Lower ability scores in Five Torches Deep

Just checking out Five Torches Deep now and I want to say something about the “lower ability scores” thing because I think that’s based on kind of a mathematical fallacy. “Lower = more challenging”.

The average chance of success of a character on a “d20+mod vs dc” is going to be based on the mod and the DC. If you lower both you haven’t changed anything and if you raise both you haven’t changed anything. With that axiom in place, that means we can select an ability mod ranged based on other criteria. And then set the game’s DC to match that.

For me, the criteria I want is:

most relevant mods to be zero or positive so that we have to subtract as seldomly as possible. I hate subtraction.

some mods to be negative because we do want generally low mods and zero mods so we don’t have to add big numbers to the d20. I hate adding big numbers.

mods to not change wildly over levels so we can run sandboxes where you can go pretty much whereever you want over a big region. I love bounded accuracy. Some progression makes sense but not PF2E “adding your level to the roll”.

buuut... all that balanced with us wanting mods to have a big range so that your character’s ability actually impacts the roll, actually impacts the chance of success, more than just the flat d20. a +7 vs a +2, both rolling vs dc 15: the former would succeed 65% of the time, the latter would succeed 40% of the time. We want this because we want characters to matter.

We particularly want rogue skills to be completely sick compared to normal skills. Remember that their sneaking & climbing & lockpicking is competing against Invisibility, Spider Climb and Knock spells! We want the rogue to shine doing rogue stuff, not “Batman” wizard with a utility belt of rogue-emulation spells.

With all that in mind, that’s why I think vanilla 5e actually does things perfectly. “4d6 drop the lowest” creates an asymmetrical curve with a long tail on the lower end, i.e. a tendency towards the higher end. Which is the perfect algorithm for all four of our criteria above. Most numbers are positive, most numbers are 0 or 1, but there’s a chance of the occasional +4 in there which will succeed 20 percentage units more often than a ±0. (5e didn’t invent 4d6 drop the lowest, it’s from AD&D 1e. But 0e and B/X used 3d6 straight up.)

We have a proficiency bonus that significantly rewards trained characters while only sloooowly rising across levels. Also good, for sandboxing purps. Expertise is fantastic, doubles proficiency but there is still some tension in rolling, some chance of failing.

That said, yes I definitely definitely agree with their advice that most of the times there is no skill roll at all. If they look in the box and the prep says the key is in the box well then they find the key. Duh. We’ve actually cut skills out of the game (we keep ability rolls and tool proficiency rolls in). Buuuut since attacks and most everything else that does involve a die is based around the mods, I’m very happy with our method of generating the mods.

So yes they’re right that 99% of the play time should not involve rolling against your mods, but no I disagree with them on what those mods should be for the rare times you are rolling against them.

Dice & rules, especially combat rules, are “the guns on the negotiation table”. You definitely hope to get through the night without having to pick them up. But when you do need to go for them, you want them to work well.