Quote from: Idinsinuation;266049That's quite likely because I think it's safe to say that most common games contain some sort of random element, dice, a deck of cards, etc.  That doesn't mean we remove the "game" status from classic games like Go, Chess, Shogi, and even Pac-Man.

Certainly, which speaks to the popularity of randomizers in games, doesn't it?  

More to the point, it can just as easily be said that chess (and the like) is a game with strong puzzle elements.  So, it's rather a hybrid.  It's certainly not an RPG.  RPGs share much more in common with Monopoly or poker.  You roll the dice, or get dealt a hand, and figure out how to make the best of it.

Importantly, what is missing from the puzzles is precisely that randomising factor.  For example, solitaire is a solo activity, but I think it falls well under the rubric of 'game' rather than puzzle.  Similarly, if one were to introduce a randomising element to Sudoku, it would be more of a game than a puzzle, rather more in line with the other popular computerized Asian pastime Mah-Jongg.

In that respect, whether the players are aware of it or not, if you remove randomizing elements, you are getting closer to a puzzle.  How do we avoid the patrol?  If we make too much noise, some of the orcs from room 15 will 'wander' out to investigate.  You can give that the guise of randomness any number of ways, but it really ends up being a triggered encounter.

Now, this certainly isn't a bad way to play.  It just seems rather superfluous to go out and drop $100 on rules if you just want to tell a collective story.  There are plenty of free rules-lite games out there that are more suited to that style of play.

Of course, that is just my preference for play.  

Quote from: droog;266052In principle, the same applies to pre-planned encounters. Ideally the GM makes principled decisions (exercising restraint and neutrality) in planning those encounters.

The question is, what is the principle a GM would use in a pre-planned encounter? If the goal is verisimilitude, in terms of approximating the risk and uncertainty due to incomplete knowledge--but also the expectation that the "world" isn't arranged according to some plan--it seems that, first, "preplanning" is inherently contradictory to that goal.

E.g., I've seen modules where, no matter what the PCs do or how quickly they do it, or where they go and how fast they travel, they'll run into a certain encounter, yet the structure of the encounter implicitly requires a coincidence rather than planning. IIRC this happens in D&D module X4 in a few places; the one that sticks out in my mind is running across an enemy encampment right before the army picks up and leaves. Implicit in the idea of the army being on the move is the idea that if the PCs arrived earlier or later, they wouldn't have run across the encampment. Yet the chance of the encounter occurring is 100% as long as the PCs move roughly within the confines of the mission. Once habituated to this type of encounter structure, players (many of them, perhaps not all, YMMV) will "know" that anything they run into has been planned. For me this sort of thing encourages too much of an expectation that the trail has been marked out specially for me and I don't have to worry about picking the right path for myself.

It seems that improvisation is less likely to be a problem--and I think it is. I have no doubt that improvised encounters can be done sort of whimsically or unconsciously, focused more on giving interesting grist for play than for pushing the game in a particular direction. But they can still deprive the players of freedom or lead to railroading--a classic case being the infiltration that never comes off without a hitch because the GM thinks that a game just isn't complete without a fight. Gygax suggested a method of dealing with this issue, but I think his advice tends to be ignored--it's as simple as backing off of deterministic improv and instead looking to possibilities and likelihoods, working the "encounter" from there. It would come as second nature to a wargamer, statistician, or anyone else focused on "modeling", and it comes from the same neutral ethos I mentioned: instead of deciding arbitrarily that X is true, the GM looks at the situation and asks how likely X is, given the context. Then dice are rolled.

This is similar to the techniques advocated in some "narrativist" games ("say yes or roll dice", "let it ride") except that the mechanics are explicitly subsidiary to the fiction, and it's understood that probabilities should be based on a vision of the game world as an objective reality, instead of arbitrating between competing visions. (This would be true even if the probabilities are arrived at by consultation among the GM and players.)

Quote from: Elliot Wilen;266127It seems that improvisation is less likely to be a problem--and I think it is. I have no doubt that improvised encounters can be done sort of whimsically or unconsciously, focused more on giving interesting grist for play than for pushing the game in a particular direction. But they can still deprive the players of freedom or lead to railroading--a classic case being the infiltration that never comes off without a hitch because the GM thinks that a game just isn't complete without a fight. Gygax suggested a method of dealing with this issue, but I think his advice tends to be ignored--it's as simple as backing off of deterministic improv and instead looking to possibilities and likelihoods, working the "encounter" from there. It would come as second nature to a wargamer, statistician, or anyone else focused on "modeling", and it comes from the same neutral ethos I mentioned: instead of deciding arbitrarily that X is true, the GM looks at the situation and asks how likely X is, given the context. Then dice are rolled.

Spot on.

Quote from: Elliot Wilen;266127The question is, what is the principle a GM would use in a pre-planned encounter? If the goal is verisimilitude, in terms of approximating the risk and uncertainty due to incomplete knowledge--but also the expectation that the "world" isn't arranged according to some plan--it seems that, first, "preplanning" is inherently contradictory to that goal.

Not really. Say your players have decided to take a trip across wild territory. You plan a set of encounters based on your understanding of the area and your estimation of probabilities. What's the difference, in principle?

There's also a middle ground, in that you can plan for random encounters, ie by making special tables for particular areas.

Quote from: droog;266860There's also a middle ground, in that you can plan for random encounters, ie by making special tables for particular areas.

That's my approach.

For my encounter tables, I'll put in some common critters, inhabitants, and events that aren't tied to a specific place. I'll also put in random encounters based on encounter locations in the area: for example, the dragon that lives on the moutain to the east may be a random encounter, or the goblin guards in room 12. This means that if the adventurers are successful in these encounters, the dragon's lair or room 12 will no longer be occupied by the dragon or the goblins if or when the adventurers arrive.

Yes, I think I mentioned that further up in the thread. I have another idea that I'll get into below, but the idea of tying random encounters to an area is basically a special case of tying random encounters to terrain type or dungeon level. Identifying specific "location-based" inhabitants with certain encounters, is also a good idea. At least, I've used it myself.

Anyway, I don't think any of the "pro random" folks have a problem with either of these approaches.

Quote from: droog;266860Not really. Say your players have decided to take a trip across wild territory. You plan a set of encounters based on your understanding of the area and your estimation of probabilities. What's the difference, in principle?

This is a little more complicated. I should state first that, contrary to what I may appear to be arguing, "random" isn't the gold standard here. Most things that are modeled by random techniques aren't really random in the mathematical sense--random (stochastic) modeling is mainly a tool to represent lack of information, or unpredictability in the strict sense of the word. For example, you might use a random model to determine the weather in a simulation, but while a true random process is defined by the fact that it can't be precisely predicted regardless of how much information you have at the moment, the weather is unpredictable because (a) there are too many variables to detect and track and (b) they interact in extremely complex ways that resist abstraction (chaos).* Another way of putting this would be to say that although the mathematical definition of a fair coin includes the requirement that each time you flip it, there's no way of knowing which side will come up, when it comes to chaotic systems like the weather it's perfectly reasonable to believe that if you could somehow exactly reproduce all the starting variables, the system would behave exactly the same way every time.

But for the sake of argument let's look at the difference between a set of encounters which have been planned ahead of time with no purpose behind them, and a set of encounters which are generated "on the fly" by rolling dice. Depending on how you go about it, there doesn't have to be a difference, but there are pitfalls. To begin with, it's not enough to generate the encounters; you also have to generate the intervals between them. Easy enough. In fact, as long as you're taking the time to prep encounters beforehand, you could simply take your probability estimates, roll dice for each day of the journey, and map out which days will and won't have encounters--as well as the encounters themselves.

If you say you can do that without dice, I'm a little dubious, but I'm also aware that most people are pretty bad at estimating probabilities and detecting small biases. Amos Tversky made a name for himself demonstrating this. (Example.) As long as you didn't make a blatant pattern (such as an obvious three-act structure, or a build up to a "boss" encounter), players would be unlikely to tell the difference--except to the extent they'd take dice rolling as a cue. And even there, I wouldn't be surprised if a sequence of pre-built encounters produced at whim would seem "more random" to players, provided you pretended to roll dice during the game, than a similar sequence that you "pre-rolled" but which you didn't pretend to roll dice for during the game.

Perhaps a bigger problem is that by trying to "appear random", you'd be likely to deprive yourself of the chance to simulate "real coincidence". Simply put, in real life sometimes you run into the kindly hermit right after you barely escape from the orcs. But if players don't see you roll dice, I think they're especially likely to see extremes of good and bad fortune as "railroading", and in a similar vein, you may shy away from planning such a string of encounters. Yet if you honestly roll the dice, I think the players are more likely to be genuinely elated by lucky coincidences, and to grit their teeth and bear down when faced with bum luck. If you don't roll the dice, they'll be jaded by the first and resentful about the second. YMMV.

Of course you could go ahead and roll dice even though you "planned" the encounters beforehand. I wouldn't be comfortable with this except maybe if the original "planning" had been via random generation; otherwise, it's dishonest. What's more, if you're caught, I think it would have negative effects on the trust that's essential to a good group.

(I'm going to take a break here. In my next post I hope to talk more about risk and "dynamicity", and to get into my idea for an alternate approach to random encounters, which is sort of a synthesis of some of the approaches raised in this thread.)

My experience is that players don't really care, and that they think you're fudging the dice anyway. Also, rolling for encounters as you go along is a royal pain and slows down the game.

I've gone so far as to roll up the weather for an entire season ahead of time, and it actually gave me some colour to use, but after that one experiment I realised that it wasn't a big deal and never did it again. It's much the same with wandering monsters. I started by rolling religiously for everything, but found in the end, unless you're running a lot of dungeons, random encounters get old.

Well, you asked what the difference was in principle, and that's what I'm focusing on; that the differences are ones that you don't care about doesn't really make them irrelevant--it just reveals a difference of taste (pretty much as I responded to Pseudo above).

You have your experience with your players, and at this remove I couldn't say what their exact tastes have been. But I think some people expect and enjoy being guided through an adventure. Furthermore there's a range of "less random" styles, from the rigidly-plotted railroad strung together via GM sleight-of-hand, to a sort of improvisational "keep the ball in the air" method--where maintaining continuity is more important than either following a particular path or enforcing the stakes of "serious" challenges. Of the two I prefer the latter, but at the same time I feel there needs to be an understanding that, since there's no "real" danger, there's also no need for the players to make a serious tactical or strategic effort.

Quote from: Elliot Wilen;267138Well, you asked what the difference was in principle, and that's what I'm focusing on; that the differences are ones that you don't care about doesn't really make them irrelevant--it just reveals a difference of taste (pretty much as I responded to Pseudo above).

No, you actually made a couple of claims about how the players might perceive things. I'm just saying that my experiences don't bear them out.

Otherwise, I'm not sure that you have identified any substantial principle, just some possible pitfalls.

Let me try one on you: the reason for rolling for random encounters is very much the same as the reason for rolling any dice at all in an RPG. Dice or other randomisers give you results you would not have thought of by yourself.

For this to work consistently and not cause the occasional stupid or boring encounter, you have to have one hell of a good table.

Quote from: droog;267142No, you actually made a couple of claims about how the players might perceive things. I'm just saying that my experiences don't bear them out.

And what the fact that what I'm writing is based on my experiences as a player?

QuoteOtherwise, I'm not sure that you have identified any substantial principle, just some possible pitfalls.

Of course, I can't slot myself into your past experiences, so we don't know precisely how I would respond as a player in your game. However I have to say I'm surprised that you would be so quick to dismiss the likelihood that different personal tastes lead to preferences for different game mechanics.

Quote from: Elliot Wilen;267148However I have to say I'm surprised that you would be so quick to dismiss the likelihood that different personal tastes lead to preferences for different game mechanics.

Not at all. I want to highlight it. I think there's a distinct strain in what you've been saying that denies it and tries to ascribe some universal value to random encounters.

Then I haven't gotten my meaning across. It's unfortunately common in these sorts of discussion for each side to think the other is declaring a single truth; at the same time, I believe that the reception of different game mechanics is associated with real differences of taste.

Then I haven't gotten my meaning across. It's unfortunately common in these sorts of discussion for each side to think the other is declaring a single truth; at the same time, I believe that the reception of different game mechanics is associated with real differences of taste.

Then I haven't gotten my meaning across. It's unfortunately common in these sorts of discussion for each side to think the other is declaring a single truth; at the same time, I believe that the reception of different game mechanics is associated with real differences of taste.