I made a comment in the Why rocks/papers/scissors doesn''t work thread about why people seem to be against the idea of introducing uncertainty in games. But as i got to thinking about it, I think this really deserves its own thread. I''ve noticed a large number (I think I may actually be the only proponent of this) of designers here feel that a random element in games (especailly RTS) will detract from them. I think the thought is that the player may lose a battle not due to his planning, but by sheer luck, and is therefore "robbed" of a victory. This in turn leads to the what I think is the real point, the loss of feeling of control. But this is exactly my whole point. Control is a fickle thing, and your ability to have control should be the paramount concern of a commander. As RTS stand today, they guarantee your control. But I think that the TRUE skill of a player, will be his ability to maximize the control of his units. What do I mean by this? I''ll repost what I put in the other thread: 1) Because games have fixed unit types, players will quickly discover their strengths and weaknesses. Even technology trees just make more powerful units that players will know the capabilities of. 2) Since there is little if any random element, if players know the makeup of the armed forces on both sides, players can calculate with precise certainty the outcome of a battle. 3) Since there are few if any external factors like morale, leadership, terrain or logistics (when have you seen a unit run out of ammo?) the player is once again free to calculate with near 100% certainty of the outcome 4) Players have absolute control of his units, know what his units know, and always have access to his units, thereby his units will carry out his plans without fail. In other words, strategy games have a formulaic determination of outcomes of battles. You are guaranteed to have certain outcomes, and you are guaranteed to always have control of your units. So where is the skill is my argument. The skill comes in making sure you have proper intelligence to make sure you know what forces your enemy has, maximize your resource potential, and to make sure you do indeed know that UnitD beats UnitE. Random factors and more importantly factors which do not guarantee points 1-4 above are I believe crucial to creating a game in which the skill lies not in what I mentioned above, but in maximizing your ability to control your units and their potential. When you have a game of this nature, you can create scenarios where the smaller though better trained or led troops can beat poorly led, motivated or trained units....even if they have the same quality weapons as your own forces. For example, Thebes defeated Sparta during one of their wars through superior leadership, not because the thebians had better units. Their organizational command had been revamped by Phillip of Macedonia...the father of Alexander the Great. And look at Alexander himself. Time and again he faced insurmountable odds to defeat his enemies. His foes were not little piddly piss-ant forces, but included the very powerful Persians and the Indians. It was only his troops incredible fatigue that stopped them from marching on to China. But how in the world can you simulate this with RTS as they are now? As I noted in the other thread, there are many examples of battles which relied on a little luck for the victory to go through. Midway was won almost entirely by luck...we found the Japanese first. The Hood was destroyed in one shot by a lucky hit to it''s ammunition bay. Custer (stupidly) decided not to bring along his gatling gun batteries along to Wounded knee. The American navy was spared because for some reason, neither Aircraft Carrier was at Pearl Harbor on Dec. 7th...even though they were supposed to be. Luck is a fact of life, and it does NOT diminish from strategical thinking. In fact I think it becomes another obstacle to be avoided and factored in. When you have uncertainty your thinking has to change, and I think it is a higher and more complex level of thinking than when it is deterministic. Players may feel that they are losing control, but instead of seeing it that way, it should be seen as a measure of skill to maximize your control. Essentially, games today guarantee you control, so there is no skill involved there. Knowing your units will obey your every command is not only unrealistic but is just too easy. Imagine a game where victory can be achieved by simply throwing your enemy''s forces into disarray and confusion. Where defeating the enemy''s units is not so much a matter of destroying them, but making them unable to act. It introduces another means of victory. Again, I will point out that I am far more realisitc than most, but I think this can benefit even casual gamers. So don''t look at uncertainty as something to be avoided, but just another gameplay element to be factored in.

I should start by saying that I don''t believe in randomness. I think everything is essentially deterministic. A game''s ''random'' factors are there to model the vast array of minor factors that would otherwise go unaccounted for.

At one extreme, we have no chance factor, and the game can be predictable and boring. At the other extreme, when the whole game is based around chance it is unrewarding for the players.

However, games like chess appear to have no random factor, and yet are still deeply engaging. This is because the ''chance'' occurs in the mind of the opponent. There are many valid ways that they can respond to a threat, and it takes a truly skilled opponent to be able to guess which response will come next. In many games, there are few ways to validly respond, and in systems that approach R/P/S, there are only 1 or 2 valid responses.

A random factor added to a largely deterministic system will add a bit of spice to the game, but essentially that is all. Instead of attacking when the odds are 51:49 in your favour, you may wait until they are 60:40 instead. The random factor just draws out the process. So although I am not against a small degree of randomness in a game, I think the true answer is to give players a range of possible responses that are all effective, rather than forcing them to choose from a limited selection.

[ MSVC Fixes | STL | SDL | Game AI | Sockets | C++ Faq Lite | Boost ]

hmmm... i think you are on to something here...
i never really liked RTS games, so i never thought about it too much... but after reading your post i realized that at least part of my dislike for them is because of the lack of randomness. i never got a thrill out of finding the ratio of unit-As to unit-Bs (or even looking them up on the internet at a fan site) just so i could know how many of each to "build". it would be much more fun to just find out that unit-As generally can beat unit-Bs, although sometimes for no apparent reason five unit-Bs can wipe out a dozen unit-As and still be quite healthy!
of course, this might annoy hard-core RTSers because they get joy from calculating the types and numbers of units to put in their armies.

quote: Original post by Kylotan
I should start by saying that I don''t believe in randomness. I think everything is essentially deterministic. A game''s ''random'' factors are there to model the vast array of minor factors that would otherwise go unaccounted for.

true, but to get a game to seem as random as real-life seems you''d have to do far more computations than a computer can handle.

quote: A random factor added to a largely deterministic system will add a bit of spice to the game, but essentially that is all. Instead of attacking when the odds are 51:49 in your favour, you may wait until they are 60:40 instead.

or you might attack with half your force (30:40) and hope you are lucky, while reserving the other half to defend the base and/or attack some other place (assuming that this isn''t one of those massive battles where both players throw every unit they have at each other until total death has occurred for everyone on one side).
but i''m not a hardcore RTSer...

There is nothing wrong with randomness, provided it does not dominate the gameplay. If a player can rely solely on luck to win a game (e.g, ludo) then the game is tedious for anyone older than the age of about 5. If the player can use strategies to influence the probabilities then you have a game that may be fun to play.

Okay, so it seems from Kylotan''s and Sandman''s replies that randomness is more a matter of degrees rather than total exclusion. And I agree. I think randomness should not be so overwhelming that it factors in more than actual actions, but as Kylotan said, it can definitely add spice to a game.

But I disagree with Kylotan in that there really isn''t chance, just the lack of forehand knowledge. Weather for example is a perfect example that can turn the tides of battle of which there is absolutely no control or perfect prediction. Officers misinterpreting orders is another. Or like I mentioned before, our Aircraft Carriers were supposed to have been docked at Pearl Harbor on Dec. 7th, but they weren''t. There is chance and random factors, and they can play a greater or lesser role in the outcomes of battles.

So now the trick becomes...how much is too much? At what point do you factor in random elements? If you start from the bottom up approach, the consequences of bad luck with one unit against one unit shouldn''t have too much of a consequence against the overall scheme of things. However, in a top down approach, having bad luck can be disastrous. Let me illustrate:

In the bottom up approach, you could have for example, a very powerful unit that unluckily only "grazed" a target, thereby allowing it to survive, and in turn the weaker unit gets a lucky shot in at a vulnerable spot. But while this may suck, it doesn''t affect the overall battleplans too much unless you REALLY counted on that powerful unit to proceed. But what if instead, an order that you wanted to send to an entire division was not received or worse, miscontrued? Then you have an entire division doing something that you did not intend them to do. Another example could be a leader dying, throwing your units into confusion.

So how far is too far? Should the randomness be limited to the bottom-up approach? Or should there be more significant random factors? How about weather for example? If your forces rely heavily on air superiority, bad weather can throw your forces to a serious disadvantage (look at the battle of the Bulge).

The simplest way to introduce some uncertainty is to allow weapons to miss, or to make critical hits on an occasional basis. The likelyhood of hitting with a weapon could increase with the experience and readiness of the firing unit.
This is the sort of model used by tabletop games, like Battletech for instance. But a model like this must be implemented very carefully to avoid causing frustration. It must be very clear what is going on. I mean, if I lose a unit to a lucky shot, I want to know how it happened, and how unlikely that lucky shot was. And if I''m pounding away at an enemy unit and it''s just not dying, I want to know how many of my shots are going wide. Without this transparency, the system just appears arbitrary, and it is difficult to learn any lessons from a seemingly random defeat.

It boils down to gameplay vs realism. While the idea of a misunderstood order may exite some players, others may find it an incredible hassle (and with instantaneous response will probably just issue the same order over again anyway. ) As a player, I would prefer the randomness to occur on the division level because its easier to notice and correct (or take advantage of.) But realistically you would probably have to use the moral or something so you can tell why your center line is running away or why the reserves won''t move. Randomness chooses when exactly, but the player should be able to know ''around when'' it might happen.

Please, please, please...

Everyone, go out and buy Shogun Total War. It''s not perfect (mainly interface issue), but it will shine such a new light on your view of RTS games, that you''ll bang your head in for not trying it out sooner.

Oh, and yes, you do run out of ammo (arrows).

Back to topic on randomness:

If we''re talking about warfare RTS games, the random factors should reflect the unexpected things that happen in warfare. Of course, you simply can’t account for everything, but arrows and bullets don’t always fly true.

It’s true, that if you have a 100 archers and each individual archer has a 20% chance of hitting a target, on average 20 targets will be hit by that group if they all shoot one arrow. You could just say ‘well, why make it random if on average they hit the same amount of targets anyway?’. True, but the chance of those 100 archers not hitting a single target, or all 100 hitting their supposed target, means that you shouldn’t rely too much on them, but also that you can put a little bit of hope into their actions. Say your 100 archers are attacked by 100 swordfighters and you know that in melee combat, your archers don’t stand a chance (read: will be wiped out). Say you know that your archers only stand a chance of winning if there are 100 of them and only 40 or less swordfighters. Now say you know that your archers will only be able to get two volleys of arrows off before the enemy closes. If you know that the skill of the archers and the armor of the swordfighters are such that about 20% of your archers will kill or disable a swordfighter per volley, you could reason that the first volley would kill 20 swordfighters, and the second would kill another 20, leaving 60 to fight your archers in melee. 60 is more than 40, so you figure your archers would be wiped out. At this point (for argument’s sake, let’s just say you have enough time to think all these things through) you decide to not even fire one arrow, but to turn and run instead, risking being attacked from behind if you’re not fast enough.
But, just imagine, that luck/randomness is in your favor and instead of the expected 20%, 30% of your archers hit their target on each volley. Now, there are only 40 swordfighters left, and your archers have a chance to win. Of course, the opposite could happen as well. Say there were only 80 swordfighters to begin with and you as the player have concluded that if 20% of your archers hit their target on each volley, only 40 swordfighters will be left. Therefore you decide to go ahead and try it. Unfortunately, only 10% of your archers hit. Now there are 60 swordfighters left and they simply wipe out your archers.

Comparing the two options:

Game without chance/randomness/luck/deterministic factor
100 archers will never beat 100 swordfighters in two volleys as they will kill only 20 swordfighters per volley and 40+ swordfighters always win against 100 archers.
60 swordfighters will never win against 100 archers, because in two volleys 40 of them will die and less than 40 swordfighters will always lose against 100 archers.

Game with chance/randomness/luck/deterministic factor
100 archers can with a little bit of luck/chance/etc win against a 100 swordfighters.
60 swordfighters can with a little bit of luck/chance/etc win against a 100 archers.

There are many factors that could change all these numbers if the game engine can deal with them. Wind might be a factor, the archers might use different types of arrows, and the swordfighters might wear different types of armor. Some archers might get off 3 arrows; some swordfighters might reach the archers before the second volley is ready. The archers might be positioned at the top of a hill, giving them chance for a third volley. The swordfighters might be running downhill, momentum speeding them up enough to beat the second volley. Some swordfighters might turn and run after seeing some of their friends die to arrows, some archers might run when they see the swordfighters still charge after the first volley. An arrow might hit the leader of the swordfighters causing general disorder amongst them.

To me, randomness should simulate logic, not necessarily realism. When my Shogun archers are standing at the top of a hill, I know they will do more damage and their arrows will have a longer range. I hope that this will affect the result of the battle, but I can never be sure. Even if I’m unlucky, I can still hope that I might be lucky at another facet of the battle, perhaps my own swordfighters strike true above and beyond expectation.

Randomness is like spices you add to a dish. Without it, you might miss out on the potential flavor fest. Too much of it, and you could ruin the entire dish. Many different spices can be used and some mix better with some dishes than others. Some dishes don’t require any spices, some need all the help they can get.

For some strange reason, Shogun would always lock up on my machine after about 15-20 minutes of play, so I could never really get that far into it The most bizarre thing is that the demo didn''t lock up on me, only the real game...so I wound up giving it to my dad instead.

But I did read the manual and did at least get a chance to look at the tutorials. I liked a lot of what I saw in Shogun, but was still disappointed by the RPS element to it. However, I liked having a general on board, factoring in terrain and it''s morale. I also liked it''s strategic turn based phase. If I had to peg a description of my game idea to people, I''d say it was 50% Shogun, 30% Close combat and 20% kohan.

But you''re right, luck can be a good or bad thing. I think the main thing is, how much should there be of it? I think you are right though...there are two kinds of randomness:

1) Variable, errr, variables
2) Unaccountable or unpredicatble situations

The first example is your accuracy comment about the archers. People have good and bad days, and their mark may be more or less true. Weather I think also falls under this category, although it could call in the 2nd as well. I remember playing the very old but still excellet Pirates! game where the wind would change randomly...so one minute your ship could be doing great...and the next minute you were struggling to tack against the wind while the other ship was about to give you his broadsides, a situation that happened very often in real life.

The second example is more like the misconstrued orders, death of leaders, or just plain old "how in the hell did that happen?" situations (like the missing aircraft carriers at Pearl Harbor). In these circumstances, the uncertainty comes not from having variables that are constantly changing, but rather from the vagaries of chance itself. Take for example the death of Stonewall Jackson, he accidentally got shot by his own men. And some think this may have been the turning point of the war for the south, because many think Jackson along with Longstreet could have persuaded Lee not to make the ill fated central charge at Gettysburg.

These sort of random elements can really throw you for a loop, and could either make the game very interesting, or a lesson in frustration. So again, is this a good idea or a bad one? Should random factors be relegated solely to the first type of uncertainty? Or should the 2nd kind of randomness also be factored? I bet Genghis Khan was savoring the moment his hordes would conquer japan...until a kamikaze blew his fleet to shreds. I don''t think many players would appreciate that happening to them (unless of course you were the Japanese player ). But who''s to say some war changing event can''t also befall the other player too? Random is random, so either side or both could have lady luck visit them with a whopper of random event.

I can see advantages and disadvantages to including the 2nd form of chance...so I''m not sure if I would include them in a game. On one hand, they ARE realistic, but on the other, they can really make you lose control of a situation very rapidly. So it can provide for some interesting scenarios, but may be a little overboard as to the loss of control and playability. what does anyone else think?