Random Success

Page: 1 2
ID:153510
 
Nov 24 2003, 10:15 am
Well since I finally have a bit of time to expend I'll reply to the thread started over in "Them lively little critters!".

Spuzzum wrote:
That's just the thing, though -- tactics are by their nature techniques which must be invented in each individual situation, not techniques which must be applied to each individual situation.

Yeah situations may change but when you develop a tactic you use past experience to try and determine the best method to solve a situation. If something you did worked out exceptionally well before and now it fails miserably with little to no indication why the player is just going to be confused. The same can happen if an effective techinique fails to bad luck. This can make it really hard to tell what is and isn't effective because of random luck. Luck in real events will make it hard to determine what's effective or not, but the luck in the realistic case isn't from random numbers it's just from too many variables to keep track of. Abstracting the millions on variables into a few using random numbers is unrealistic and therefore it isn't the best way to simulate this since it can generate very unbelievable situations which would work much differently in reality. Removing the random element from game mechanics and using a simple model is just as unrealistic but for gameplay reasons will make winning the game a matter of developing effective tactics/stratagy instead of doing the same thing over and over again reloading when you fail.

Someone can't win a war just by reading Sun Tzu's Art of War and being able to shoot a gun or swing a sword -- you need to be able to adapt to situations as they're presented. The tactic of throwing one enemy into another to bowl both off the edge of a cliff will not usually work twice.

If you are more experianced with a weapon than your opponent, have a larger knowlege of tactics, and past experience. You have little chance if any of failure except in extreme situations.

But as I noted in the real world success isn't determined by random numbers, just a large number of variables that you can't keep track of but if you're smart you'll know all the important ones. When you fail it isn't because of a bad dice roll it's because of some concrete reason which you can figure out given the time. This isn't the case in a game where all these variables are encapsulated into a random number.

The real problem is that most games use hard-coded tactics, making combat just a matter of learning one tactic to forever be capable of defeating the same enemy again.

This is the case in a very poorly designed game. But if the game is designed well and in a later level the game could add in a unit which could counter previous tactics forcing the player to rethink their actions. This doesn't involved random numbers just good game design. Starcraft doesn't involve any random mechanics and has a few stratagies planned out by the designers but that doesn't mean that players can't come up with thier own more effective stratagies/tactics.

Granted, not everyone checks out every monster they meet, but if the way the game worked needed them to, and they liked the game, they'd certainly learn to do so.

Or they could just keep reloading until the random numbers favored them instead of the monster. In a deterministic game it becomes more important to figure out whats effective vs what since if they keep doing the same stupid thing they'll fail in the same stupid way over and over again.

The command interface and interactivity should always remain the same, but the actual world shouldn't be predictable (in my opinion, for this hypothetical game -- not for roguelikes) by any means.

Like I said a well designed game will present progressivly more difficult challanges by adding new game elements into the game which force the player to rethink their actions. Random numbers just make how powerful a player or enemy is a fuzzy value. That just means the player will occasionally have to work with a worse case scenario which usually isn't much different than a good sceneraio it just involves either more healing or being more careful.

For example, whenever I drive, there are a billion things that could happen. That car that I'm watching on that side road could either drive out in front of me or wait in its lane. It could obstruct my view of a pedestrian. It could even get rear-ended and sent out into the street. No amount of training can make me react better to those situations -- it's intuition and coordination.

No just one thing can happen. What outcome that is is hard to predict but it's all based on deterministic data not random rolls. The reason past experience works is because you can think up reasons for the actions that occur and work accordingly. You may not be able to predict every event but most things shouldn't be hard to predict except in extreme cases but these themselves aren't random and occur for a reason even if you don't know it.

As I noted earlier it's folly to beleive that taking a large number of factors and make them one random roll is in any way shape or form close to being realistic. You are much better off implementing a complex deterministic system, where the player might not initially be able to figure out by can figure out most things by the end of the game. But you shouldn't even worry about this from a realism stand point since that matters little in the end. Having random results will hamper tactical play since the player can win by being lucky so they'll end up just doing the same thing over and over until they're successfull.

Say you have some nasty demon guarding a powerful item. The player at his current level and under his current tactics has a 25% chance of winning. This isn't much and in any situation that the player will lose a lot by attempting the fight will probably be warded away. But in many games(single player) it's just a matter of going back to the last save point which is usually not far away especially since many games put save points next to tough fights or you're allow to save at any time. In this case even with bad luck the player may only have to try 6-7 times before winning a fight they reasonably shouldn't be able to. If you make things deterministic it becomes a matter of you can or can't win under specific tactics not how good you are at getting good random numbers. In the previous example the player would not be able to win until his character was tougher or the player found a more effective way of taking out the demon. This rewards the player for being perceptive rather than punishing them at random.

Leftley wrote:
Yeah. Like one time, I cast a death spell and it killed the target, and then I cast the death spell on another monster of the same type and it didn't work. Or one time I did 24 damage with an attack, and then the very next swing only did 9. And then the next attack against the same monster missed, which is entirely different!

Ahh the death spell. That is the perfect example. I'm assuming you're refering to the Final Fantasy variation where the spell has a small chance of finishing off the enemy. How often have you used the instant death spell in a Final Fantasy game? If you're like just about everyone else and it isn't FF6 then you probably never use it frequently if at all. You can't incorperate it into any tactic since that tactic will more than likely fail since the spell is too unpredictable to have any decent effect when you need it to have an effect. It worked so well in FF6 because you could make it predictable then it became a very effective tactic and was actually used. Same thing goes for most systems. The instant death type stuff is never effective while it's unpredictable and is only effective if the system allows you to somehow make it predictable(ie in DnD you use stuff to reduce the enemes save vs death or you only use it on monster you know to have low save vs death like low constitution monsters like mages).

Also in the Final Fantasy games only a wierd weapon would have such a high damage disparity(unless one of those hits was a critical). For just about every weapon the difference between the maximum damage of an attack and the minimum is usually about or under 5% of the maximum damage the weapon can deal. So in the end getting all minimum damage attacks wouldn't be much different than getting all maximum damage attacks. Removing this difference would have little to no effect on the gameplay whatsoever. The Final Fantasy games also all but removed the chance to miss a monster except in special cases. So except for a few status effect spells and instant death spells the combat system is very deterministic in the Final Fantasy games(well past the NES ones which usually did have high dispariaty in damage done by spells). And most the status effect spells are completly worthless since they tend to not work in battles where theyd actually be useful(ie boss ones). All the status effect spells that affect the player always work and have fixed results. This is great but the Final Fantasy games are generally bone head easy and leave little room for tactics except to make an easy fight easier.

Many people(me included) can't stand the first few levels in AD&D because the outcome of many situations is entirley random and isn't affected much by your choice unless it's to avoid them. In general fighters have a 50% chance to hit and will deal anywhere from 1 to 8 points of damage with a long sword. Combat generally is resolved just by the person with the better dice rolls. The most useful wizard spell is magic missile because it always hits making it very effective. Some spells may cause more damage when you first get them but they generally require a to-hit roll which means more thn likly they'll miss.

In Diablo 2 how many high level sorceresses do you see using lighting or chain lightning? I never see many but it's probably due to the fact that their minimum damage is always one so you may have to hit a single monster many times to kill it even if it would only take two hits at the average damage. This is annoying and can result in the death of your character which is real bad in hardcore mode. The effective sorceresses I've see use static charge followed by ice orb. Static Charage reduces the enemy health by a fixed percent and after enough uses can then be finished off with an ice orb which deals a good amount of damage even in the worst case scenario.

The most effective tactics are those which will work well and not fail most of them time. So in general good players of a certain game will tend to stick with things which will have a predictable outcome. This is NOT a design flaw. Players like to be rewarded for figuring out this type of stuff. You shouldn't punish players by making their past experience inneffective or by having them lose from streaks of bad luck. That's just frusterating not fun.
Nov 24 2003, 12:25 pm
On the flip side, deterministic games have considerably less replay value (and to a certain extent, less "play value"), because there is a finite number of situations that the player must be able to handle, and once they have mastered all of these situations there's just not much point. It's largely up to the player to seek out unique new situations, whereas in a more random game, most situations are likely to be unique. Deterministic games reward players who systematically try tactic after tactic until they develop a foolproof tactic for a certain situation; more random games reward players who are more adaptable and can roll with the punches.

Yes, random games suffer from reload syndrome... but are deterministic games really any better on that count? I mean, sure, in a random game players will reload and try again until they get better results, but in a deterministic game players will do the exact same thing--the only difference being that they'll vary their tactics slightly. "Oh, spell A doesn't work. Reload, try spell B. Oh, spell B doesn't work. Reload, try spell C. Hey, spell C gives me an easy win."

Actually, that's not true, because players will generally vary their tactics in a random game too, and there is one other difference: in a random game, if a player keeps reloading until they get really lucky and win, then they still have to keep doing the reload thing every time they encounter a similar situation. In a deterministic game, once a player arrives at the killer tactic for a certain situation, then their success is pretty much assured in all similar situations in the future.
Nov 24 2003, 3:47 pm
In response to Leftley
A few comments.

For determinism:
As you say, in reality there are no "random" rolls (actually, the scientists say there are at a quantum level, but I don't believe them.)

However, I disagree that random rolls should be completly removed. There are just too many things that simply cannot be tracked that random rolls simulate.

I think your bottom line is that too much randomness takes any sort of strategy or planning, and ultimatly fun, from a game (D2 Chain lightning = perfect example).

For Randomness:
If a game is predictable then it loses all replayablity. Once you figure out what works in a given scenario, then that scenario loses all of its fun.


What I propose is an alternative in the middle. I agree that when making rolls more things should be factored in. In an attack roll, for example: the surface you're on, the condition and types of both weapons and armors, your stamina, your health, the weather, if you can see, smell, hear well, et cetera. This gives the players a lot more to use when figureing out what to do, creating more tactics, and more fun.

The place where randomness is appropriate, is in making two identical encounters with goblins very different. Not more or less difficult, but different. So make it random whether the goblin pulls out a sword or a bow. Add 5-10% randomness on damage dealt. Make it random whether the goblin coats his arrows in blind poison or unbalance poison. Make a battles very varied, but more or less the same difficulty.

In the end, deterministic or random, I think that there are two main things here.

1) Huge amounts of both randomness and blatent determinism sucks.

2) The options, and the more effective things that a player can do, the more strategy and fun will be created.

Anyway, just my 3 cents.

Peace
Nov 25 2003, 6:58 am
In response to Leftley
On the flip side, deterministic games have considerably less replay value (and to a certain extent, less "play value"), because there is a finite number of situations that the player must be able to handle, and once they have mastered all of these situations there's just not much point. It's largely up to the player to seek out unique new situations, whereas in a more random game, most situations are likely to be unique. Deterministic games reward players who systematically try tactic after tactic until they develop a foolproof tactic for a certain situation; more random games reward players who are more adaptable and can roll with the punches.

A deterministic game with preset scenarios and no random AI logic will result in no replay-ability since it's just a matter of following a specific set of commands to win. But a deterministic game with either fluctuating AI or randomized scenarios will present just as many unique challanges whether or not the damage is random. How much replay value a game has has nothing to do with whether the game mechanics are random or not. Just look at all the RTS games. Rarely do they have any random factors at all(except for maybe some random AI behaviours) and they provide all sorts of challanges that require different techniques to win. If the damage or success of hits were random all it would do is have the difficulty become random. The same thing happens with RPGs. The only thing random damage and hit success adds is a random level of difficulty. Difficulty in a well designed game is never random and should pace well so the player can learn the mechanics gradually as well as being challanged the whole way through.

Yes, random games suffer from reload syndrome... but are deterministic games really any better on that count? I mean, sure, in a random game players will reload and try again until they get better results, but in a deterministic game players will do the exact same thing--the only difference being that they'll vary their tactics slightly.

Isn't that the whole point though? You want players to try new tactics not repeat the same process over and over until it works.

"Oh, spell A doesn't work. Reload, try spell B. Oh, spell B doesn't work. Reload, try spell C. Hey, spell C gives me an easy win."

In either a poorly designed game or a learning scenario this may be the case. But if that's as complex as the game gets random numbers won't make it any more fun.

Actually, that's not true, because players will generally vary their tactics in a random game too and there is one other difference: in a random game, if a player keeps reloading until they get really lucky and win, then they still have to keep doing the reload thing every time they encounter a similar situation. In a deterministic game, once a player arrives at the killer tactic for a certain situation, then their success is pretty much assured in all similar situations in the future.

In both random and deterministic games if the scenarios are that similiar the whole way through it's going to be a boring game. It's only obvious that a well designed game with random numbers thrown in is going to be more entertaining than a deterministic one that throws the same situation at you over and over again.

[Edit] And as far as the title goes. If I have an awsome and clever tactic that makes sense it better work provided I have the resources to do it effectivly and I know what I'm up against. Having to reload because of a bad roll is no fun.
Nov 25 2003, 8:53 am
In response to DarkHorsePlague
However, I disagree that random rolls should be completly removed. There are just too many things that simply cannot be tracked that random rolls simulate.

If you're making a game you should be more concerned over how fun it is not how well it simulates everything. It's the fun well balanced games that sell well not the super realistic ones.

If a game is predictable then it loses all replayablity. Once you figure out what works in a given scenario, then that scenario loses all of its fun.

Starcraft is entirely deterministic. If you play through the campaign the same way everything will happen the same way. If this is no fun why is it so popular? The answer is simple the campaign may be predictable but when you throw in multiplayer it's no longer predictable(for the most part :P). Despite having set rules that never deviate the tactics/stratagy involved must be developed as you play since the scenarios you are put in deviate. This can be simulated in single player games by having somewhat randomly set up scenarios.

What I propose is an alternative in the middle. I agree that when making rolls more things should be factored in. In an attack roll, for example: the surface you're on, the condition and types of both weapons and armors, your stamina, your health, the weather, if you can see, smell, hear well, et cetera. This gives the players a lot more to use when figureing out what to do, creating more tactics, and more fun.

It's simple enough to remove the random hit/miss while still factoring chance to hit. Just reduce the damage by the percent chance to hit. If you normally do 200 damage and have a 50% chance to hit something just reduce the damage done to 50% it's normal value and have it inflict 100 damage every attack. Over long periods of time this will amount to the same thing but in the short term it aleviates those annoying times where you miss 5 times in a row despite you should be able to hit every other time. You can't plan for this kind of bad luck and it usually just makes the player reload from the last save and try again since the player may have had a good tactic the first time through and only lost from some small chance that they'd miss that much despite the decent odds of hitting. It also prevents the reverse of a player hitting 5 times in a row effectivly being able to take out stuff he shouldn't be able to.

The place where randomness is appropriate, is in making two identical encounters with goblins very different. Not m
more or less difficult, but different.

If I manage to never miss and critical each hit it's much easier than if I end up missing 10 times in a row. That's dumb luck but could happen and has a direct impact on the difficulty of the fight. If you reduce the random ranges you eventually make the match more or less deterministic. A fight with 2 goblins should be pretty much the same every time you fight otherwise you just confuse the player, they'll get frusterated and decide your game is no fun. But if you change the fight somewhat by adding some other monster then it should take a different tactic to win that is some way related to just fighting 2 goblins but with a new constraint to make it more challanging.

So make it random whether the goblin pulls out a sword or a bow.

This is an AI decision which isn't part of the game mechanics itself and is something that would make it interesting.

Add 5-10% randomness on damage dealt.

Why bother? At this range on most monsters hitting everytime at the minimum damage would result in winning in the same amount of turns as doing all maximum damage. This does nothing for the game.

Make it random whether the goblin coats his arrows in blind poison or unbalance poison.

Again this is an AI or level design descision not a game mechanics one.

Make a battles very varied, but more or less the same difficulty.

Yeah this again has more to do with random level design which I stated was good.

1) Huge amounts of both randomness and blatent determinism sucks.

No it's were it's used that makes it suck. It's good to have deterministic game mechanics. Just look at Starcraft there isn't anything random there and people love it despite not being random. Randomness is good to have but not in the way core things work like combat mecanics.

2) The options, and the more effective things that a player can do, the more strategy and fun will be created.

Stratagy is formed through past experience and knowlege of how things work and then applying that information to a new situation. When you introduce a random element the player may or may not run into a worst case scenario. This might not be entirly bad in a single player game(but can be quite annoying especially if you can't save often or load times are bad), but in a multiplayer game this can lead to severe balance issues and allow players to win through dumb luck even if they had no strategic sense and the other play did.
Nov 25 2003, 2:03 pm
For example, whenever I drive, there are a billion things that could happen. That car that I'm watching on that side road could either drive out in front of me or wait in its lane. It could obstruct my view of a pedestrian. It could even get rear-ended and sent out into the street. No amount of training can make me react better to those situations -- it's intuition and coordination.

No just one thing can happen. What outcome that is is hard to predict but it's all based on deterministic data not random rolls. The reason past experience works is because you can think up reasons for the actions that occur and work accordingly. You may not be able to predict every event but most things shouldn't be hard to predict except in extreme cases but these themselves aren't random and occur for a reason even if you don't know it.

As I noted earlier it's folly to beleive that taking a large number of factors and make them one random roll is in any way shape or form close to being realistic. You are much better off implementing a complex deterministic system, where the player might not initially be able to figure out by can figure out most things by the end of the game. But you shouldn't even worry about this from a realism stand point since that matters little in the end. Having random results will hamper tactical play since the player can win by being lucky so they'll end up just doing the same thing over and over until they're successfull.

It's also folly to assume that anything more than abstraction is practical. The fundamental principles behind chaos theory dictate that systems always reproduce themselves precisely if all of the variables were the same, but those variables cannot always be the same. At the very least, it establishes that humans are absolutely without-a-doubt incapable of recognising more than a shallow fraction of those variables, and furthermore are limited to five senses, causing variables which cannot be detected at all by any of them (thus, from my limited perspective, billions upon billions of things could in fact happen as I'm driving along). When it boils down to it, randomness suits those unknowns extremely well.

Instead of wasting years and years of programming effort to model every possible aspect of the world, everything must be abstracted at one point or another. Determinism is not the solution because the more variables that are added, the less deterministic any given situation appears to be. That's the basis of randomness. Sometimes you might strike an arm, sometimes you might strike a leg -- if the game manually calculated the path of the blade as opposed to the other person's bodily movement, and produced a deterministic result, then sure, it'd be more realistic... but it wouldn't be practical.


Say you have some nasty demon guarding a powerful item. The player at his current level and under his current tactics has a 25% chance of winning. This isn't much and in any situation that the player will lose a lot by attempting the fight will probably be warded away. But in many games(single player) it's just a matter of going back to the last save point which is usually not far away especially since many games put save points next to tough fights or you're allow to save at any time. In this case even with bad luck the player may only have to try 6-7 times before winning a fight they reasonably shouldn't be able to. If you make things deterministic it becomes a matter of you can or can't win under specific tactics not how good you are at getting good random numbers. In the previous example the player would not be able to win until his character was tougher or the player found a more effective way of taking out the demon. This rewards the player for being perceptive rather than punishing them at random.

However, there are two points to be made; one, all randomness eventually boils down to tendency; and two, knowing the precise result of any action ahead of time takes out much of the thrill of it. Regarding the first, if I rolled a six-sided die 100 times, then I could reasonably be expected to achieve a result of around 350. Each random roll would be different, and the final result wouldn't be precisely 350, but the tendency is there nonetheless. It's masking the determinism, which forces the player to think less in terms of strategy and more in terms of tactics. Regarding the second, sports, games, and gambling are based on the thrill of the unknown. If I knew that I would win my next Poker hand, I'd bet, and if I knew I'd lose, I'd pass. It just wouldn't be a game, because it would eliminate all of the worry out of it.

Determinism is the same way -- it's extremely practical for strategic-scale, because all randomness eventually boils down to a tendency, but it's incredibly impractical for tactical-scale, which is what almost all RPGs are set in.


(By the way... it's "frustrating". ;-))
Nov 25 2003, 4:31 pm
In response to Theodis
Randomness in combat? Combat is, to some extent random. Obviously its not really random since everything has its cause and effect, but as far as the combatants are concerned there is randomness to it. This applies especially to a large battle, but this even applies to a duel in a very static environment. Sometimes armor deflects a blow and sometimes it dents in. The structural integrity and molecular structure of the material of the armor determines this, but as far as the combatants are concerned sometimes a good blow will dent the armor and sometimes it won't. A skilled warrior will know how to best deliver his blows, but sometimes the blows will work and sometimes they won't. Sometimes one combatant will see a sign that his foe is kicking out of the corner of his eyes and react and sometimes he won't see it. Unless its a very small foe he can't concentrate on every part of him at once. A good warrior will know when to concentrate were and how to concentrate on a larger area, but he'll still miss some subtle things. Whether he misses something is based on where he was looking so its not random, but since he doesn't know how his opponent is going to react its random to him. There are thousands of factors in everything. Can they all be calculated? sure. But at some point a player can't account for all of them, especially in the heat of battle. At this point the result appears random to the player whether it is or whether it isn't. I've heard suggestions on messageboards before for making rpgs where archery is based on air humidity and temperature and wind speed and arrow smoothness and the direction and intensity of the sun and elevation differences and different air temperatures in different spots and string wear and so forth. But when it comes down to it, unless you have little dials everywhere, the player won't know every factor when he's pulling that bowstring and then the result is random to him. It might be nice to design the archery system like that for simulated experiments, but for a game all that matters is the players perception. If he cannot percieve the inner workings to some process than it is as good as random.
That isn't to say just make everything random. Flip a coin and heads the player wins and tails he loses. And the attack-reload-attack-reload-ad-infinitum method of playing a game is annoying. And getting 5 bad rolls and dying at the hands of a goblin is very annoying. Its more realistic, as combat is chaotic and veterans die and new soldiers live, but its less fun to the player. Absolutely it should be addressed. However, addressing it does not require the abolition of randomness in combat mechanics. Take the rougelike games for an example. Those are extremely random games, combat to some extent included. Yet the player doesn't die because of a string of bad rolls? why? because they can easily deal with most foes they'll face? So where's the challenge? The player faces hordes of monsters. He needs to know when to go deeper into the dungeon and when to stay where he is, when to flee and when to fight, when to walk into the middle of the room and when to stay in the hallway. Combat is random, the situations the player is presented with are random, but how the player progresses through the dungeon is deterministic. Angband, Rouge, Nethack, or any other of those vein don't suffer from the reloading problem (at least with non-cheating players) because they have permadeath. Its one way to solve the problem.
Letting the players actions directly effect the outcome of the game is definately important to making a game fun. Making a strictly deterministic game (like Starcraft) is definately one way of doing so. However, it is not the only way of accomplishing the goal. I come back to Angband variants time and time again. Making the player a slave to the dice is not a fun way of making a game, but that doesn't mean that randomness must be cut out entirely. If a combat system is engaging, challenging, and doesn't make a player feel cheated than its a good battle system. This can be accomplished with or without random numbers. Its the issues (reloading, combat where player input doesn't matter, too high chance of loss) that need addressing, not nescessarily random numbers. Of course, cutting them out is one way. Just not necessarily the only or clearly best way.
Nov 25 2003, 6:47 pm
In response to Theodis
Isn't that the whole point though? You want players to try new tactics not repeat the same process over and over until it works.

I believe the honorable represenative is suggesting is that this method does encourage one process over and over: the process of elimination. That's no more creative than the other way, and unless the possible strategies are extremely and subtley varied, it actually encourages less flexibility of thinking.
Nov 25 2003, 7:10 pm
In response to Theodis
Theodis wrote:
A deterministic game with preset scenarios and no random AI logic

You started this entire chain of discussion by asserting that AI logic should be transparent and predictable. Random AI logic is what you're arguing against, remember?

But a deterministic game with either fluctuating AI or randomized scenarios will present just as many unique challanges whether or not the damage is random. How much replay value a game has has nothing to do with whether the game mechanics are random or not.

I think you're drawing distinctions that don't really exist here; there's not that large of a difference between random scenarios and random damage/success rolls. The only difference is the degree of transparency: you might be able to see part of an encounter before you commit to your course of action, whereas you only see the results of a random action after the fact. But even making encounters too transparent will dull your replay value; if the player can peek into every factor of a randomly generated encounter before engaging, it's probably a foregone conclusion anyways.

Just look at all the RTS games. Rarely do they have any random factors at all(except for maybe some random AI behaviours) and they provide all sorts of challanges that require different techniques to win.

I think you have a typo. You accidentally wrote "all" in the first sentence instead of "the one or two most exceptional, outstanding RTS games ever".

If the damage or success of hits were random all it would do is have the difficulty become random. The same thing happens with RPGs. The only thing random damage and hit success adds is a random level of difficulty.

Whereas, of course, it's child's play to make randomly generated dungeons that will provide exactly the same level of difficulty every time!

Difficulty in a well designed game is never random and should pace well so the player can learn the mechanics gradually as well as being challanged the whole way through.

Difficulty in a well designed game is always random, unless the game is either absolutely unwinnable or absolutely unloseable; you don't know all the player's capabilities when you design the game, and the player doesn't know all your design tendencies when they play the game. There's very little practical difference between a quantity which is random and one which is unknown (and unknowable).

Sure, you can present players with identical obstacles... but even barring the fact that you yourself say that this leads to boring games with little replay value, identical obstacles don't mean an identical level of challenge.

Yes, random games suffer from reload syndrome... but are deterministic games really any better on that count? I mean, sure, in a random game players will reload and try again until they get better results, but in a deterministic game players will do the exact same thing--the only difference being that they'll vary their tactics slightly.
Isn't that the whole point though? You want players to try new tactics not repeat the same process over and over until it works.

By extension, you seem to think that in a random game the point is to repeat the same process over and over again until it works. In either a poorly designed game or a learning scenario this may be the case. But if that's as complex as the game gets non-random numbers won't make it any more fun.

Random failures (at least of the variety that induce repeated reloads) are Mother Nature's way of telling you that your strategy, while interesting, needs improving.

In both random and deterministic games if the scenarios are that similiar the whole way through it's going to be a boring game. It's only obvious that a well designed game with random numbers thrown in is going to be more entertaining than a deterministic one that throws the same situation at you over and over again.

Similarly, it's only obvious from your argument that a well designed game with non-random numbers is going to be more entertaining than a random one that throws the same situation at you over and over again (by way of continuous arbitrary game overs and subsequent reloads).

[Edit] And as far as the title goes. If I have an awsome and clever tactic that makes sense it better work provided I have the resources to do it effectivly and I know what I'm up against. Having to reload because of a bad roll is no fun.

If your strategy goes from flawless victory to total disaster over one bad roll, it's not quite so awesome and clever, now is it? Having to reload because of a bad roll is no fun, but I have no problem with having to adjust my tactics on the spot because of bad rolls.

While we're discussing the relative merits of random rolls... how much fun is it to have to reload because the encounter generator came up with a group of monsters that you can neither escape from nor kill since the game is deterministic and you don't have the resources necessary to beat them? How about replaying entire dungeon levels because you got crappy useless treasure?
Nov 26 2003, 8:58 am
In response to Leftley
You started this entire chain of discussion by asserting that AI logic should be transparent and predictable. Random AI logic is what you're arguing against, remember?

Must be changing my view as I go along then :P.

But a deterministic game with either fluctuating AI or randomized scenarios will present just as many unique challanges whether or not the damage is random. How much replay value a game has has nothing to do with whether the game mechanics are random or not.

Well I think the worst experiances I've had involved me failing to do something I had a good chance of doing not the decisions of the AI. When I write a game I want it to be challanging but not incorportate a large amount of aggrivation due to random losses. It's the least aggrivating games that people will play for long streches of time not the ones that constantly promote it. Avernum 3 is probably the game I've played for the longest strech since most of the experience is entirely fun there are random annoyances due to bad rolls, but this is alleviated by being able to save in any non-combat situation and loading is next to instantanious even on computers a decade old. In the end how fun a game is is all the matters as far as keeping a player playing(though to some graphics may be important for starting off :P)

I think you're drawing distinctions that don't really exist here; there's not that large of a difference between random scenarios and random damage/success rolls. The only difference is the degree of transparency: you might be able to see part of an encounter before you commit to your course of action, whereas you only see the results of a random action after the fact. But even making encounters too transparent will dull your replay value; if the player can peek into every factor of a randomly generated encounter before engaging, it's probably a foregone conclusion anyways.

Players(well at least me) tend to be more annoyed by when their character fails to do something more than when an enemy succedes. By removing random rolls at least on their side can reduce a lot of frustration which is good.

I think you have a typo. You accidentally wrote "all" in the first sentence instead of "the one or two most exceptional, outstanding RTS games ever".

Well the only RTS I can think of off hand that involves random success is Warcraft 3 but that's only for certain special abilities and mostly everything else is based on a non-random algorithm(actually I can't remember but the damage ranges might have been somewhat random). But all of the Westwood RTSs and older Blizzard ones don't use any random mechanis as far as I'm concerned. I don't know about the AI though.

Whereas, of course, it's child's play to make randomly generated dungeons that will provide exactly the same level of difficulty every time!

Entirely randomly generated dungeons that provide the right challange is definantly difficult and hasn't been done well. Diablo 2 pulls it off but doing so removes the feel of randomness since dungeon exits are generally the same distance apart each time and the enemy selection isn't very random. But setting up a few random scenarios on a static map(should be decided on the creation of a new game so players can't easily just keep reloading to get the scenario they want) can bring in some of the effect but without sacrificing control of the pacing of the game.

Difficulty in a well designed game is always random, unless the game is either absolutely unwinnable or absolutely unloseable; you don't know all the player's capabilities when you design the game, and the player doesn't know all your design tendencies when they play the game. There's very little practical difference between a quantity which is random and one which is unknown (and unknowable).

Just go and play any Nintendo made game :). They are the masters of building well paced games.

Sure, you can present players with identical obstacles... but even barring the fact that you yourself say that this leads to boring games with little replay value, identical obstacles don't mean an identical level of challenge.

Just look at Nintendo games like Super Mario 64. It's deterministic and paced very well. It may be boring after you've managed to get all the stars but by the time you get all the stars you played the game for long enough to be worth your money.

By extension, you seem to think that in a random game the point is to repeat the same process over and over again until it works. In either a poorly designed game or a learning scenario this may be the case. But if that's as complex as the game gets non-random numbers won't make it any more fun.

That's part of the point :P. There are many RPGs that don't have much as far as combat goes and just try to obscure the fact with random numbers. Console RPGs suffer much less from the reload syndrome because combat is much more deterministic compared to a game based on D&D rules like Balder's Gate.

Random failures (at least of the variety that induce repeated reloads) are Mother Nature's way of telling you that your strategy, while interesting, needs improving.

There are cases when your stratagies are perfect and you still fail because of bad luck. This is just randomly punishing the player.

Similarly, it's only obvious from your argument that a well designed game with non-random numbers is going to be more entertaining than a random one that throws the same situation at you over and over again (by way of continuous arbitrary game overs and subsequent reloads).

Yeah so stop assuming that a deterministic game will play out the exact same way everytime you run through it :P. It would be very hard to completly reproduce the exact same situation in a Starcraft scenario despite it's not random.

If your strategy goes from flawless victory to total disaster over one bad roll, it's not quite so awesome and clever, now is it? Having to reload because of a bad roll is no fun, but I have no problem with having to adjust my tactics on the spot because of bad rolls.

One or two bad rolls is bad but if you're like me you run into situations where you get them constantly throughout the fight :P. Atleast my magic missiles never fail(except for lousy spell resist)!

While we're discussing the relative merits of random rolls... how much fun is it to have to reload because the encounter generator came up with a group of monsters that you can neither escape from nor kill since the game is deterministic and you don't have the resources necessary to beat them?

In a deterministic game if I can avoid it this is better since I know I can't win and won't be tempted to just get through with constant reloads.

How about replaying entire dungeon levels because you got crappy useless treasure?

This is just bad game design. In most roguelikes the difficulty of the dungeon is based on how much good stuff there is so if it's a nasty map you're pretty much garunteed some good loot.
Nov 26 2003, 9:17 am
In response to Luap
If he cannot percieve the inner workings to some process than it is as good as random.

Exactly simulating a situations may be fun sometimes but is generally not a good idea. I much prefer racing games that give me some kind of weapon or attack so I can do something to other racers than just getting in the way. This isn't real but it is a lot of fun. Losing from a lot of factors you don't know about is not fun. This is why it takes a long time to actually get into a game like Sim City since if you neglect on minor thing it can completly turn success into rampant failure through a large chain of events. This may be real but is generally annoying until you know how to handle them. In cases like random combat you'll never be able to completly handle long streaks of bad luck which can inevitably cause you to lose. This may be somewhat realistic but isn't in any way fun.

And getting 5 bad rolls and dying at the hands of a goblin is very annoying. Its more realistic, as combat is chaotic and veterans die and new soldiers live, but its less fun to the player.

This is my main argument and why you should weed out stuff like this. In console RPGs they pretty much already have by limiting the random ranges so much.

Absolutely it should be addressed.

If it's not fun why bother?

However, addressing it does not require the abolition of randomness in combat mechanics. Take the rougelike games for an example. Those are extremely random games, combat to some extent included.

The levels are random and there is a high chance of failure for using skill/powers you aren't good at. But once you're high level you're next to garunteed to be able to cause a certain amount of damage to a certain amount of monsters if you know what you're doing. This is also the case in a D&D game. This is great but should be applied at the lower end too.

Yet the player doesn't die because of a string of bad rolls? why? because they can easily deal with most foes they'll face? So where's the challenge? The player faces hordes of monsters. He needs to know when to go deeper into the dungeon and when to stay where he is, when to flee and when to fight, when to walk into the middle of the room and when to stay in the hallway. Combat is random, the situations the player is presented with are random, but how the player progresses through the dungeon is deterministic. Angband, Rouge, Nethack, or any other of those vein don't suffer from the reloading problem (at least with non-cheating players) because they have permadeath. Its one way to solve the problem.

But this is at higher levels when you have higher amounts of control over what you know and having garunteed success at things like causeing certain amounts of damage. Early on you suffer from restarting many many times until you get a character that is capable of getting more predictable results in combat. This is real bad for new players. When I got into roguelikes it was from playing Ragnorak which allows backup saves. It wasn't until that that I got into Nethack and Angband.

Letting the players actions directly effect the outcome of the game is definately important to making a game fun. Making a strictly deterministic game (like Starcraft) is definately one way of doing so. However, it is not the only way of accomplishing the goal. I come back to Angband variants time and time again. Making the player a slave to the dice is not a fun way of making a game, but that doesn't mean that randomness must be cut out entirely.

I don't think it's really been done yet so I'll have to try and see how it turns out :). It may be the greatest thing since sliced bread.

If a combat system is engaging, challenging, and doesn't make a player feel cheated than its a good battle system. This can be accomplished with or without random numbers. Its the issues (reloading, combat where player input doesn't matter, too high chance of loss) that need addressing, not nescessarily random numbers. Of course, cutting them out is one way. Just not necessarily the only or clearly best way.

Yeah but what leads to failure when you know what you're doing? It's bad rolls so comming up with a good way to remove them is one step toward the solution. Anyway how else to you remove random failure without making the game super easy or killing the random aspect :P? If you have a good idea speak up cause I'd like to know :).
Nov 26 2003, 9:38 am
In response to Spuzzum
It's also folly to assume that anything more than abstraction is practical. The fundamental principles behind chaos theory dictate that systems always reproduce themselves precisely if all of the variables were the same, but those variables cannot always be the same. At the very least, it establishes that humans are absolutely without-a-doubt incapable of recognising more than a shallow fraction of those variables, and furthermore are limited to five senses, causing variables which cannot be detected at all by any of them (thus, from my limited perspective, billions upon billions of things could in fact happen as I'm driving along). When it boils down to it, randomness suits those unknowns extremely well.

But in a well done deterministic system in the long run it will turn out the same numbers as a random system which is my point. The only difference is on the short term there is no luck good or bad.

If you just reduce damage by the percent to miss and make it always hit it'll have the same effect over a large number of attacks, but on the short term it's easier to develop plans of attack.

So I don't think either is better or worse at trying to simulate an abstract situation it's just that the deterministic way will remove the inconviences that rise from a large number of random rolls. Console RPGs have delt with the problem by reducing the disparitiy of the rolls essentially removeing random loss since the worst case scenarios aren't much different than the best.

Instead of wasting years and years of programming effort to model every possible aspect of the world, everything must be abstracted at one point or another. Determinism is not the solution because the more variables that are added,

Not neccesarily you can easily keep in percent chance to hit but instead use it as reducing damage done. This doesn't add anything new, has the same effect in the long run, and makes things laregly less random. Instead of having damage ranges just take the average.

the less deterministic any given situation appears to be. That's the basis of randomness. Sometimes you might strike an arm, sometimes you might strike a leg -- if the game manually calculated the path of the blade as opposed to the other person's bodily movement, and produced a deterministic result, then sure, it'd be more realistic... but it wouldn't be practical.

Dealing with this is challanging to accuratly model in either a deterministic system of random. The best I think would be to take(as long as you don't care to offer descriptive results of the combat action) the percent chance to hit each body part. Apply each percent to the damage that would be cased by hitting that area, then take the average of this damage and use it as the final damage caused.

However, there are two points to be made; one, all randomness eventually boils down to tendency; and two, knowing the precise result of any action ahead of time takes out much of the thrill of it. Regarding the first, if I rolled a six-sided die 100 times, then I could reasonably be expected to achieve a result of around 350. Each random roll would be different, and the final result wouldn't be precisely 350, but the tendency is there nonetheless. It's masking the determinism, which forces the player to think less in terms of strategy and more in terms of tactics.

Yeah but does your character actually get 100 rolls? In the short run you can lose a lot faster if the demon gets a few good hits in a short amount of time preventing you from recovering. If the damage done by the demon is relativly the same each time the player has a better chance of determining the best time to recover. By removing random rolls the short term results won't vary as much if any from the long term results.

Regarding the second, sports, games, and gambling are based on the thrill of the unknown. If I knew that I would win my next Poker hand, I'd bet, and if I knew I'd lose, I'd pass. It just wouldn't be a game, because it would eliminate all of the worry out of it.

A good gambler knows what his chances are of winning and how to effectivly increase them.

Determinism is the same way -- it's extremely practical for strategic-scale, because all randomness eventually boils down to a tendency, but it's incredibly impractical for tactical-scale, which is what almost all RPGs are set in.

It just hasn't been done yet :). If I ever get around to making this type of game I'll make sure to make it deterministic to see what happens.

(By the way... it's "frustrating". ;-))
Heh I'll try and get it right in the future but that's not how I pronounce it so chances are I'll keep making the mistake :P.
Nov 26 2003, 11:00 am
In response to Theodis
Theodis wrote:
Players(well at least me) tend to be more annoyed by when their character fails to do something more than when an enemy succedes. By removing random rolls at least on their side can reduce a lot of frustration which is good.

I'm annoyed most by bad luck in random factors that will have the most impact: randomly generated loot, randomly generated party members, etc. If I have a party of 4-6 characters and one of them dies because they failed a defensive maneuver that had an 80% chance to work, that's probably a minor setback in most games.

I think you have a typo. You accidentally wrote "all" in the first sentence instead of "the one or two most exceptional, outstanding RTS games ever".

Well the only RTS I can think of off hand that involves random success is Warcraft 3 but that's only for certain special abilities and mostly everything else is based on a non-random algorithm(actually I can't remember but the damage ranges might have been somewhat random).

Oh, I agree that most RTS games tend to have a pretty small random factor. That doesn't stop the majority of them from being fairly simplistic and tactically repetitive.

Difficulty in a well designed game is always random, unless the game is either absolutely unwinnable or absolutely unloseable; you don't know all the player's capabilities when you design the game, and the player doesn't know all your design tendencies when they play the game. There's very little practical difference between a quantity which is random and one which is unknown (and unknowable).

Just go and play any Nintendo made game :). They are the masters of building well paced games.

But they still don't make perfectly paced games, or anywhere close to it. In fact, really it's not the pace so much as it is the range; Nintendo's games tend to feature a great variety in difficulty, so that both novice and expert players will find some situations somewhere to provide suitable challenge.

By extension, you seem to think that in a random game the point is to repeat the same process over and over again until it works. In either a poorly designed game or a learning scenario this may be the case. But if that's as complex as the game gets non-random numbers won't make it any more fun.

That's part of the point :P. There are many RPGs that don't have much as far as combat goes and just try to obscure the fact with random numbers. Console RPGs suffer much less from the reload syndrome because combat is much more deterministic compared to a game based on D&D rules like Balder's Gate.

Yes, combat is much more deterministic in a lot of console RPGs, at least in newer generations. It works something like this: if you press a lot of buttons, you generally win.

Random failures (at least of the variety that induce repeated reloads) are Mother Nature's way of telling you that your strategy, while interesting, needs improving.

There are cases when your stratagies are perfect and you still fail because of bad luck. This is just randomly punishing the player.

If a strategy can fail, it's not a perfect strategy, which is a good thing because allowing for a perfect strategy is bad game design.

If your strategy goes from flawless victory to total disaster over one bad roll, it's not quite so awesome and clever, now is it? Having to reload because of a bad roll is no fun, but I have no problem with having to adjust my tactics on the spot because of bad rolls.

One or two bad rolls is bad but if you're like me you run into situations where you get them constantly throughout the fight :P. Atleast my magic missiles never fail(except for lousy spell resist)!

While we're discussing the relative merits of random rolls... how much fun is it to have to reload because the encounter generator came up with a group of monsters that you can neither escape from nor kill since the game is deterministic and you don't have the resources necessary to beat them?

In a deterministic game if I can avoid it this is better since I know I can't win and won't be tempted to just get through with constant reloads.

So, to sum up: your complaint against random games is that you lack willpower?

How about replaying entire dungeon levels because you got crappy useless treasure?

This is just bad game design. In most roguelikes the difficulty of the dungeon is based on how much good stuff there is so if it's a nasty map you're pretty much garunteed some good loot.

Good loot doesn't automatically equal useful loot. Nor does powerful monster equal difficult monster, except in very simple games. If a deterministic game has the complexity to call for a wide variety of tactics and the tools to implement them, then it leaves open plenty of possibilities for mismatches between the challenges the player's encountering, and the tools they're equipped with the meet those challenges.
Nov 26 2003, 11:45 am
In response to Theodis
But in a well done deterministic system in the long run it will turn out the same numbers as a random system which is my point. The only difference is on the short term there is no luck good or bad.

If you just reduce damage by the percent to miss and make it always hit it'll have the same effect over a large number of attacks, but on the short term it's easier to develop plans of attack.

But that's at the cost of realism. No one can always hit the other target unless the technology is utterly outstanding. =)


So I don't think either is better or worse at trying to simulate an abstract situation it's just that the deterministic way will remove the inconviences that rise from a large number of random rolls. Console RPGs have delt with the problem by reducing the disparitiy of the rolls essentially removeing random loss since the worst case scenarios aren't much different than the best.

However, console RPGs involve more strategy than roleplaying. You rarely control only one character at a time, and most of the characters' personalities are already prescripted. As I mentioned, determinism is great for strategy games, because wide-scale strategies aren't based on chance, but deployment -- everything is abstracted to the simplest set of numbers to govern situations, such that a troop of heavy infantry will always defeat a troop of light infantry on open ground, etc. It's not so great for narrow-scale strategies because the tactics you develop are based on the resources you have on hand, and none of those tactics can be guaranteed successful.


Instead of wasting years and years of programming effort to model every possible aspect of the world, everything must be abstracted at one point or another. Determinism is not the solution because the more variables that are added,

Not neccesarily you can easily keep in percent chance to hit but instead use it as reducing damage done. This doesn't add anything new, has the same effect in the long run, and makes things laregly less random. Instead of having damage ranges just take the average.

However, that eliminates one of the better parts of randomness -- unexpected success or failure. Anyone could fluke out and down a big bad beastie with a penknife if the monster was having an exceptionally bad day and the would-be victim was having an exceptionally good day. In a deterministic situation, said would-be victim is a will-be victim. No amount of tactics could save you if you took on that beastie with that penknife. Of course you'd probably lose, but that's the meat and drink of an RPG -- acting the way your character would actually act in such a situation, even if it's detrimental to you.


However, there are two points to be made; one, all randomness eventually boils down to tendency; and two, knowing the precise result of any action ahead of time takes out much of the thrill of it. Regarding the first, if I rolled a six-sided die 100 times, then I could reasonably be expected to achieve a result of around 350. Each random roll would be different, and the final result wouldn't be precisely 350, but the tendency is there nonetheless. It's masking the determinism, which forces the player to think less in terms of strategy and more in terms of tactics.

Yeah but does your character actually get 100 rolls? In the short run you can lose a lot faster if the demon gets a few good hits in a short amount of time preventing you from recovering. If the damage done by the demon is relativly the same each time the player has a better chance of determining the best time to recover. By removing random rolls the short term results won't vary as much if any from the long term results.

Regarding the second, sports, games, and gambling are based on the thrill of the unknown. If I knew that I would win my next Poker hand, I'd bet, and if I knew I'd lose, I'd pass. It just wouldn't be a game, because it would eliminate all of the worry out of it.

A good gambler knows what his chances are of winning and how to effectivly increase them.

But the key point here is that the result may in fact be physically preordained, but the player himself has no idea. Wasting memory on preordainment is pretty irrelevant if no one would ever have access to that information. Transforming it into an abstraction takes randomness.


Determinism is the same way -- it's extremely practical for strategic-scale, because all randomness eventually boils down to a tendency, but it's incredibly impractical for tactical-scale, which is what almost all RPGs are set in.

It just hasn't been done yet :). If I ever get around to making this type of game I'll make sure to make it deterministic to see what happens.

*shrugs* Your prerogative. ;-)
Nov 26 2003, 1:20 pm
In response to Spuzzum
But that's at the cost of realism. No one can always hit the other target unless the technology is utterly outstanding. =)

The point of using random numbers in the first place for damage and hit probability was an abstraction of what was happining. By removing random probability by just factoring chance to hit as damage reduction is just another layer of abstration to remove the game play issue that arrises from using random numbers. Causing that much damage every attack doesn't mean the player is actually hitting every time anymore than a range of damage successfully implies and simulates exact situations. If you always hit that would simply be represented by causing the max damage every round. If you only hit every other time that means you would only do that much damage after 2 attacks which is how long you would have to wait randomly if you got the statistical average of the random rolls. By making it deterministic you're not implying anything unless that's what you convey to the player.

Another sleezy trick you can do is like what they did in Jagged Alliance. Whenever you saved in that game it also saved a certain number of future random numbers and the seed. So if you reloaded and tried the exact same thing you ended up with the same rolls. So if you had bad luck you were doomed to it until you figured out when the random numbers were used then you can shift the bad numbers onto the enemy.

However, console RPGs involve more strategy than roleplaying.

As far as I'm concerned I haven't experienced any real roleplaying other than when I did it off the computer. Roleplaying has nothing to do with random numbers or game mechanics which is why it's so hard to simulate without a human dictating what happpens rather than a computer.

You rarely control only one character at a time, and most of the characters' personalities are already prescripted. As I mentioned, determinism is great for strategy games, because wide-scale strategies aren't based on chance, but deployment -- everything is abstracted to the simplest set of numbers to govern situations, such that a troop of heavy infantry will always defeat a troop of light infantry on open ground, etc. It's not so great for narrow-scale strategies because the tactics you develop are based on the resources you have on hand, and none of those tactics can be guaranteed successful.

Why should this be any differnt at a tactical level? At a tactical level you should have more complexities than at a strategic level but random numbers help/detract as much at a tactical as they do at a strategic level.

However, that eliminates one of the better parts of randomness -- unexpected success or failure. Anyone could fluke out and down a big bad beastie with a penknife if the monster was having an exceptionally bad day and the would-be victim was having an exceptionally good day.

How is this a good thing? It just makes success/failure more of a random nature rather than one that's from well though out plans.

In a deterministic situation, said would-be victim is a will-be victim. No amount of tactics could save you if you took on that beastie with that penknife.

No but in a good system you could keep yourself alive until external factors help out your situation. In something like a console RPG fight where your resources are limited to what you enter combat with this would make for a bad situation that would inevitably lead to a reload.

Of course you'd probably lose, but that's the meat and drink of an RPG -- acting the way your character would actually act in such a situation, even if it's detrimental to you.

Yep but this is something that you can't really do in a computer game. But neverthless is something that could happen regardless of the combat rules.

But the key point here is that the result may in fact be physically preordained, but the player himself has no idea. Wasting memory on preordainment is pretty irrelevant if no one would ever have access to that information. Transforming it into an abstraction takes randomness.

Yeah but in many cases you can further abstract the randomness into a deterministic system that has the same results in the short and long term. For a gambling game this would make no sense because it relys on you losing in the long term, so making the short term also like this would garuntee the player always loses.

*shrugs* Your prerogative. ;-)

heh And you got after me for adding extra letters to my words :). Though this would seem like a typo since it doesn't sound anything close to what it should.

Anyway unless I get my phone problem resolved I probably won't be able to post until next monday when I'm in class :P. So if you don't hear my reply until then that's why.
Nov 26 2003, 1:46 pm
In response to Leftley
I'm annoyed most by bad luck in random factors that will have the most impact: randomly generated loot, randomly generated party members, etc. If I have a party of 4-6 characters and one of them dies because they failed a defensive maneuver that had an 80% chance to work, that's probably a minor setback in most games.

But if it happens frequent enough combined with not being able to save often enough or loading takes a while this can lead to not wanting to play the game.

Oh, I agree that most RTS games tend to have a pretty small random factor. That doesn't stop the majority of them from being fairly simplistic and tactically repetitive.

The only random factors I can think of is the cool down delay of weapons. This is very minor and is only used for visual purposes and chances are don't affect anything unless a fight is between equal amounts of forces. It's tactically simple because most stratagy games focus at a strategic level rather than a tactical one :). Though Blizzard tends to focus more on the tactical aspect than most other companies that do RTSs which I why I generally don't like then as much as other RTSs.

But they still don't make perfectly paced games, or anywhere close to it. In fact, really it's not the pace so much as it is the range; Nintendo's games tend to feature a great variety in difficulty, so that both novice and expert players will find some situations somewhere to provide suitable challenge.

Uh maybe we are thinking of two different things. I'm refering to the difficulty pacing which Nintendo does extremtly well going from easy to hard at a good rate that lets you learn as you play. This is something many other companies botch and either have the game easy the whole way through or difficult the whole way through.

Yes, combat is much more deterministic in a lot of console RPGs, at least in newer generations. It works something like this: if you press a lot of buttons, you generally win.

This has nothing to do with random numbers just with how simple the combat models are. And occasionally they do a decent combat model but they make everything so easy that if you use the worst possible attack constantly you can still come out victorious only taking a bit longer.

If a strategy can fail, it's not a perfect strategy, which is a good thing because allowing for a perfect strategy is bad game design.

Something that is perfect has no room for improvments. So you can have a perfect stratagy that fails if it is impossible to win which can be the case if the random number generator decides to hate you(if the game design allows for such wide range of random occurances).

In a deterministic game if I can avoid it this is better since I know I can't win and won't be tempted to just get through with constant reloads.

So, to sum up: your complaint against random games is that you lack willpower?

Well in most cases difficult games are designed such that you have little to no chance of winning if you don't get the most powerful equipment. Getting the most powerful equipment usually involves fighting powerful monsters who you can only win through good tactics and hoping the random number generator doesn't decide to hate you. This in itself is a bad way to make a game difficult but is done anyway.

Take for instance in the current game I've been playing somewhat off and on "Temple of Elemental Evil". There is fight near the end with a Balor. For those that don't know a Balor is one of the more powerful types of Balrogs. To win this fight without cheesy tactics that involve exploiting the AI you need one of the weapons that always hit or have a 10th level character with enough enchantments to hit something with an AC of 38. So the main "challange" is either having a weapon that always hits or having a well enchanted fighter roll extremly well 18+ on a 20 sided die. But the game is quite good and there are more challanges to overcome than just being able to hit the beast but to do so involves cheap things like having to know about the special weapons or have a real good understanding of the 3.5 edition ruleset of AD&D. That and I don't think any DM would be sadistic enough to have a 10th level party face a greater demon :P. I can only imagine how much it would suck to try and win that fight in ironman mode where you lose and must start over if your party dies off.

Good loot doesn't automatically equal useful loot. Nor does powerful monster equal difficult monster, except in very simple games. If a deterministic game has the complexity to call for a wide variety of tactics and the tools to implement them, then it leaves open plenty of possibilities for mismatches between the challenges the player's encountering, and the tools they're equipped with the meet those challenges.

Yep it would be one of the larger challanges of effetivly beuing the generator! Not that it can't be done but I'm sure I'd just use a simple solution and just apply hacks after some testing :P.
Nov 26 2003, 6:37 pm
In response to Theodis
I like some degree of randomness to my rpg battles. Obviously, huge amounts are not fun. If the weakest foe has a chance to kill you at the highest level its just going to get annoying. However, most rpgs aren't like that.
Combat systems in rpgs are, by and large, extremely low on user interactivity. A first person shooter requires aim and strafing skill and good mouse control. Sidescrollers often require distance judgement and timing. Those games where you fly up the screen in a ship shooting other ships requires you to identify safe paths on the screen and move into them. Rpgs, however, often require you to select your chosen attack, be it a spell or a sword, repeatedly. Are there exceptions? Of course, there always are, but most games don't deviate too far. If the combat was testing some skill of yours, like First-Person Shooters or Sidescrollers or spaceship games than random vs. deterministic wouldn't be much of an issue. You succeed at aiming at the foe faster than he aims at you or you don't. You're given a test and pass or fail it. While some aspects might be random, the principal control of combat is in your hands. You jump over the evil mushroom or you jump too late. It is based on your timing ability.
RPGS, on the other hand, don't usually test some skill of yours, other than maybe knowing when to drink that potion you are holding. The randomness creates some sort of skill to test: risk management. The player must ask themselves What do I serve to gain? What do I stand to lose? What is the chance I'll win? If the player sees that the risk isn't too great and the reward is good he'll go for it. If the reward is low and the danger is high he won't. Of course, a bad player might overestimate the worthwhile challenges and underestimate the tough ones and wind up underrewarded and killed alot. Hopefully as he plays the game more he'll become better at judging the risks in the game and act better in the situations he is faced with.
An example: I started playing a new MUD recently, as I often do. Combat was standard Diku fair: type kill mob and wait. Combat was fairly random, sometimes I'd hit sometimes I wouldn't. As I exploed the Ice Caves I found some foes I could dispath easily and would give me about 100 xp. As I got deeper I ran into an Ice Dragon. Killing it took great risk and much more time. I had to flee and heal a few times. And each time I fled there was a chance I would fail. When it was over I got a decent item and 300 xp. The risk and time was not proportionate to the reward. Therefore, I stuck to the easier foes. I judged the risk, decided it to not be worth it, and acted accordingly. Where I a higher level the dragon would have been easier to dispatch and the 300xp would have been a fair reward. typing "kill dragon" was not a challenge or a test of my skill, acting with the knowledge I learned from that encounter was.
If that situation was done in a purly deterministic system it would have been much less interesting. It would have taken me, say 20 attacks to kill the dragon and 10 for him to kill me. I would fail every time. The other foes might kill me in 15 hits and I'd kill them in 5. I would succeed every time. I would, therefore, kill foe after foe until I was ready to progress. Once I killed one enemy I am fully capable of killing all like them. It wouldn't be as fun.
If all combat consists of is typing or clicking attack repeatedly, then the randomness is the only thing keeping it from being boring. If you have a system that is more involving of the player and testing of his talents than absolutely, do away with randomness. However, with a system that doesn't actively test something of the player than the risk management aspect is tested instead. Of course, this assumes that the player is given options. If the player must kill a foe to advance, then the risk managament is negated because he has no choice. On the other hand, if there is plenty to keep him occupied for now where he is and its not important that he defeat that boss yet, then he is given the choice: Take the risk with the boss and get the rewards now, or wait and get stronger, but get rewards then that don't mean as much, plus wait longer before accomplishing something.
I guess my point is that randomness in a relatively uninteractive system when the player is given choices creates some sort of challenge. Poker is largely a game of risk management and its fairly popular. If the combat is a measure of some skill of the players, including strategy or tactics, then the randomness may be superflous and if the game is well designed might not be needed. However, for rpgs with the standard "choose to stab or cast spell" system of combat the randomness is good.
Nov 26 2003, 7:13 pm
In response to Theodis
Theodis wrote:
*shrugs* Your prerogative. ;-)

heh And you got after me for adding extra letters to my words :). Though this would seem like a typo since it doesn't sound anything close to what it should.

*cough* =P
Nov 26 2003, 9:02 pm
In response to Theodis
Your example of the Balor in the Temple of Elemental Evil is actually a wonderful example of how the problem isn't actually a matter of random vs guaranteed but simply one of bad game design and poor in-game decisions... you list two ways to defeat the AC 38 fiend: a Magic Item of Always Hitting The Sucker, or a powerful warrior with freakishly good rolls.

Why is this a perfect example? Because we have one situation where you can't miss, but you also can't possibly hope to defeat the Big Bad without the necessary (and presumably rare and hard to get equipment), and we have one case where you're relying wholly on the luck of the dice. Is either situation really the best way, from a game design standpoint, to handle a major confrontation's difficulty level?

I personally can't stand rock-scissors-paper combat models, and I think if you step back and view determinism through that particular filter, you'll see why. You define "perfect" as something with no room for improvement... that's a strange idea of perfection... I mean, something perfection does have no room for improvement, but that doesn't mean that everything which lacks room for improvement is perfect. "Room for improvement" is a pretty handy definition of "potential." I think the term I'd use in place of "perfect" for the thing you're trying to describe is, "As good as you can get it."

You can have a battle plan that's "as good as you can get it" and still fail. Is it frustrating? Yes. Failure ought to be. Otherwise, what's the point of success? You rightly point out to Spuzzum that gamblers know how to maximize their odds, and indeed, you can view games like draw poker, hearts, and blackjack as combat sims where you choose an appropriate weapon/tactic for an appropriate situation, and there are definite strategies that you can employ, there's even such a thing as a "right" move and a "wrong" one in a card game. End result? Still random. Your choices help you make the most of the randomness, that's all, but wouldn't be much of a game if the outcome was fixed. Play tic-tac-toe if you wanna fixed outcome. Or, since you're the one designing the game, why not just replace combat with a rousing round of "I Win!". You just need one command: "Win."

Yes, it sucks to have your fabulous hero fight his or her way to the 29th level of the dungeon and then fight back out only to be killed at the entrance by a half-dead goblin, but that's the way the cookie crumbles. Here's a relevant quote from the book of Buffy, 5:7...

Spike: "But you can kill a hundred, a thousand, a thousand thousand, and the armies of hell besides, and all we need is for one of us, just one, sooner or later to have the thing we're all hoping for."
Buffy: "And that would be what?"
Spike: (whispering) "One... good... day."
Nov 27 2003, 5:54 pm
In response to Theodis
Off-topic:

Another sleezy trick you can do is like what they did in Jagged Alliance. Whenever you saved in that game it also saved a certain number of future random numbers and the seed. So if you reloaded and tried the exact same thing you ended up with the same rolls. So if you had bad luck you were doomed to it until you figured out when the random numbers were used then you can shift the bad numbers onto the enemy.

I've actually saved a few JA2 games moments before I performed an outstanding kill shot. One file which I still have is where a guard sneaks around a corner only to be greated by my sniper. The bullet passes straight through the enemy's head and then ricochets off of a rock, coming dangerously close to another one of my troops.

I've considered saving random numbers in much the same way in some of my games.

(Incidentally, JA2 doesn't save a seed -- it just saves a set of ~50 (64?) numbers. After those numbers pass, the rolls are always different, even if you perform the exact same actions.)


On topic:

Why should this be any differnt at a tactical level? At a tactical level you should have more complexities than at a strategic level but random numbers help/detract as much at a tactical as they do at a strategic level.

However, it all boils down to tendency -- on a small scale, it is impossible to predict the exact course of any given action. Determinism makes it possible, preventing the suspension of disbelief. On a large scale, however, the results of most actions become highly predictable because all randomness leads to tendency. In such situations, determinism makes things a lot more believable.

It's possible (however remote) for a lowly goblin to somehow sneak under the radar and deliver a mortal wound to an elite swordsman. However, in a large scale battle, there's no way an army of these swordsmen could ever expect to take more than minimal casualties from an army of these goblins.

On a small scale, randomness makes sense -- on a large scale, not so much.


However, that eliminates one of the better parts of randomness -- unexpected success or failure. Anyone could fluke out and down a big bad beastie with a penknife if the monster was having an exceptionally bad day and the would-be victim was having an exceptionally good day.

How is this a good thing? It just makes success/failure more of a random nature rather than one that's from well though out plans.

Those sentences can't really be separated from the point below. ;-)


In a deterministic situation, said would-be victim is a will-be victim. No amount of tactics could save you if you took on that beastie with that penknife.

No but in a good system you could keep yourself alive until external factors help out your situation. In something like a console RPG fight where your resources are limited to what you enter combat with this would make for a bad situation that would inevitably lead to a reload.

That hypothetical example is assuming that combat is inevitable and avoiding the enemy isn't an option. It's a fight to the death. In a random system, the person has a glimmer of hope, however small, making the fight seem satisfying even if it results in failure -- in a deterministic system, the person has pretty much already lost, and there's no fun in knowing that you're dead before the first blow is even struck.


Of course you'd probably lose, but that's the meat and drink of an RPG -- acting the way your character would actually act in such a situation, even if it's detrimental to you.

Yep but this is something that you can't really do in a computer game. But neverthless is something that could happen regardless of the combat rules.

Says you. =) I personally think online roleplaying games are entirely feasible -- it just takes the right mindset of player, and a staff willing to educate people on the difference between roleplaying and metagaming. (For our spectators: roleplaying = doing things your character would do; metagaming = making your character do things that you would do.)


*shrugs* Your prerogative. ;-)

heh And you got after me for adding extra letters to my words :). Though this would seem like a typo since it doesn't sound anything close to what it should.

The word really is prerogative. Everyone always pronounces it incorrectly. =)

[edit]Just noticed Crispy already linked to it. Heh.[/edit]
Page: 1 2