Preliminary Dice Guide

DICE ROLLING - YOUR GUIDE:
Dice can add the appearance of fairness and risk to your quest, which are important! Players want to believe that you're fair and that they can fail. But they also take away control from both you and the players, putting it instead in the hands of random chance. That can be good - it makes things unpredictable - but it can also be bad, by derailing the quest completely.

Seriously consider whether your system gains or loses by its random elements, and what, specifically, it does gain or lose, before you add them. Consider making success and failure based on player decisions before you decide to make it based on random chance.

---

AN IMPORTANT REMINDER:
If the results of failure are not something you're willing to have happen in your quest (for example, the MC getting killed like a chump), don't roll for it. Just say it succeeds.

If the results of success are not something you're willing to have happen in your quest (for example, convincing people of an obvious lie), don't roll for it. Just say it fails.

If the task is trivial, don't roll for it. Just say it succeeds.

If the task is impossible, don't roll for it. Just say it fails.

In quests, players won't really notice that you aren't rolling for this or that thing in particular, or if they do notice, they generally won't care, so long as the result is respectable.

---

THE CRITFAIL PROBLEM:
An individual roll can have a positive, negative, or negligible effect. For the moment, we will consider the case of "best of three" systems, such as "1d20 best of three." When I speak of a roll for the following bit, I am referring to the result of a single case of dice+1d20, not all such rolls taken together.

A positive effect would be, for example, rolling a 19 when previously the highest roll was a 7. Positive effects produce some "oh yeahs" and "thank yous," but little more. For particularly important rolls, they produce sighs of relief. They cause your players to experience a temporary elated mood, which is good.

A negligible effect would be, for example, rolling a 7 when previously the highest roll was a 13. Although this roll does not help, it also does not directly hinder. These rolls are almost universally glossed over.

A negative effect would be, for example, rolling a 1 in a system with critfails. Such rolls produce an amount of bitching directly proportional to the importance of success in whatever particular case is being dealt with. You will hear epic whining about 1s, blaming whoever rolled them forever.

Consider in crafting your system what sort of moods you want to evoke in your players. Consider how an individual dice roll may influence the overall rolling system. If you enjoy the lattermost case - the epic whining - then feel free to include critfails. Otherwise, please remove them from play.

---

PROBABILITY DISTRIBUTIONS:
Best of X - looks like a curve upwards. 20 is more likely than 19 is more likely than 18, etc. The probability of getting at least X on YdZ is 1-((X-1)/Z)^Y. So, for example, the probability of getting at least a 19 on 3d20 is 1-(18/20)^3, or 27%.

Best of X with crits - looks like best of X, but with a big spike at the start where 1 is.

Average of/sum - Approximation of the normal distribution.

Generally speaking, you should only use a particular dice system if you understand what sort of probability distribution it produces. Best of 3d20 is a tight clump centered around 17, for example, which only very rarely goes below 6.

anydice.com is a helpful website if you want to get an idea of what your distribution looks like. For more complicated distributions, you may have to use some programming knowledge to figure out what the distribution looks like. If you don't know how to program, and can't figure out the probabilities involved using anydice or a text document and a calculator, you probably shouldn't use that system.

---

SHIT YOU PROBABLY SHOULD NOT DO:
DCs determined by a random number generator (i.e. contested rolls). It's stupid because 1d100 versus 1d100 amounts to a fancy way of flipping a coin; you are just obfuscating things for yourself so you have less of an idea of what the probability of player success is.

Criticals being a huge deal on 3d20. They happen more than every fourth roll. Don't.

Exploding crits. On /tg/, every explosion takes a minute or so to resolve. Further, exploding crits are wildly overestimated in their relevance. They have very marginal effects on medians and averages.

Rolling for the number or kind of dice. To explain, I mean that your "final" roll is XdY, and you roll some dice to determine X and/or Y. It's like exploding crits, except you can't even hope to take numbers from extra rolls! Also, it makes some rolls more important than others, which is probably inadvisable.

Do note that you can break these rules. They are not necessarily hard and fast, but it is the sort of thing you should try working with, before you go around breaking them.

---

SELECTING A DICE MECHANIC:
At its core, the purpose of a given dice mechanic is not to be mechanically interesting in and of itself.

A dice mechanic is, fundamentally, a way of segregating and organizing probabilities between 0 and 100. Best 1 of 3d20, for example, organizes them into twenty discrete segments of reducing size, starting from a subsection of 14.2% labeled "20" and continuing down to a subsection of 0.01% labeled "1." All systems are basically ways of organizing probabilities.

Certain ways of organizing probabilities allow for different things. For example, crits in best 1 of 3d100 can easily mean a lot more than crits in best 1 of 3d20, since they happen a lot less (~6% versus ~27%). Some systems enable a broader number of segments by adding secondary elements beyond high/low: whether or not there are any duplicates or multiples of 11, say, or the number of successes.

Allowing for tiered success (e.g. mitigated, unmitigated) is something any system can accomplish, but some do it more easily than others. Tiered successes or an independent "luck" element add something to dice rolls which the more conventional pure pass/fail lacks. It can be fun to try to figure out exactly how a given action wound up failing but the players got lucky, or how the players succeeded but at a cost.

---

BONUSES, SPECIAL ABILITIES, ETC:
Just add a number to the roll. A single dimension is easy to understand and control, multiple dimensions are difficult and complex for little gain. A +2 is not, mathematically, that different from an additional 1d20 in a take best of X, but fuck if it isn't easier to work with.

It does not need to be cool and special mechanically. Just make the fluff interesting and make it marginally better than the alternatives. Adding numbers is all you're doing anyway, if you hide it behind some weird dice mechanic (exploding crits, whatever), you're just making it harder for yourself.

If you're not adding numbers, the ability should be qualitatively different (e.g. a buff instead of an attack, or area of effect rather than single target). There's a lot of different variations on qualitatively different abilities, so try to take a moment to think of some interesting ones before you just do +2 damage or +1 die or whatever. This is true for noncombat stuff too. Try to offer the players new options for solving their problems, not just better versions of old ones. It's more interesting and more fun.

---

COMBAT SYSTEMS:
Quest combat needs to be simple and short. Optimally, you're going to do 20 minutes a turn for however long it takes; more likely, thirty to forty minutes a turn. A thread lasts maybe eight hours, ten or twelve at the most. That mean combat needs to resolve in only a few turns. My general recommendation is that important characters should die in 2-5 hits, and should be hit around half the time. Combat should be simple, and should require few rolls. The MC should never be at risk of dying in a single turn: do what you have to in order to ensure this is the case.

One thing to remember is the risk of players instakilling your big bad on the first turn! If players have a party, and they're all united up against a single big bad, just give the big bad more HP. A lot more. Possibly additional attacks, or passive effects, or just generally extra shit to do. Or abstract combat somehow so that having a boatload of allies doesn't let the players instagib the bad guy.