Randomness in RPGs

[CrazyCrab]

Hi everyone,

recently a friend of mine made me play Fallout 2 and as much as I'm having fun, one element is somewhat ruining the experience for me. I know that it's a characteristic of most old western RPGs, but I just cant get over it.

It's how random everything is.

One turn I can shoot the enemy for 65 dmg using my SMG. The next turn I hit for 12. An enemy has +- 80 hp at this point.

At the beginning of the game I had 31% chance to hit a scorpion with a spear. I missed 11 times in a row and almost died during the tutorial.

With a high Stealing skill, I can pickpocket small items really easily and I almost never fail. The one time I failed I was machine gunned for something like 65 damage and given that I have like 50 hp this meant an instant reload and a couple of minutes wasted.

Those things annoy me. I can understand when there is some chance, it makes the game a but more exciting I suppose, but when you miss 11 times in a row you know that there's something wrong.

At the same time the friend that recommended me the game absolutely loves it. He also loves Warhammer though, so he really doesn't mind randomness in general. When we talk about it he says that without randomness the game is too predictable and just boring.

What do you think? Should RPGs be this random? Is there a ''limit'' on how random it should be, for ex. by giving you +accuracy whenever you miss? Should it always be a clear number and just be reduced when there are things that make you miss etc?

Thanks! 

 

[Razzazaki]

I love fallout 1 and 2 and i love the randomness in it

I would love to have the same turn based system in a rpg maker game

 

[Meoix]

Randomness in games I feel is an interesting and important aspect to increase realism.

In a way it fills in the blanks for hidden variables that would be too complex or annoying to consider implementing into the game.

Example, say in Fallout 2, you attack somebody with a spear and even with high agility you miss because of the random factor.
Well to make the game more interesting or realistic you could argue that maybe your character missed because a rock gave way underfoot and your character lost balance hence missing the target. In real life the perfect soldier/warrior/martial artist can still lose to a novice under the right conditions. Randomness creates this illusion.

That all said, I feel that simply using a uniform random number generator isn't ideal I believe (if you want a random number 1-10 then each number has an equal chance of being selected).

I would use a non-uniform random generator using exponential distribution. That's a whole topic in of itself so feel free to google it if you want to know more about it, but essentially
with a non-uniform random number generator, numbers nearer to, for example say, '5' would be exponentially more likely then numbers further away. 
 

Non uniform randomness creates a 'soft-cap' on the randomness so you're far less likely to get say that 11 misses in a row.

Btw, Fallout 1&2 were amazing I definitely recommend sticking it through and maybe even try Fallout Tactics: Brotherhood of Steel if you also have it (commonly found in bundles with FO1&2)

 

[omen613]

I was gonna make a post about randomness myself but looks like you beat me too it lol.

I haven't played Fall out 2 but i have played both fall out 3 games and if I recall correctly the stats you chose in the beginning of the game really factor in those % rates on skills like throwing a spear (Str I think it was?) because my first build was to be a sniper but I put the starter stats wrong and I couldn't hit anything with a sniper rifle...restarted the game..allocated the right stats I looked up online and BAM I was an awesome sniper for the rest of the game.

-----------------

As for randomness in RPGs in general if your not gonna factor chance on stats like the above paragraph then the risk has to match the reward. Doing a basic attack like throwing a spear to do damage should have a very high success rate like at least 75% otherwise your just gonna make players turn off your game lol.

Example A: I have a Blind spell that works 30% of the time....i'm probably never gonna use that spell (FF vets could remember this)

Example B: I have a skill that I use often that has a 25% to poison the foe. When it procs I'm happy about it because I picked that skill to do damage and not just for it's poison move. Now say I have a skill that does double damage if the foe is poison that I don't use on a regular bases now i'll use it because of something random that happened.

Example C: Hitting this slime monster has a 25% of causing a random status effect on me for hitting it. I know not to use multi-hitting attacks on it because my chance of getting a status effect multiplies and If I do happen to get the status effect anyways it's just an obstacle to overcome not a turn off the game kind of frustration.

-------------------

So I like randomness in the battles for I think it makes battles and actions a little more unique then the last battle.

 

[Omnimental]

I prefer predictability with pockets of skill to modify the outcome.  Timed button presses or input combos to increase damage or add effects.  I don't mind certain skills having random chances to hit or effect, but I don't want to be fighting a boss and have a streak of bad luck cause me to lose when I did everything perfectly.

 

[C-C-C-Cashmere]

I think randomness is OK to have in a game, but it just matters about how fun the experience is in the end. If, like with your experience in Fallout 2, you find the randomness is too arbitrary and thus frustrating instead of fun, then maybe it isn't a good design. I've never played it, so I wouldn't know. But I do have an idea for an RPG where everything is deterministic, that is, nothing is random. Imagine dealing the very same damage every single time according to a static formula. Then you can predict how battles will play out like a puzzle. I'm not sure if many RPGs have tried that out, but I think it'd be a cool thing to explore. Imagine if you encountered a battle every 10 steps, no more, no less, and you could see this counter via the on-screen HUD. That would be a really interesting idea. But right now that's on my list of games I should make later, because I should finish the one I'm doing right now.

Anyway, I ranted. I just want to say that randomness is good if it's not too random so that your stats almost don't even matter. But as long as it's fun in the end, I think it doesn't matter how you implement randomness. If the end result is that it's an enjoyable experience, it doesn't really matter how unfair the combat is, as long as it's a good experience.

 

[Harmill]

I recently wrote up 3 blog posts on this site about RNG. If you've played Disgaea 4, Ni No Kuni, Pokemon, or Suikoden, perhaps check them out for a detailed analysis of how these 4 games used RNG in some of their mechanics. But here, I'll be more general.

I have not played Fallout so I cannot say to how their systems work, but missing 11 times in a row (even if it's a hyperbole, missing 3 times in a row SUCKS) or having a damage range of 12-65 sounds horrible.

To me, RNG NEEDS to be there for a reason - and for it to provide meaningful decisions for the player. Games that give you 80% Accuracy and then give enemies 30% Evasion are doing it wrong.

1. It feels horrible to miss twice in a row. If this "miss streak" becomes a common occurrence, I am less likely to partake in the mechanic (so if it's something as core as Hit Rate in an RPG, that's a BAD thing).

2. Accuracy and Evasion should have a relationship in which the player KNOWS what his chances are going into it. Games that either tell you what your chance of hitting are, or games that have direct stats that influence Accuracy and Evasion at least give the player a way to INTERACT with the randomness.

RNG needs to have a PURPOSE and in most cases, the player should have a way to manipulate them. Give your Status Ailment spells 100% accuracy, and THEN give certain enemies varying resistances. That way you should always hit enemies without any resistance to it. Or, go the Pokemon route, and remove an evasion factor and tie it all into Accuracy. The player knows with EVERY use of Sleep Powder that it only has a 75% success rate. There is randomness, but the player is completely aware of it, and so it doesn't feel like you're being "cheated" if you happen to miss twice in a row. I've had Tackle (back when it had 95 Accuracy) miss 3 times in a row. It sucks, but it's hard to blame the game in this case. It's not hard to blame the game if you miss 3 times in a row due to 80% Accuracy vs 30% Evasion scenarios. The main takeaway of Pokemon's implementation, is that Accuracy is a HUGE factor in how players make turn by turn decisions on which moves they are going to use. Many RPGs, especially the ones made in RPG Maker, hardly emphasize Accuracy and yet set random Accuracy and Evasion stats to their party members and enemies with hardly an afterthought.

You need to consider the consequences of failing an RNG check. Did you use a Normal Attack and miss? It slows the gameplay down, AND it provides the enemy with another chance to kill the player. If the player gets a Game Over because of a miss, that's going to be a slap in the face. Are you partaking in a Harvesting system where you find plants in the field and try to harvest them? Well, if they grow back over time, it's not a big deal if you fail an RNG check trying to harvest them. Did you just defeat a one-time boss who had a 30% chance to drop a unique item that unlocks a bonus dungeon? Did you "fail" the RNG check and not get that item when you defeated the boss? Now you're locked out of an entire bonus dungeon! OK, that last one is not a likely occurrence in an RPG, but it's meant to highlight the potential for unnecessarily bad consequences for failing an RNG check. The point is, RNG can be more forgiving if the consequences don't cause frustration to the player. I'm not likely to get frustrated if I fail harvesting two plants in a row, but I AM likely to get frustrated if I miss an enemy twice in a row. And you can be sure that I'd be livid if I defeat that difficult boss and it doesn't drop that unique item for the bonus dungeon. It's a waste of time because I'm just going to reload my save and fight the boss again and again until I get that unique item (of course, that depends on how important accessing that bonus dungeon is to you).

Anytime you want to add a mechanic influenced by RNG into your game, you should really think hard about WHY it is influenced by RNG and whether it's only going to frustrate the player more than it's going to help or enhance your game. Bad implementation of RNG is one of the biggest reasons I'm going to shut off a game because of how frustrating it is.

 

[Onomotopoeia]

I can respect the need for some randomizer; but if the random variance in your example is too great, it breaks the "feel" of the gameplay.

I would start with a base value before randomization, and add or subtract a randomized value that is no greater than ten percent of the base value. Does that sound like a decent principle?

 

[hian]

I don't really like randomness at all. Most RPGs provide the player with statistical numbers to keep track of the likelihood of success in a given mechanic, and as such, players like me, come to expect that the game will behave accordingly. If there is a large margin of error to that variable despite the number the game provides me and tells me to expect, then that becomes a problem.

I remember playing the original Final Fantasy Tactics on my emulator a few years back, and I was annoyed to no end by the seeming randomness of that game.

The game would tell me I had 79% hit chance for an attack, and I would miss over and over again. Finally, reaching a peak in my frustration, I decided to start using save/load states to actually check how well the numbers added up.

I thought that if I had a 79% hit rate given before an attack, that I should expect to be hitting about 8 times out of 10, yet, in using the save/load state function to reset whenever I missed, I found that I missed at a ratio of about 7/10 times.

That's ridiculous, and too much randomness for my taste. It completely ruined the experience, and I simply ended up clearing the game resetting whenever I made a miss in a critical situation.

Does randomness have a place in games? Well, it can, if it's to be expected and the player has an incentive to accept the random aspect of the game-play (fluctuating drop-rates, or rare encounters, in games about collecting resources can be examples of that).

If the game delivers you a formula for how things are supposed to work, and provide you numbers that you're supposed to base your play-style on, only to break that expectation with large margins of random errors, then it simply causes frustration.

That shouldn't happen at all in my opinion.

 

[C-C-C-Cashmere]

Well, maybe randomness is a way of creating realism because reality seems so random. Lol.

 

[hian]

Well, maybe randomness is a way of creating realism because reality seems so random. Lol.

Well, that can be true to an extent. I think game designers should be wary of realism for realism's sake though.

Apparent randomness is probably one aspect of reality that we don't generally want our games to emulate.

Another thing that happens seemingly at random (although it really isn't), is natural disasters killing off important people. That being said, you wouldn't want to have your game end randomly by the entire party being killed off by a tsunami as they visit a town by the sea, for no apparent reason.

 

[C-C-C-Cashmere]

That being said, you wouldn't want to have your game end randomly by the entire party being killed off by a tsunami as they visit a town by the sea, for no apparent reason.

For some reason, I really want to play a game where this happens now.

 

[hian]

For some reason, I really want to play a game where this happens now.

lol, well it could make for a fun oddity in terms of story-telling though. When you think about it, it really is quite silly how almost every single story ever written is this statistically super-unlikely series of events where everything is tied together at the end, and no people ever get struck by random accidents etc that don't inevitably end up serving the greater direction of the plot.

 

[whitesphere]

In my current game, I use randomness to balance out certain very powerful abilities.  In particular:

- Bow and arrow.  Without scripting, they have infinite ammo.  My solution: Bows and arrow miss 10% of the time, which balances out the extra damage potential.  

- To AVERT the "Useless Useful Spell" trope.  This is the "Damn, why can't I use Instant Death on the Final Boss?"  More powerful opponents have a reduced (in some cases significantly) chance for the spell to stick.  But there is always the remote possibility that, maybe just maybe the Final Boss MIGHT fall prey to the Instant Death spell.

- To irritate players less by having ultra-powerful boss attacks and status effects NOT always connect.   To compensate, when they DO connect, they are more painful.

Basically, I think randomness is an excellent way to balance out ultra-powerful, otherwise game breaking abilities.   This gives the player a risk/reward choice to make.  Use the powerful, but sometimes missing weapon?  Or use the less powerful, but reliable one?  Playtesting will ensure the damage output averages to be close.

I always ensure the spell or skill which has a good chance of failure specifically mention this in its description.

Randomness also allows a spell or attack to be more powerful by having a much larger random range.  Again, it becomes a risk/reward choice for the player.  And more combat options can be a very good thing.

That being said, I don't like it if something is so random the player can't count on it at all.  But as you see, a degree of randomness is a great asset to game development, at least for me.

As others have said, the random number stands in for a lot of details the player doesn't care or need to know, so the game flows smoothly without requiring the player to micro-manage every little thing.  

 

[Ralpf]

I don't really like randomness at all. Most RPGs provide the player with statistical numbers to keep track of the likelihood of success in a given mechanic, and as such, players like me, come to expect that the game will behave accordingly. If there is a large margin of error to that variable despite the number the game provides me and tells me to expect, then that becomes a problem.

I remember playing the original Final Fantasy Tactics on my emulator a few years back, and I was annoyed to no end by the seeming randomness of that game.

The game would tell me I had 79% hit chance for an attack, and I would miss over and over again. Finally, reaching a peak in my frustration, I decided to start using save/load states to actually check how well the numbers added up.

I thought that if I had a 79% hit rate given before an attack, that I should expect to be hitting about 8 times out of 10, yet, in using the save/load state function to reset whenever I missed, I found that I missed at a ratio of about 7/10 times.

That's ridiculous, and too much randomness for my taste. It completely ruined the experience, and I simply ended up clearing the game resetting whenever I made a miss in a critical situation.

Does randomness have a place in games? Well, it can, if it's to be expected and the player has an incentive to accept the random aspect of the game-play (fluctuating drop-rates, or rare encounters, in games about collecting resources can be examples of that).

If the game delivers you a formula for how things are supposed to work, and provide you numbers that you're supposed to base your play-style on, only to break that expectation with large margins of random errors, then it simply causes frustration.

That shouldn't happen at all in my opinion.

That's not really a fair test, because it depends on how they set up their RNG, it sounds like they have at least part of it locked (meaning a reload doesn't change that part, I know I'm not using the right terminology), not to mention you used save states (missed that the first read) which would definitely lock parts of it in (I'm surprised it wasn't a guaranteed miss, using save states). I got annoyed at seemingly random misses in FFT, too, but it is ultimately fair....I have had plenty of experience with it, played through probably 10 times.

 

[hian]

That's not really a fair test, because it depends on how they set up their RNG, it sounds like they have at least part of it locked (meaning a reload doesn't change that part, I know I'm not using the right terminology), not to mention you used save states (missed that the first read) which would definitely lock parts of it in (I'm surprised it wasn't a guaranteed miss, using save states). I got annoyed at seemingly random misses in FFT, too, but it is ultimately fair....I have had plenty of experience with it, played through probably 10 times.

It's a perfect;y fair test, because the number that's shown once you highlight the enemy for attack in FFT is supposed to be the final hit calculation. You're not, as the player, supposed to look at that stat and go "and now I have to detract the enemy's evasion points etc. and then I'll know what my hit-rate is".

Then hit rate in the box as you set up your attack is provided as the final calculation and meant as a tool for players to decide whether the attack will be worth it or not. If the game doesn't follow it's own design, that is a bad design decision. You might as well leave that stat out of the picture all together then, because it isn't consistent with what will actually happen most of the time in game.

And, the calculation isn't locked before command select.

If it was locked before, it wouldn't change at all after load, yet it does.

Besides, I already took that into account when I started testing. I've done save/load state tests from before I move my character at all, to up until after I've selected attacks.

The calculations are done pre-select. Some variables change based on where you stand. That's it.

Despite that, based on two play-throughs of the game, if you're hit rate isn't in the 90% range, you'll miss more often than not on your first attempt. That to me, is neither fair, nor good game design, because it means that more often than not, the death of your characters etc. will be based on statistical flukes because you gambled on a hit that didn't stick despite being above 70% but below 95%. That's just bad balancing.

Case in point, the PSP version seems to have more consistent hit-rates, so it would seem they actually went in and balanced in later version.

Don't get me wrong, I loved FFT - for it's class system, and it's story.

It had several issues pertaining to balancing though.

 

[TheoAllen]

Randomness is good. It adds plan B once your move is failed (I love plan B!)

However, even the game has randomness, it shouldn't have a chance to instantly killed the player. For a little example, you got a surprise attack and because enemies are too strong, your party knocked out in single turn and don't have a chance. That is the limit of the randomness shouldn't have.

 

[C-C-C-Cashmere]

Every single action should have a completely random outcome.

 

[Ralpf]

It's a perfect;y fair test, because the number that's shown once you highlight the enemy for attack in FFT is supposed to be the final hit calculation. You're not, as the player, supposed to look at that stat and go "and now I have to detract the enemy's evasion points etc. and then I'll know what my hit-rate is".

Then hit rate in the box as you set up your attack is provided as the final calculation and meant as a tool for players to decide whether the attack will be worth it or not. If the game doesn't follow it's own design, that is a bad design decision. You might as well leave that stat out of the picture all together then, because it isn't consistent with what will actually happen most of the time in game.

And, the calculation isn't locked before command select.

If it was locked before, it wouldn't change at all after load, yet it does.

Besides, I already took that into account when I started testing. I've done save/load state tests from before I move my character at all, to up until after I've selected attacks.

The calculations are done pre-select. Some variables change based on where you stand. That's it.

Despite that, based on two play-throughs of the game, if you're hit rate isn't in the 90% range, you'll miss more often than not on your first attempt. That to me, is neither fair, nor good game design, because it means that more often than not, the death of your characters etc. will be based on statistical flukes because you gambled on a hit that didn't stick despite being above 70% but below 95%. That's just bad balancing.

Case in point, the PSP version seems to have more consistent hit-rates, so it would seem they actually went in and balanced in later version.

Don't get me wrong, I loved FFT - for it's class system, and it's story.

It had several issues pertaining to balancing though.

Edit: Just wanted to say right at the beginning that after typing this out and thinking it through in the process, playing a game with a RNG through an emulator may mess with certain aspect of it, depending on how it is set up, which may account for your issue of hitting things with middle of the road percentages.

What you are describing sounds like they are doing it exactly right, and your save state test proved it...No game it going to tell you your hit rate after a (general predetermined) calculation, that would give the player way to much information, as you would generally see wildly different numbers (probably see 100% more often then not for any attack from the rear) then they want you to see.

Have you played any of the Civilization games? In the modern ones (can't remember if they did it before 4) there is an option for 'New Random Seed on Load' I believe is how they put it. What that does is change the seed on load (obviously), which essentially controls the random actions of the game, if that is on you can reload a save if you failed an action and you may have a chance at success, but if it is off if you reload and repeat the same action you should have the exact same results, the later is how most games handle it.

From your description of your test it sounds like FFT uses two or more seeds, one which is predetermined (only for that specific turn, most likely, or maybe for the turn and spaces on the grid) and one or more based on other factors (probably system clock, which may be screwed playing on an emulator), if the former is very low or very high depending on how their system is set up, you will probably see the same result every time, or close to it, but they are not going to tell you the percentage based on that number, they are going to tell you the percentage based on a neutral number (if it is 0-255 they probably base the percentage on 128, for example), because if they did tell you the percentage based on that number the game would be too easy. If you see 10% instead of 70% you are probably going to use an item, guard, use a 100% attack, or if no other options....get out of the way. If you see 100% when it would be showing 60% you would know to take the action, may as well not have any randomness in both of those cases. Actually both of those are probably wrong....it should be showing 0% or 100% if they are telling you after the calculations are done.

Again, probably botching terminology.

 

[hian]

I think you're failing to get the gist of my argument.

My argument is that in either case, that is bad game-play design.

Either the game is set up to give the player the illusion of having an over-view of failure/success rates, when they in fact don't, or the game's hit-rate equation has a much too large margin of error. Both are bad.

Just play the game again. When you move your character and set up an action, a box shows up, telling you the odds of success. This box shows the player what the player can expect based on the total amount of factors that go into the action.

It isn't showing you the base accuracy or something like that.

It shows you the percentage chance you have of hitting after adding character accuracy, equipment and buff effect, and the enemies evasion rates.

I know this, because the number changes based on equipment, level, where your character is standing, the opponents evasion rates, and buffs. You can easily check this, in game, in battles, by simply changing any of the above factors prior to selecting an enemy to attack.

If there is additional factors influencing it, then again, that's bad design. If the number clearly changes based on discernible factors that we can reasonably expect, then additional non-discernible factors only confound the player.

This is not "giving the player too much information". It's the minimum amount of information a game should give a player in order to allow a player to make informed decisions.

A game that doesn't provide you this, means that the player has to either leave things to chance, or always plan for the worst possible scenario at all times. Non of this is particularly enjoyable for most players, just as it wouldn't be funny to play sports or other competitive forms of play, if random unknown elements suddenly came into play and messed up your game.

If it was any other way, then that would be ****ty design choice. Why? Because the game can tweak difficulty without hiding the math that decides the margin of error behind an action. If a game tells me that I need certain gear and certain level, and certain stats to clear a challenge, then that's reasonable.

Losing in an RPG because of lacks like that is a given.

Losing in an RPG because you don't have a good strategy is a given.

Losing in an RPG because your entire party gets struck by "death" and dies despite "death" having the official success-rate of 65% is unreasonable and silly. Games should function around players making strategies based on factors that are known, not on factors that aren't.

Imagine playing football(or soccer if you're from the U.S), and then having random wind-gusts made by randomly moving wind-cannons along the field, blow the ball off course when you tried to kick it. Or, having the coach and judge tell you that the cannons will only interfere 65% of the time, only to find that they blast your ball away 9/10 times on average.

That would be completely inane. The only reason people defend randomness like this in games, is because they belong to a camp of players who're willing to grind, and considers grind an acceptable part of game-play, whether they realize this or not.

That's what it is essentially. Random elements that ruin accuracy etc, only do one thing - it takes the control out of the hand of the players, and threatens to set you back by having you lose, and thus having to spend time going back to the point where you previously lost, because you didn't have fate on your side.

You should lose games because your skill doesn't suffice, or because your characters aren't strong enough - not because the A.I threw out a random number that by all accounts should only make you lose 1/10 of the time. Especially so, if the game doesn't even tell you this in a reliable manner, or give you any clear cut way of making those calculations before taking an action.

As for emulator - you're basing your arguments on factors that are completely unknown to you, and ignoring the arguments that are made based on the factors that are known to us.

Knowing how emulators work, knowing how the original PSX works, and I can say with a pretty high level of certainty that there is nothing about the emulation that is screwing with the statistics. The PSX didn't have a system clock last time I checked (and even if it did, my emulator is running on the bios loaded my from my own native system, so it shouldn't matter anyway), and the time in FFT is a local functionality of the game itself. There is no reason to assume that emulation should do anything to the in-game timer that would mess with the statistics system. Furthermore, the idea that in-game clocks should have an effect on accuracy is inane too. Again, that would be poor design in my opinion. You don't take important game-play factors determined by a numerical system, and then link that to other numeric systems that have nothing to do with it for no apparent reason.

The fact of the matter, is that I retried several times, using save/load states, as I said several times now, taking into account various factors that might change the hit rate.

At the end of the day, there are only two possibilities :

1.) The random distributor, or variable range of any given action equation, of the game is extremely over-sensitive. Bad game-design.

2.) The box that gives the player the information of hit-rate prior to action doesn't reflect the actual hit-rate. Bad game design.

You're not getting around that. You might think that's okay, and that it's fun to play with a game that might kick you in the balls unexpectedly. I don't. I consider that flawed game design. If I wanted that, I'd be playing Yahtzee, not Playstation.