It's a good question, and it's something that often bugs me in games as well. Randomness is often introduced as a sort of handwaving "simulating this is complicated, so I'll just use a random outcome" kind of way. Which isn't very good. Winning or losing games on the strength of pure luck is not really much fun. However, there are many games which rely heavily on random outcomes which can be very interesting games.

The trick is to allow players to manage their risk. To do this, they need to know up front what the risks are, and have the tools to manage them. You also need to take a statistical view of the overall gameplay. Don't hang too much on one big dice roll - lots of little dice rolls allow things to average out so pure luck is less of an issue, and good risk management over the course of a game wins out overall, rather than a random lucky roll.

The real answer to your question depends on the type of game you want to create. Personally, if it's just underdog victories/unexpected defeats you want to throw in, I'd be more inclined to add depth to your game rules so that those can occur due to good/poor strategy rather than blind luck. Randomness is no replacement for strategic depth. However, if you want players to be making strategic decisions based on risk factors - and expect them to delve into that side of things in enough detail to appreciate it - then go right ahead.


EDIT: Oh and one other thing: Play Diplomacy. It's effectively a non-random version of Risk - and it's awesome.

I have to agree with Orymus. A tactic should never rely on whether or not units are simply lucky enough to be more or less effective with any given set of rolls. If a lesser force defeats a greater force based purely on luck, where is the sense of accomplishment for the player controlling the lesser force?

Case in point: I had a 1v1 Starcraft game where the other player had to leave early, so we agreed to meet our forces in the center. Turns out the other player was vastly better with economy and preparation, and had a much larger army including a crap ton of Carriers. Based on resource expenditure I was completely outclassed 3/2, but by properly utilizing a group of Scouts (my only advantage over him) I was able to attack from the rear and eliminate the majority of his Carriers early on, allowing me to (very) narrowly win the battle.

I'll never forget that match, but it was because I solved the problem of being overwhelmed with tactics. Had I simply watched my inferior forces decimate a larger one it would've been an empty victory quickly forgotten.

I think it’s fine to have some amount of randomness as long as the player won’t feel like the random values are actually changing the outcome of the game. Ideally, your random calculations should have a low variance and a large sample size over the course of any battle, so they will average out to relatively constant values and the player won’t be able to blame a win/loss on high variation in the random damage. Generally, you don’t want the outcome of the game to come down to a small number of battles that are decided with a few rolls of unweighted dice where the result of the battle can very possibly be drastically different from the expected result.

There is also in my opinion a very different mindset in playing turn based games vs. playing real time games in that turn based games (Civilization, Risk, etc.) can get away with being a lot more reliant on random outcomes than real time strategy. In an RTS, the outcome of an evenly matched battle is primarily determined by who has the better micro, while in a turn based game, the concept of micro might not exist at the level of individual battles, so random damage can make the battle a little more interesting. The way most of these games are designed, they would be a lot less fun if the outcome of a battle was set in stone before it even began, so the random outcomes are essentially simulating the effect of micro between the two players. From a strategic perspective, this also essentially forces each player to overcommit to each engagement just as you would want to in a real time battle where the outcome is uncertain. You don’t want to go into a battle with a slightly larger army where you only have a 51% chance of victory, instead you would rather wait until you have a much larger army or better positioning, maybe with a 70% chance of winning.

I think random events can work as well, especially if there’s a chance for the player to respond to them. For example, if a unit trips and simply takes 2 points of damage that’s not too interesting, whereas if a unit trips and takes 2 extra turns to get up, that’s an opportunity for the player to make a new decision in how to mitigate or take advantage of the situation.

I can understand the arguments for/against both types, but I vastly prefer random. I'm coming from a background of Iron Crown Enterprise's Rolemaster RPG system though, so.... I've always liked the idea that a lvl 1 nub with standard weapons/armor could take on and potentially get a critical hit on a vastly superior foe and take them out (i.e., a 'you strike foe through the eye killing him instantly' kind of thing). It also preserves a lot of tension (and yeah, presents a huge potential source of frustration) when your high level unit can be taken out by vastly inferior units. I think it also makes for a much more exciting and memorable game when there is at least some degree of randomness present. Otherwise no matter how many more strategic/tactical elements you introduce into the game you're still going to have something akin to chess.

But I guess it goes back to what some have mentioned above. It depends on what kind of game you're shooting for. No matter what kind I make it's going to incorporate RPG elements. Unit progression at the least, where each individual unit can "learn", get stronger/faster, etc. And RPG elements...I dunno', they just seem to go hand in hand with a large degree of randomness.

Interesting discussion!

Take care.

Random outcomes in turn-based games require appropriate player reactions (e.g. retreating hurt units, concentrating attacks depending on damage needs, placing units depending on how far other units are able to move, abandoning unfortunate units as decoys, etc.): far more interesting than applying a single optimal cookbook tactic. I recommend playing all units of all armies in Battle for Wesnoth for inspiration; it's an example with mostly medium-high damage randomness (depending on unit type; some have very low or very high randomness) mitigated by significant nonrandom defense bonuses.

Random outcomes in turn-based games require appropriate player reactions (e.g. retreating hurt units, concentrating attacks depending on damage needs, placing units depending on how far other units are able to move, abandoning unfortunate units as decoys, etc.): far more interesting than applying a single optimal cookbook tactic.

Surely these apply to deterministic games as well? Just because a game has deterministic mechanics doesn't mean it can be played optimally according to a script.

For that matter randomness doesn't necesarilly mean you can't. Dominant strategies are just as likely to occur as in deterministic games, and risk management techniques could be applied to mitigate unexpected results and create optimal strategies.