokona]

Let our choices, in conversations and actions, dictate the non-combat results instead of fake numbers designed to mathematically describe a morals/personality.

[Teahive]

With combat you have a string of tactical decisions and dice rolls, and a bad roll can add to the tension, might make you use that expensive scroll or potion you saved up for a rainy day, and generally adds to the feeling of overcoming a real challenge.

 

Non-combat parts of the game need to take a slice from this. Bad rolls are ok if at least something interesting happens. Instead of a single check failure leading to "I don't like you, I'm never going to talk to you again", at least give the player some opportunities to respond to a bad roll. Maybe you can do the NPC a favour to win back their trust (which might anger someone else), bribe them, threaten them, make them forget, or maybe after repeated check failure they give you misleading information, directing you into an ambush. There should be multiple levels of success, and different paths with interesting decisions to get there.