...
Well, that's great.
Meguca Micro Empire Quest (PMMM)
- Thread starter inverted_helix
- Start date
Well, that's great.
How did either of our diplomacy actions score a zero? Does area 11 really hate us enough to cancel out Mami's bonus?
This kind of thing is why I hate dice having so much say over action resolution. Is there any wonder why there's so little debate over ensuring that casualty percentages are at or below zero? A 1% monthly chance of casualties becomes 12% in a year.
This kind of thing is why I hate dice having so much say over action resolution. Is there any wonder why there's so little debate over ensuring that casualty percentages are at or below zero? A 1% monthly chance of casualties becomes 12% in a year.
Well crap. Those are indeed rolls that lead to a mini turn.
This is kind of a cornerstone of the medium. I mean there are a couple quests that don't use dice, but without dice you're completely beholden to the QM's feelings, which kind of removes any sense of chance. Either you have a QM that is too nice most of the time, or you have a QM that's evil all the time. To a degree that's the case even with dice, but the dice provide some much needed moderation to the QM's tendencies.
The main problem I see with quests that rely heavily on dice is that most GMs don't consider the distribution model that they're using. For example, a single d100 for all rolls means rolling a 1 is just as likely as a 50, is just as likely as a 100. Regardless of whatever N% chance of success there is, the results for any given action tend to be extremely random, and there's a feeling of a lack of control. There's a sense that, when you fail, it wasn't because you weren't good enough, or skilled enough, or didn't plan well enough, but rather that you failed "just because".
Just like with D&D's d20 system, the focus then becomes one of enhancing or mitigating the extreme results (such as aiming for 0% casualty rates, here), rather than aiming for challenges that match the 'typical' results, because there are no 'typical' results. For example, I don't like aiming for 0% casualty rates in our group because it feels so artificial, but I can see that it's necessary because of the dice mechanics. We don't tune our hunting to match the skills of the group, we tune it to match an artificial boundary. I'd prefer it not be possible to hit a 0% casualty rate (it just feels unrealistic), but at the same time the statistical distribution makes risk of death rise extremely quickly as soon as you go above 0%.
Many game systems avoid this by using mechanics of multiple dice rolls, either as a total count of individual dice that pass a given threshold (eg: the Storyteller system with multiple d10's; the Shadowrun system with multiple d6's; the Fate system with d3's; etc), or as the sum of multiple dice (eg: 2d6 for Lace & Steel, xd6 with GURPS, etc). Other systems get more creative, such as the card-based system of Castle Falkenstein, where the cards drawn by each player are random, but they get to choose which cards to use in any given situation (ie: some cards are better for combat, some are better for social, etc; if all you have are combat cards in a social situation, you're just kind of screwed, but you might still want to hold onto them if you think you're going to get into a big fight later).
For most of the dice system variations, the key is to give a strong tendency towards the average result, producing something more like a bell curve that is heavily weighted towards the central tendency, and extremely unlikely to reach the outer edges (at least without special additional mechanics). Shadowrun, for example, considers a 5 or a 6 on a d6 to be a success. The more dice you roll, the greater the tendency towards 1/3 of your dice being successes. Greater skill means greater numbers of dice, and the number of successes you roll indicates how well you did (which means it's not a pure black & white pass/fail). Storyteller and Fate systems work similarly, just with different types of dice.
So, if you were to roll 12 dice, you're very likely to get 4 successes. You have a reasonable chance at 3 or 5 successes. You're extremely unlikely to get 8 successes, or 8 1's (chances for crit failures). You then want to aim for challenges in the 3-5 success difficulty range. Anything that required 1 or 2 successes is practically guaranteed, and anything that requires, say, 6 successes is something where you want to start using special bonuses to boost your results.
The focus moves away from worrying about the extremes, and turns more towards aiming at challenges that are appropriate for your skill level.
Anyway, that's just my view on the dice mechanic thing. SayakaQuest has seriously soured me on heavily dice-controlled outcomes with flat d100 distributions. This quest doesn't bother me as much because there is so much dice rolling behind the scenes that the random distribution doesn't feel as heavily arbitrary; there are enough rolls that things start to even out. However it still has these moments...
(Aside: I'm guessing it was area 11 again, and despite your cautions and our brief concerns about it, we really should have seen something like this coming. I think it partly comes from that lack of a proper sense of the risk of an action, particularly in the whole random distribution deal.)
See I don't really agree with this logic. Yes you're as likely to roll a 1 as a 50 as a 100. However results are still considered in a form of a bell curve. Your odds of getting a perfect 1 or 100 are 2%, your odds of getting 2-30 are still 29%. Your odds of any individual number are identical, but your odds of getting a "moderate" result are still the same, it's simply the breadth of numbers representing a moderate result is the same.
For instance how I've calculated courier company results is doing a number of d100s equal to your workforce, summed the results and then determined the percentile using AnyDice. Anything above a 90 percentile result is a bonus, anything below 10 percentile a penalty. This gives the same statistical likelihood as if I just rolled a single d100.
Honestly it is, however I considered a fairly wide range of possible ways to do it. The problem I ran into is that without allowing you to reduce casualty chance by a meaningful amount, research felt pointless. And without making base casualty chance ridiculously high then you swiftly reduce it to 0%. And if base casualty chance was ridiculously high the number of girls dying begins having too significant an effect on population for the masquerade to make sense. I mean as it is, it could be statistically ascertained as people have discussed. Now imagine if I'd made base casualty chance 5x higher. (The fact that people apply statistics so heavily to this quest kind of limits how extreme I can go without it breaking people's SoD, and I did roughly consider how casualty rates would look from a total population perspective before I set them, they weren't completely pulled out of a hat.)
This seemed like the only way to really make it work at all, even if not well.
I interpret how high you roll to a degree in how well it goes (though there's a degree of smoothing as I said, where ranges tend to give same results).
Yeah, I tend to roll around 50ish dice each turn? There's a lot of smoothing effect from that many dice being rolled.
I'm not sure what you mean here? I mean you guys haven't behaved in the past as if you feel there's no risks associated with actions.
It's much too late to change everything now, but the research/training could've done things like "reduce casualties by 50%"- as in, 1/2 of casualties are prevented. Stack two and you get 75% reduction. That way we wouldn't be able to ever get to 0 but it'd get smaller and smaller as we research more and more (and if well-balanced, would shrink approximately the same amount our group grows)
If I understand correctly, you're just taking the sum, and seeing whether there's a 10% chance or less of rolling that on Xd100? Like, for 3d100, the 90th percentile is at about 217 (72% of 300), so if the 3d100 added up to 217 or more, it would count for the bonus?
Then yeah, in that case you're just using a convoluted 1d100 roll, effectively.
This is a common problem in the choice between multiplicative reduction and additive reduction.
A multiplicative reduction is more valuable the higher the starting risk (eg: 50% of 50% drops it to 25%), but rapidly diminishes in absolute value as the starting risk declines (50% of 2% is 1%, which is only a 1% absolute reduction).
An additive reduction is less valuable at higher starting risks (-2% from 50% only drops it to 48%), but increases exponentially as the starting risk approaches 0 (2% - 2% = 0%).
Heading out, writing more later.
@inverted_helix, I think what he's getting at is that a natural 1 has exactly 1% probability, just like a natural 50, but a lot of people like to treat the 1 like a major event, even if it happens more often than people think. If you're rolling 50 d100s pet update, then you have a 40% chance of turning up a 1 every time you do an update.
Most of those natural 1s don't amount to much of anything though. You've had some before. It's just when they come up on relatively important rolls that you have problems.
It's no different from rolling 2d10 and making a roll of 2 a bad result though. Increasing number of die doesn't really change probability. It simply shifts what numbers represent what probability.
Aranfan
Team Plasma Grunt
There are a lot more ways to get 11 on 2d10 than to get a 2 on 2d10. There is the same bomber of ways to get a 1 on a 1d100 as a 50 or 100.
There are exactly 10 ways to get 11 on a 2d10 roll. So it's essentially defining 46-55 as one value instead of 10. It doesn't make mid range results any more likely than a single 1d100 roll, it just compresses those results into a single value instead.
Similarly, I tend to play anything in the top or bottom 5% as a critical success or failure respectively. So it would be roughly equivalent to <=4 and 18<= results being critical on a 2d10 roll (though this would make the critical range 6% on each end instead of 5%).
Changing what dice pool you use simply makes the center results pool into fewer numerical outputs, it doesn't change the basic probabilities. It may have some psychological impact on what probability the GM assigns to critical effects, but fundamentally makes little difference.
Last edited:
Except that's life. Sometimes you fail because of events outside your control.
Not only do other people's choices have an impact on you, but random chance can ruin your day. It can even kill you.
People die from slipping on the stairs and falling down.
Or look at Freddie Gray in Baltimore. From the initial evidence it appears that the police roughed him up in a manner that they had probably done hundreds of times before without ever causing much more than bruises. But through a combination of bad choices and bad luck, they broke his neck, probably didn't realize it, and then killed him with the "rough ride" tactic. (All of which makes them criminally liable for at least 3rd degree murder.)
This then became a spark that resulted in massive riots, and attacks on a large majority of the Korean businessmen in the area.
Did the Korean businessmen not take enough precautions? Is it their fault for not being skilled enough to avoid the destruction of their businesses?
Rolling a natural 1 does happen in life. Sometimes for the lack of a nail the kingdom is lost. And when it happens, there is often not much you can do.
A more reasonable complaint is that the 1% chance of rolling a 1 on a 1d100 results in much more common disasters than expected in real life. This is why I think GMs should be careful about over interpreting natural 1s (and natural 100s too). They should not mean that common sense is suddenly thrown out the window and the laws of physics becomes melted cheese.
helix seems to be doing a good job at this from my viewpoint.
A natural 1 on a major important diplomacy check resulting in open hostility seems pretty reasonable.
A natural 100 resulting in some major success/advantage for the enemy seems reasonable.
What would not seem reasonable, to pick an example, would be for a 1 followed by a 100 to result in Mami's death by betrayal at the negotiations, and having Kyouko take over the Serenes. (If for nothing more than the fact that the Mami + Taura combo being successfully killed in a betrayal at the negotiating table seems far less likely than a 1 in 10000 chance).
Yet that's how some GMs I've known would in fact interpret it.
Helix's instinct to treat it more as: You rolled a 1 on diplomacy, they are very hostile. They rolled a 100 they have a big advantage. Now here is a mini turn, try and get out of this mess with as low a loss as possible.
That's a reasonable response to that kind of dice roll.
Yup.
Last edited:
As an addition to this portion of my above post: Multiplicative reduction gets less valuable the more of it you have, while additive reduction gets more valuable the more of it you have.
Caveat: I am in no way suggesting that the current system be changed. This is just a discussion on theory.
Hmm.. Take something like 3d100/3. That gives you a decently bell-curved shape.
With that formula, the odds of rolling between 20 and 80 (roughly 2 standard deviations) is 92.8%. That means the vast majority of your rolls will be in the 'normal' range — +/-30 from your base skill level (assuming our starting point is 50). You only have a 3.5% chance (on each side) of rolling either very high or very low. With a flat 1d100 distribution, the odds are just 60% that you'll roll within the 'normal' range, and a 20% chance each of rolling in the very high or very low ranges. (Put another way, a bell curve would have 95% normal, 5% abnormal, while a flat distribution would have 60% normal, 40% abnormal.)
Say we have a track runner that can run 100m in 12 seconds. If he does a hundred races, how many of those are going to be within about 0.5 seconds of that 12 second mark? How often will he suddenly run a 10 second race? How many times will he fall on his face?
Yeah, a beginner might fall down a noticeable number of times, but the more skilled the individual, the more likely they are to perform at the 'expected' level. The expected level may improve, but the frequency of performance well above or below the expected level becomes more and more rare. You get exceptionally good results by training to be better, not by getting 'lucky'. At the same time, outright failure also becomes almost unheard of.
With a flat distribution, that nat 1 and nat 100 result remains the same, forever. Even with an exhorbitant skill modifier, like 1d100+150 or whatever, you can never improve beyond that chance to crit fail, and you always have that same chance at a crit success.
In my personal view, the beginner has a much lower base score, but a much greater chance to either fail dismally or succeed spectacularly (relative to that skill). The expert, on the other hand, has a much higher base score, but an extremely low chance of either failing or succeeding in some exceptional manner, because they know exactly how well they can do, and how to perform at that level.
In terms of this quest, it would be something akin to saying that greens get to roll 1d100 (flat distribution; 95% of rolls centered on 50 are within +/- 47.5), vets get 2d100/2 (triangular; 95% of rolls centered on 50 are within +/- 38), and elites get 3d100/3 (bell curve; 95% of rolls centered on 50 are within +/- 32.5). With 4d100/4, the range is +/- 28; with 5d100/5, the range is +/- 25. (A specialist might get a bump, or there might be variations in the choice of roll types assigned to each class.)
Anyway, the distribution narrows, so that the final value tends to adhere closely to the actual skill level, rather than being more broadly dispersed.
The downside is that you can't hope for a "lucky shot" kind of win when fighting against enormous odds. That nat 100 just isn't going to happen. Winning thus skews heavily towards planning, preparation, innate skill, and methods of shifting the base modifier, either to buff yourself or debuff the opponent.
Just noting that I don't disagree with any of this.
Indeed. And, to an extent, you can view that as a different type of roll than what I was discussing above.
The bell curve distribution is more suited towards things where the results are a direct effect of your actions — primarily things like skill checks. Checks for purely random events (such as several of those you mentioned) are more easily treated on a flat distribution, as you're just saying, "This is how likely X is to occur."
The check for whether Mami and Taura could be killed in an ambush isn't a purely random event — it's a skill test — so I would expect it to use a skill-type roll, where the chance of a critical failure is extremely low. A pure random check would be something like Mami and Taura walking across the street, causing a driver (who would have run the red light anyway) to suddenly swerve away from them and hit the mom of one of the girls they were there to meet with. That's just one of those pure random WTF moments that no amount of skill can affect.
You've actually had natural 1s before, it's just that their effect is really dependent on how important the roll was in the first place.
(Though anecdotally, I vaguely remember one quest where players managed to roll a 3 on a 3d1000... on a nuclear fusion research project, I'm not sure that luck that bad has ever been found before. The GM properly (seriously, 1 in a billion chance) interpreted this as them destroying the continent. Not sure if that quest actually continued.)
I can kind of see your point. However my concern was not only once you had a bit of multipliers they'd seem less and less appealing but also it means that demons would have to scale multiplicatively to make further advancements useful to you in the way they are now.
I didn't actually say these rolls were with the same group. Enemy activity roll is from the same people as last turn though.
Will hopefully have first section up tonight or tomorrow (still busier than normal as I said) since it seems people are up for another miniturn series (which no one directly answered if they wanted it or not). Keep in mind this is a mini turn event to mitigate a bad though, whereas the other one was along the lines of a world event (which you managed to wring best possible results out of) so your good result is basically that it's less terrible, not that it's good on the whole.
I know you guys have been crazy obsessive over letting no one die and so haven't had any deaths in a while, but I want you to keep in mind this is a PMMM setting, people die. Mami is quite unlikely to die in any given situation (for a few reasons, some plot, some meta), her dying would basically take total player stupidity layered on top of bad rng.
To a degree all of your elites are a lot tougher to kill than the rest, but your people can die. (Though none are dying just from this roll to be clear, depending on how you handle it).
Thought you were arguing with my roll system and interpretation in general, so not quite as confrontational as I thought.
Oh! You mean something like that. Okay, I misunderstood. I thought there was an argument for replacing a 1d100 with a 3d100 but measured the same way you would a 1d100. Not switching to a 3d100/3
Yeah, that makes sense, you are trying to get a more bell shaped curve, gotcha.
Okay, that's cool. It's a more complicated mechanic that I would certainly consider for a quest that I want to be more planning based instead of luck, but the current system works fine too as long as the GM is being reasonable in his interpretation of natural 1s. (Which he has been).
Oh... interesting...
Yeah, I know. But there are expectations being set. Having horrible rolls against the level 3 Demon and Mami getting killed would have been devastating, but understood to be possible. Mami dying on a diplomacy roll would be infuriating in how blindsided we were by the event.
Yeah, for that, it's part of an analysis that shows that there is no single contest system (offense vs defense, assault vs resist, whatever) that always works well. If you take it far enough, you find notable flaws in any of them. It can be frustrating.
Ah, sorry. Yeah, I'm fine with another mini-turn. When things go wonky, we need to step in to deal with it.
As much as I'd hate for it to happen, I also recognize that it's a necessary and important part of the story. Part of why we're pushing so hard against it is because it can happen; it's the monster just waiting to pounce the instant we let down our guard.
Mini-Event 2
Mini-Event (to be named)
You take a bite of a dango as your legs swing off the side of some random three story office in Iwata, lightly hidden from human sight. One benefit of being a magical girl you're not afraid to abuse is you don't need to worry about eating right. You still regret the wish you made, perhaps if you'd not been so quick to make it… but it had made sense at the time, more time to think wouldn't have changed it.
You like being a magical girl at any rate, and in your time alone after you had… split from Mami you'd come to greater terms with your power despite what it had caused. The spear is no more evil than a car. Your abilities are only a tool, one you needed to survive, to live.
Your other half is idly killing demons dozens of kilometers away in another city. You couldn't really explain it to Mami, the feeling of being two places at once and able to do two things at once. The two of you together couldn't outsmart one of Mami probably, yet you could still do two completely different things at once without too much trouble.
You swallow the last of your snack and toss the stick downwards, piercing some rich guy's coffee cup. You pull a tiny grief cube from your pocket and hold it close to your gem, draining away a tiny amount of darkness.
The magical equivalent of a heat wave appears to your north, time to get back to work.
You make it to the miasma without incident, but you can already tell someone's been screwing with your territory again.
Trying a bit of perspective shift since it's appropriate since Kyouko is your only Elite in the targeted area and people seem to want less monotony. Tried to shift the word choice as well to highlight it.
Kyouko has very limited perspective on what's going on. This is in a sense a puzzle where you have very little information to go on for what's happening, try to keep that in mind. You'll gain more information as you go depending on your choices. This sort of style is going out on a limb for me so will probably be awful, but at least it's different. Instead of giving you a command center view, an individual view as Haman has suggested in the past.
This is really short for how long it took to write, it took me quite a while to determine how an enemy with their degree of intelligence on you would strike, and how that would look from the perspective of individuals on your side.
Kyouko is currently in your southern Iwata territory, Kyoclone is currently in the northern territory. Kyouko can handle a second body with minimal issues because magical degree of multitasking (superhuman, but still finite) she's kind of being introspective here and not paying attention to her other half which is just doing routine demon hunting. (And somehow character development is leading towards minor literal Multiple [near identical] Person Disorder).
Kyouko has very limited perspective on what's going on. This is in a sense a puzzle where you have very little information to go on for what's happening, try to keep that in mind. You'll gain more information as you go depending on your choices. This sort of style is going out on a limb for me so will probably be awful, but at least it's different. Instead of giving you a command center view, an individual view as Haman has suggested in the past.
This is really short for how long it took to write, it took me quite a while to determine how an enemy with their degree of intelligence on you would strike, and how that would look from the perspective of individuals on your side.
Kyouko is currently in your southern Iwata territory, Kyoclone is currently in the northern territory. Kyouko can handle a second body with minimal issues because magical degree of multitasking (superhuman, but still finite) she's kind of being introspective here and not paying attention to her other half which is just doing routine demon hunting. (And somehow character development is leading towards minor literal Multiple [near identical] Person Disorder).
Last edited:
What do we know about the problem?
1. It's a miasma spot that Kyouko can tell was caused by someone screwing with our territory.
2. Last turn, the miasma was caused by depleted grief cubes.
3. The vote did not explicitly specify what armor the Solo Elite in the Southern Territory should be wearing, but if we bought the Kevlar clothes and vests at the beginning of the month, then she should be wearing them.
4. As a Solo Elite, Kyouko should be able handle up to DS+12 without danger.
5. The demons from last month were at least DS+11 in order to have been able to injure the Veterans.
6. Kyouko is alone; no other hunters are within range.
7. No matter what we do, we can only mitigate the disaster, not eliminate it entirely.
What can we speculate?
1. Since the sabotage attempt from last month failed (and the dice rolled high), the saboteurs have probably upped their game.
2. The saboteurs may have added a lot more depleted cubes to summon a stronger demon (perhaps even a Class 3).
3. Alternatively, the saboteurs are waiting in ambush.
4. Alternatively, the saboteurs are using the demon as a distraction so that they can assault our house in Iwata.
[] You have a phone right? Use it to tell Mami about this shit, then go in. But be ready to bug out.
[] Call the girls at the Iwata house and have them be on alert.
[] Ask for clairvoyant support.
[] Check the surrounding area for an ambush or any traces of the saboteurs.
1. It's a miasma spot that Kyouko can tell was caused by someone screwing with our territory.
2. Last turn, the miasma was caused by depleted grief cubes.
3. The vote did not explicitly specify what armor the Solo Elite in the Southern Territory should be wearing, but if we bought the Kevlar clothes and vests at the beginning of the month, then she should be wearing them.
4. As a Solo Elite, Kyouko should be able handle up to DS+12 without danger.
5. The demons from last month were at least DS+11 in order to have been able to injure the Veterans.
6. Kyouko is alone; no other hunters are within range.
7. No matter what we do, we can only mitigate the disaster, not eliminate it entirely.
What can we speculate?
1. Since the sabotage attempt from last month failed (and the dice rolled high), the saboteurs have probably upped their game.
2. The saboteurs may have added a lot more depleted cubes to summon a stronger demon (perhaps even a Class 3).
3. Alternatively, the saboteurs are waiting in ambush.
4. Alternatively, the saboteurs are using the demon as a distraction so that they can assault our house in Iwata.
[] You have a phone right? Use it to tell Mami about this shit, then go in. But be ready to bug out.
[] Call the girls at the Iwata house and have them be on alert.
[] Ask for clairvoyant support.
[] Check the surrounding area for an ambush or any traces of the saboteurs.
From the perspective of Kyouko, I'm thinking she would probably call it in, but is unlikely to wait for backup unless Mami has been getting on her case about playing it safe. She might, however, do some looking around if there's any possibility that whoever planted this demon-bomb is still around (since she just detected it being released a few moments ago).
Oh, bah. I'll pulled armor off earlier during the various re-tweaks of the hunting assignments because there wasn't enough. With the addition of the IRT research, the number of hunters went down enough that everyone should have access to full kevlar.
The demons from the demon-bomb last month were noted as being around DS+20. If Kyouko waits for backup, a pair of elites can handle DS+20 safely.
Largely agree with boone's suggested actions.
[] Inform dispatch of another ongoing sabotage event.
-[] Dispatch is likely to then inform anyone else in the area, including anyone at the Iwata house, and start setting up planning for the demons.
[] Look for who set this off. The saboteurs are of more interest than the demons at the moment.
-[] Kyouko can be sneaky-like as she scouts the area.
[] If nothing is found, prep for taking out the demons.
@inverted_helix: does Kyouko have the option of waiting around for support before entering the miasma? What support is available? Does she have the ability to remotely dismiss and reform her clone in order to get quick support?