Hello

There is an area of game design that I have been wondering about for a while. Since it is something that must have been discussed before, I would appreciate it if you could point me to any good threads. Failing that, pointers to games that have tackled the issue, or opinions based on actual play would be most welcome.

As a GM, there are two things that are problematic to me: target numbers and NPC stats. In many games, these are often the same thing - target number equals some NPC stat.

In games that strive to do away with the rule zero or GM fiat, these numbers still represent that fiat. Well, yes, it is not really "GM fiat" since the numbers in most systems represent probabilities rather than certainties, but this is no respite: while utilizing his or her power to set the numbers, a GM needs to constantly judge statistical probabilities based on the system mechanics. In most cases, the GM does not have the time, inclination or capability for real statistical analysis, but has to rely on his or her gut feeling - which is rarely very reliable, especially when playing a new game.

The problem is even worse in games that advocate minimal prep time and ad-hoc story, NPC and conflict creation, while still requiring the use of NPC stats and target numbers. The GM must come up with these numbers on the spur of the moment. At least to me, this takes away some of the joy of GMing and replaces it with the wrong kind of mechanical stress over fairness and consistency.

Now, some of you might be comfortable with this power and say that I am just weak or lazy and not fit to be a GM, but for those that share my handicap: which systems and solutions would you recommend to deal with it?

Here is one approach we are currently experimenting with:

GM has a pool of, say, 10 suitably-sized dice (or cards, if you have a card-based system). GM throws these, in secret, behind a screen or something. Whenever GM needs stat or target numbers, one die is selected and removed from the pool, and the number recorded on the NPC sheet or target number reference for future use. When the pool is empty, all the dice are rolled again. Repeat.

Advantages:
GM still controls the order in which he or she uses the numbers, even if the numbers are random.
Handling time is very short.
Can be used both in prep and during play.
Can be adapted to many different systems.

Disadvantages:
If the pool use rate is low - not many numbers are needed per session - the contents of the partly-used pool should be recorded and remaining pool used in the next session. Recording can be a bore or just forgotten.
Towards the end of the pool, GM might not have any numbers left that fit an NPC concept or the rough difficulty level. (Although the GM can circumvent this creatively: The big guy is seriously hung-over and does not feel like doing anything physical, the system administrator left himself logged in, and so on).

As an example, with Andrew Cooper´s Balance of Power rules, we use the following pool of dice: 2d12, 2d10, 2d8, 2d6, 2d4. This matches the dice used in the game, and seems to fit with the character Attributes being in the 1-6 range - the GM will probably have some seriously powerful numbers, but is also assured of having some weak numbers.

Comments/suggestions/pointers?

Thanks!
Mikael

Everspinner wrote: In most cases, the GM does not have the time, inclination or capability for real statistical analysis, but has to rely on his or her gut feeling - which is rarely very reliable, especially when playing a new game.

The problem is even worse in games that advocate minimal prep time and ad-hoc story, NPC and conflict creation, while still requiring the use of NPC stats and target numbers. The GM must come up with these numbers on the spur of the moment. At least to me, this takes away some of the joy of GMing and replaces it with the wrong kind of mechanical stress over fairness and consistency.

Orx has something very similar to what you describe. While, in my Great Ork Gods the players are responsible for assigning difficulties. It is possible you could work out some varient of this system.

However, in general, I think if you're having a GM they need to have the power to control difficulties otherwise their ability to fulfil the role they are there to fill in most games is horribly hamstrung.

The Imp Game bypasses set difficulty entirely with a universal Target Number for all of the players. Players can wager points to make the TN easier or not wager anything and attempt more difficult TNs in an effort to win something for nothing. Success resets the pool to a difficult level and the succeeding player rakes in points while a failure means the TN remains at the adjusted level for the next player to roll on.
In this set up, however, the GM does NOT perform anywhere near the normal level. He or she becomes a player, complete with their own PC, and only becomes GM after the setup to make sure everyone is playing nice.

Hi Mikael,

I think it depends on what you want. I think for some styles of play, deciding a target number is actually a primary pleasure of play. Some people really savour determining just how the game world works and what exact number should be used.

I don't think you enjoy this. What I've thought of myself is a reward based on the size of the target number. For example, a target number of 10 gives the player 10 points. A target number of 20 gives 20 points (change the TN / point ratio to whatever you want). These points can be given if the roll passes, or even given both pass or fail.

You can see here that the GM, if he tries to covertly say 'no', is providing reward (particularly with the 'you get the reward on a pass or fail' method). It basically means you can't screw up as a GM, because whatever number you come up with, your providing adversity AND reward at the same time in perfect scale. With this, players will probably moan if your target numbers are too low, rather than moaning at high TN's! A healthy turn around!

My solution is just to use opposed resolution.  If I need a stat for an NPC I just use whatever logic I use on PCs to decide what their stats should be.  If I'm GMing I just play like a player.  If I happen to be playing Joe Bob the average guy then he gets an average rating.  I don't concern myself with whether or not it's scaled too hard or too easy.  Scaling things directly character effectiveness instead of consistency just ends up with deprotagonizing events like getting owned by a ninja doorman.  Some people I play with don't get how I do this and end up with decent scenes, but I've as of yet been unable to decipher why the concept is so opaque.

There are also the nice player assigned difficulty tricks others have mentioned.  For example, player chosen difficulty wherein the higher the difficulty the greater the reward for success, but the more severe the price for failure.

Check out Verge for an example of another game where players set their own target numbers. Players choose the level of risk, which corresponds directly to the level of reward.

Some kind of risk-reward connection is required to keep such a game from devolving into a make-believe with no system. If a game requires target numbers or difficulties or what-have-you, then someone has to set those numbers. Without a GM role (ostensibly with no personal stake in a conflict), a player's ability to set a target number impartially is suspect. That is, unless you tie risk to reward and let the player pay an opportunity cost for wimping out: if you don't take risks, you don't earn rewards.

Alternatively, make the target number buy off an imposed penalty that comes with every risk. This discourages players from entering conflicts, however, and conflicts are the interesting part of a game.

Mikael,

I think your system could work, especially if you want to still have some suspense on the players' part on what they face next (as they don't know what the GM rolled up earlier). But the pool remains stagnant, unless you have a mechanic to widen it/use bigger dice as the characters get more powerful, so that's something to think about.

I'm surprised no one has mentioned it yet (or I missed it), but Primetime Adventures actually gives the GM a Budget of extra dice. All rolls are made on a baseline (two dice, I think?) and the GM can spend additional dice only out of the Budget. Those dice then go into a different pool that players use to reward each other. It's a very cool system and addresses that same issue.

Thanks to all who responded, and thanks to Ron for moving this thread to the proper forum.

I will try to summarize the responses here, and reply to some of the points made.

GB Steve immediately confirmed my suspicion that my "little problem" might not be shared by the masses. This seems to be supported by Callan´s and cruciel´s responses and the fact that not too many people dropped in with "me, too!" comments. Sigh. Anyway, this is still an area I would hope that designers would give some thought to, with optional rules if nothing else.

GB Steve also raised an objection that my proposed system is just adding another layer of statistical probabilities on top of the old ones and not really fixing anything in terms of GM fiat. While this is certainly true, I would argue the following:

1) In this system, the additional probablities are a challenge for the game designer, not the individual GM trying to run a smooth game. And I guess it is reasonable to expect the game designers to be familiar with the statistical probabilities for the randomizers they are using - or at least that they spend sufficient time playtesting them to develop a good "feel" for them. I restate that the "remedy" I presented was just one option, but I still think it could be a good candidate for easing the burden of a GM struggling with unfamiliar rules.

2) As to the system resulting in situations where the players are calling for absurd or trivial conflicts just to spend the GM´s secret pool at a desired rate: First of all, while certainly possible, I consider this to represent the kind of approach to roleplaying to be something that would not happen in our group. Second, this is even mechanically impossible if we remember that the pool was meant to be used for both target numbers and NPC stats - a GM in a tight spot can conceivably just populate some NPCs with the remaining low numbers and then reroll the pool to make some high numbers available again.

This was very good critique which made me see that what I was proposing was not a "gamist" way of removing GM fiat, but rather just a possible way of making the GM´s life easier.

In terms of game design, several existing, in-the-works and just-an-idea systems were suggested, all with some sort of connection between the difficulty of a target number and the cost or reward to the players. I think the responses covered the space of possible combinations rather nicely, in terms of who sets the initial target number and whether points are spent, gained, or both. I have some work to do before I´ve checked all the systems you mentioned.

Adam Dray raised the excellent "theory-level" point that there must be either perceived objectivity or some kind of risk/reward system, because players are not altruistic. I was going to agree, then thought about "my system", which does not simply rely on GM objectivity but neither has a risk/reward system. Then I realised that the GM pool system can just be seen as a subcategory of the objectivity category, namely "objectivity supported by a randomizer". Thus Adam´s theory holds, and should be etched in The Great Book of RPG Design Theory. :-)

xenopulse, thanks for the very valid point that the GM pool needs to take into account changes in character power level. Currently this is not a problem with us, but definitely should be solved at some point. I would prefer widening the range (i.e. adding 2d20 to the pool) or shifting the average (only one d6, but three d8s) to raising the baseline (adding a +1 or +2 to the result of every die). This is because the players should still encounter normal people, and I want to avoid any sort of calculations when using the pool.

Well, we will definitely go on playtesting this system, as the others are not easily integrated into the BoP rules. So far I have just persuaded our GM to use the system, but I guess I can not really say anything before I´ve eaten my own medicine.

Once more, thanks for your valuable responses!

+ Mikael

Actually, I've always had a problem with this. A realization I've had recently was that whenever this crops up, the game becomes all about me running an illusionist sim. Whatever the numbers are, they don't really matter to me as a GM, and the whole decision in hindsight just feels fake and arbitrary.

I've talked about this before (with similar mixed replies to the ones you've recieved).

http://www.indie-rpgs.com/forum/index.php?topic=15219.0

All in all, I've found PTA has the best way to handle this.

And though I hate to doublepost, how could I forget about this recent beauty...

http://www.indie-rpgs.com/forum/index.php?topic=15993.0

Which is just a big brainstorm that tries to figure out solutions to this problem.
-Dave

I have this problem too.

I have come up with this solution:

• Reduce the set of possible target numbers. Humans are not capable of giving a meaningful description of too many different values, only of about 5 to maximum 7. Unfortunalty in most games one has to choose the target numbers from a range of 20 oder 100 values. Thus I reduce this to less values. For example in D&D I only use 15, 25 or 35 as target value. In Hero Quest only 17, 5M or 13M. and so on... Thus reducing the target value desision to the very easy choice between easy-medium-hard.• Choose target numbers relative to the stats of the characters -- combined with the solution above this becomes: either lower, higher or equal. With both lower and higher with a constant bonus/malus.

The same with game design -- when I design game, I take care that the range of possible target numbers is as small as possible.

In target number games a common practice I've witnessed, and practiced myself sometimes, is to simply let the players roll their dice, see how good a result they got, and then decide on how well they did. As GM you often have a reasonably good idea of the relative difficulty of a task, and if there is trust and goodwill between players and GMs this approach can save an awful lot of time and energy. It's still a bodge though.

Hero Point type systems are a great help on this issue. With a decent Hero Point mechanic the GM can confidently decide on exact target numbers, and state them publicly in anvance if apropriate, and still know that everyone at the table still has the ability to influence the outcome no matter what the dice may decide. Such mechanics reduce the need for the GM to fudge difficulty numbers because essentialy they can always be adjusted after the fact, and also mean the players get some kind of say in the flow of the game and what is important to them. Thus both GMs and players have more flexibility by essentialy allowing the players and GM to adjust target numbers after the fact.

Simon Hibbs

Someone else said it, and I have to agree: to some extent target numbers (and manipulating them) are part of the enjoyment of play for many people, myself included. I can see where the "stack of random numbers" would work for some things, but many things are both repeatable and non-random, and thus this system does not work for them. These are usually cases where known real-world quantities are involved.

If Robin Hood is facing down Sir NPC 1 through 12 at an archery match, you can't draw a random difficulty for each of them, nor can you just draw one random difficulty and have it apply to all of them, otherwise Robin's final challenge might be against Sir Schmuck, who just happened to beat a low, randomly drawn target number by more than everyone else. In this situation, the GM needs to assign a number that is sufficiently challenging to Robin, and which eliminates all but the most skilled competitors for the final match. Either the GM has to set this number arbitrarily, or the system itself has rules for generating a plausible number.

Or you just "go narrative" and zip straight to the final match with the NPC who happens to have some plot connection with Robin. But even then, you need an objective way to figure who is going to win top honors.

Greg Porter
BTRC guy

I'm with Greg on the 'stack of random numbers' idea. Sorry, but I think it realy sucks badly.

You are still expecting the GM to decide whether the difficulty should be high or low within the pool of available numbers, so you're not reducing the expectation that the GM should know what the difficuilty should be. You are simply reducing their ability to assign the exact number they would otherwise choose. In other words, th process is:

1. Decide what you would nrmaly assign as a target number
2. Determine if that number is in the pool.
3. If it is in the pool, decide whether keeping it for a later contest is more important than using it now.
4. If you don't have the number, or decided to reserve it, look at the other numbers available.
5. If you have a choice of higher or lower numbers, choose whether you can accept a higher or lower target.
6. Unless you think you need to reserver those numbers for later.
etc...

All of which adds up to a huge increase in the burden on the GM.

Simon Hibbs

Another composite reply:

Thanks, Dave. Those two threads were mighty interesting reading. I knew I could not have been the first to grind this particular axe, but the search function is just next to useless when trying to locate relevant threads. And thanks to Callan for the behind-the-scenes support - oops, sorry, now it´s public... :-)

Greg, your Robin Hood and 12 NPC knights example was a good one. It made me realize that this whole issue is a lot different depending on which kind of game you like to run. If you like your games to be consistent and realistic (in-game) overall, then figuring out the consistent and realistic target numbers/NPC stats is obviously a mandatory part of reaching that goal. From that point of view, the target numbers in your example archery context need to be carefully set so that they can be seen as an interesting test of the Robin character´s capabilities - and luck.

However, if we have a more narrativistic CA, the example would more likely be a single conflict where the player´s goal might be "Rise out of anonymity and be noticed by the fair maiden", and the GM´s "Robin realizes that he is a country boy, no match to professionals, but earns the respect and even friendship of his future arch-nemesis." In this case, the target number(s) might not easily derive from in-game constraints, but from the player´s and the GM´s story priorities. If I were the GM, either outcome would be interesting, and I would personally appreciate the help of the dice pool to set the number(s). And, if I used the pool to set any NPC stats during this conflict, I would definitely record those and use them in the future as well. Thus the pool system is a sort of crutch for me: I need a number, I do not have one defined, I do not want to go through the dubious process of justifying a certain number, thus I just pick one from the pool.

Simon, I´m afraid I cannot accept your statement that the proposed system "really sucks badly", when at least two of our group´s GMs and all the players have no problem with it.

I think that there is a difference in perspective at work here. Let´s see if I can put it into words.

Point #1: In your procedural example, the GM knows, in the first step, what the number should be. Now, if the GM is both happy and capable of determining that number, you are quite correct in that the GM does not need this system at all.

Point #2: More significantly, you are talking about what the target number should be. From this perspective, the pool can indeed be seen as restricting access to the full range of numbers. However, if the numbers are instead seen as a set of GM´s story priorities, then the pool is just a convenient way of reasonably spreading the priorities across the full spectrum, and for the GM to stay relatively "honest", not setting everything "high priority." The difference between the approaches is clear: if you can happily determine what the target numbers should be, there is no need for such a distribution; rather, the distribution is determined "online" by the game events and by the perceived difficulty of the challenges the players choose to get involved in.

Thanks for the constructive criticism. I think I fell into the trap of trying to make the initial problem statement too broad, i.e. I did not clearly define which sorts of games and situations I am interested in. If we are still not on the same page here, I think this is getting close to the point where we just have to agree to disagree.

While writing the above, I realized that I had overlooked the option of having a fixed pool, a set of numbers that gets used over and over again. While that would surely cover the same function as the randomized version, I do prefer the feel of rolling the dice and the ease of just removing the used values.

Thanks again for your input,
+ Mikael

I had this light bulb thing happen in my head after this Monday's game session, in relation to my early post in this thread and Difficulty/Success Needed as it relates to GM Fiat.

I'm going to quote a couple things (emphasis mine).

David Bapst, In that other thread wrote: My personal tastes make me dislike GM fiat: maybe it's just that I'm normally the GM/referee/whatever that runs the game, but in my experience it just leads to stories that don't make me, the GM, satisfied (when I'm going for Narr). I have too much control over what will happen to actually make what happens interesting or surprising.

See, hypothetically, the GM will come up with "reasonable" difficulties. We've all seen the "rulers" as I like to think of them, with the "Climbing Wet Slippery Rocks" tables and such. But, as a GM for the past three years (college student: lots of opprutunity to run lots of games), I haven't ever done used them like that. I've never opened up the books suddenly and gone, "Man, did it rain the night before? Is the roof a little slippery? I wonder if there's moss or something?"

Hell no. Normally, I say some arbitrary number. It's almost random sometimes. Most of the time it's pathetically easy to average, unless it's something I don't want them to do or doesn't fit the assumed genre, in which case it's rather hard.

Everspinner wrote: However, if we have a more narrativistic CA, the example would more likely be a single conflict where the player´s goal might be "Rise out of anonymity and be noticed by the fair maiden", and the GM´s "Robin realizes that he is a country boy, no match to professionals, but earns the respect and even friendship of his future arch-nemesis." In this case, the target number(s) might not easily derive from in-game constraints, but from the player´s and the GM´s story priorities. If I were the GM, either outcome would be interesting, and I would personally appreciate the help of the dice pool to set the number(s). And, if I used the pool to set any NPC stats during this conflict, I would definitely record those and use them in the future as well. Thus the pool system is a sort of crutch for me: I need a number, I do not have one defined, I do not want to go through the dubious process of justifying a certain number, thus I just pick one from the pool.

...

Point #2: More significantly, you are talking about what the target number should be. From this perspective, the pool can indeed be seen as restricting access to the full range of numbers. However, if the numbers are instead seen as a set of GM´s story priorities, then the pool is just a convenient way of reasonably spreading the priorities across the full spectrum, and for the GM to stay relatively "honest", not setting everything "high priority." The difference between the approaches is clear: if you can happily determine what the target numbers should be, there is no need for such a distribution; rather, the distribution is determined "online" by the game events and by the perceived difficulty of the challenges the players choose to get involved in.

Jason, I think I do pretty much what you do.

I set difficulty based on the importance of the desired outcome and the effectiveness of the opposition. But I explicitly disconnect resolving the outcome from the narration of the player's action. Say you're in a swordfight, and the outcome at stake is "Do I beat this guy". Both characters are expert swordsmen. The difficulty will be high -- but that's the difficulty to beat him, injure him, or whatever. The dice will say nothing about how well they fight, just the outcome.

So even a badly blown roll on the player's part won't result in "You whiff". There'll probably be a flurry of thrusts, parries, and ripostes, but no actual hits. We know he's a great fighter, and the description of the action is true to that, even though the outcome is failure.

The benefit of decoupling outcome from description is that if you do get some difficulties wrong, it doesn't show the character as having inconsistent abilities. I think that's what you're saying with "the range, standard deviation, and all that jazz impact the feel (favoring luck, skill, etc) the game creates, but should not impact character effectiveness within that feel." Is that about right?

Andrew wrote:
Jason, I think I do pretty much what you do.

I set difficulty based on the importance of the desired outcome and the effectiveness of the opposition. But I explicitly disconnect resolving the outcome from the narration of the player's action. Say you're in a swordfight, and the outcome at stake is "Do I beat this guy". Both characters are expert swordsmen. The difficulty will be high -- but that's the difficulty to beat him, injure him, or whatever. The dice will say nothing about how well they fight, just the outcome.

So even a badly blown roll on the player's part won't result in "You whiff". There'll probably be a flurry of thrusts, parries, and ripostes, but no actual hits. We know he's a great fighter, and the description of the action is true to that, even though the outcome is failure.

The benefit of decoupling outcome from description is that if you do get some difficulties wrong, it doesn't show the character as having inconsistent abilities. I think that's what you're saying with "the range, standard deviation, and all that jazz impact the feel (favoring luck, skill, etc) the game creates, but should not impact character effectiveness within that feel." Is that about right?

Jason Lee wrote: The heroic advantage of a protagonist comes from player control mechanics (FitM, hero points, etc).

The probabilities of the dice are irrelevant to assigning difficulties in play (even with unopposed rolls if they have a chart) - the range, standard deviation, and all that jazz impact the feel (favoring luck, skill, etc) the game creates, but should not impact character effectiveness within that feel.

Everspinner wrote: However, if we have a more narrativistic CA, the example would more likely be a single conflict where the player´s goal might be "Rise out of anonymity and be noticed by the fair maiden", and the GM´s "Robin realizes that he is a country boy, no match to professionals, but earns the respect and even friendship of his future arch-nemesis." In this case, the target number(s) might not easily derive from in-game constraints, but from the player´s and the GM´s story priorities.

I've been reconsidering some of my motives behind the topic I made. I think reasons for my complaints are implicit in my long held bad GM habits... they become ready tools for illusionism. I graduated from the WW school of gaming, where the numbers of successes gained are used as a vague range for the GM to narrate the result without any idea of the real difficulty having existed in the first place.

I am adding this to this discussion as remarks have been drawn from my posts, and thus I feel it is important that they are interpretated correctly.

If one person both establishes the situations and decides the likelihood of success/failure -- from moment to moment -- then that person is the author of the game. Period. It may be very subtle. It may still be loads of fun for everyone. But the authorship resides in one place only. The other players are being carried along on a wave of probability.

In this kind of play:
"We go in the room. What's there?"
"It's a warehouse full of wooden crates. As you enter, 6 ninjas leap from the shadows and attack! Since they're ninjas, they should be CR 12 and they get two attacks per round..."

That second person is authoring the game. If that isn't obvious, we're not going to have a very useful conversation on this topic. The input of the first person is simply to decide (sometimes) in what order things are authored.

If you care about authorship authority, you need to address this issue. PTA is probably the best example of a fix for this. Dogs in the Vineyard does it too, although using a much more traditional method (the "pre-made encounter").

John, that's hardly authorship...it's almost a bang (it's the same as 'a goon walks in the door, guns blazing' plot device). This is a better example of authorship:

GM "Six ninja's attack you! Their CR 12, their pretty tough!"
*Players look like they are going to attack*
GM "Err, CR 12 AND they have 3 levels each of ass kickery!"
*Players look worried but one of them might still lead the others to attack*
GM "And....they have vorpal swords!"
*Players declare that they run*
*GM is happy, flipping over his notes to the page entitled "When the PC's run, this happens..."*

Authorship/illusionism can come from adjusting the numbers to force just the right behaviour from the players, that the illusionist intended. However, being able to fully control numbers doesn't make someone 'the' author. It just makes it terribly easy for them to enter illusionism at any given moment.

Yes, Callan, you're right. Control of difficulty gives the controller instant access to illusionist techniques when they want them. Add to that the likelihood of illusionism without intention (the GM doesn't even know he's doing it) and you get the "sole author" thing I was talking about.

David wrote:
I've been reconsidering some of my motives behind the topic I made. I think reasons for my complaints are implicit in my long held bad GM habits... they become ready tools for illusionism. I graduated from the WW school of gaming, where the numbers of successes gained are used as a vague range for the GM to narrate the result without any idea of the real difficulty having existed in the first place.

I am adding this to this discussion as remarks have been drawn from my posts, and thus I feel it is important that they are interpretated correctly.