I'm reposting (with some minor edits) an old blog post where I considered the impact of the concept of Bounded Accuracy as it applied to my custom m20 games. Maybe here, I'll get some discussion on the topic. If you don't think that the concept of bounded accuracy is desirable, then this topic probably won't be of use to you; I discussed what bounded accuracy is, how well my games met the requirements of it, and then because I thought that the concept was a good one, I discussed some fixes to give me a better result, which are now baked into the current versions of my games. Anyway, on with the reposting...
Another discussion on one of my favorite topics; or rather, one of my favorite ideas from 5e. Not that I spent a lot of time investigating exactly how it was implemented, but the concept is, I think, the first time that the the developers tried specifically to excise one of the main problems that I've always had with D&D of literally every edition ever. Although to be fair, it only got worse as editions went up up through 4e. For 3e and even moreso for 4e, the designers recognized the problem but rather than try to eliminate it, they embraced it and designed around it, deciding for reasons of "sacred cowedness" or something that it just wouldn't be D&D without it. That lead to the debacle that is higher level 3e and 4e play. There's a reason why hardly anyone plays at that level, and why for the most part it's seen as fundamentally broken, or at least highly undesirable. Arguably in the case of 4e it's not broken so much as it's just stupid.
E6 was the solution to fix it in 3e, but that wasn't nearly as elegant a solution as the bounded accuracy model of 5e, because it essentially made large swaths of the Monster Manual unusable. Even a continuously improving E6 character who's been around for a long time kind of tapers off and tops out at an effective CR of about 10 or so. So, the really high CR monsters, like say a CR 20 balor or old red dragon are just not really usable at all, because an effective character level 10 party can't expect to ever really defeat them.
Which maybe isn't terrible, because it encourages players to do something to deal with them other than just charge in to standard combat. But at the same time, it's not really the intention of anyone to design monsters that aren't really going to be usable to most players, so E6 had a rather steep cost, I think, associated with it.
With 5e, a system designed around the idea of bounded accuracy from the very beginning, on the other hand, we get a system that just works a lot better. Sadly, however, it means that you need to start over. So, let's have a quick discussion about it, and then let's have a quick discussion about how well Fantasy Hack and my other m20 offerings meet the design requirements that 5e spelled out. I'm using the text here to define bounded accuracy (https://www.dandwiki.com/wiki/Understan ... Guideline)). If you follow that link, be sure and expand the box to see the original developer's text (from an article on Wizards' website that now is unavailable.)
And because I'm using that link, let me address some of the text there first; specifically the "What Bounded Accuracy Is Not bullet points.
BA has nothing to do with checks of any sort. The limits imposed upon ability modifiers thanks to BA alters the standard DC range, but this is an unintended side-effect.
Uh... no, that's not true. Did the guy writing this not even read the text right above this point from the developer's own pen? It may be a side effect, but it's not unintended. This is a key component of bounded accuracy, and was discussed at length in the developer's notes. Any system which doesn't address skills, saves, and other checks in addition to attack rolls is not a full bounded accuracy system.
BA has nothing to do with saves of any sort. Because saves are just a keyworded check variant in 5e, they are affected in exactly the same way.
No. See my comments directly above.
BA has nothing to do with damage. It incidentally has implications for how damage winds up being delivered, because the same modifiers apply, and because it alters the hit frequency for attacks.
Again, it's not incidental. It's integral. Granted, the impact is less dramatic than it is for attack and AC comparisons.
BA is not about reducing the power of character level. Level is still king. It does increase how long a lower-level character or lower-CR monster will last against a character of a given level.
Yes, I suppose, but that's relative. Level is "still king" in a relative sense, but character power by level is lower than it was in, say, 3e at the same level. However, since the world doesn't "level up with you" as it rather dubiously does in 3e and even more overtly in 4e, that doesn't matter, because the whole concept of how powerful leveling is supposed to be and what kinds of threats you are expected to deal with and how is rejiggered in a bounded accuracy system. Depending on how charitably you want to treat this statement, it's pedantically correct, but context makes it actually incorrect from the point of view of a player, or the whole thing is irrelevant. In other words, in a strictly traditional sense, you can still face the same gamut of threats at the same levels in 5e under a bounded accuracy paradigm as you could during a non-bounded accuracy paradigm, but you also moved completely out of being threatened by things that were threatening to you at lower levels without bounded accuracy, which fundamentally changed the genre of the game from one of fantasy adventurers to one of fantasy superheroes. Bounded accuracy, among other things, maintains the genre through the entire spectrum of levels.
BA is not about increasing the difficulty of lower CR enemies. Rather, it allows lower CR enemies to still produce some degree of actual threat- no matter how little- against a PC of any level, and the same for a PC against a high CR monster.
This is mostly true. But it's not just enemies and CRs, it's also checks and DCs.
BA is not intended to alter the overall difficulty or risk of the game. Ultimately, how difficult the game is depends entirely on what the DM decides to throw at the players. BA just makes that job a lot easier, by giving them a wider range of options for how to achieve a given threat level.
Yes, and that's a great thing. But after describing the things that the writer of the wiki entry (often incorrectly) believes that bounded accuracy is not, as well as a historical digression to discuss the nature of the problem that bounded accuracy is getting rid of, he lists four things that it is, and I think this is spot on. So, let's discuss how bounded accuracy works in 5e and compare it to the situation in Fantasy Hack to see how well I meet the expectations of bounded accuracy. If I'm a little off, will I want to make minor modifications to Fantasy Hack and it's derivatives? Possibly. Let me "talk" through this as I make this post.
First, they deserted the magic item economy. This was an effect generated by developers assuming players would have magic items providing a minimum bonus at given levels and preemptively building those bonuses into monster ACs to compete. It made magic items worthless, because players could only use them for a short time before being forced to upgrade, and also made magic items mandatory because you couldn't function without them. It forced DMs to plant a regular progression of magic gear as rewards during play, regardless how shoehorned-in it became. The magic item economy was the main driving force behind the treadmill. Instead, monsters would be built on the assumption that players do not have any magic items.
I disliked the magic item economy for aesthetic reasons that have little to do with the mechanical implications, but the fact that the magic item economy fundamentally forced us into the so-called "treadmill" means that it absolutely had to go. Keep in mind that in Fantasy Hack, as in older versions of D&D, there is no CR, so I'm not trying to calibrate level vs CR the way 3e, 4e and 5e have done.
Now, also keep in mind that this doesn't mean that there may not be threats out there that require magic items to deal with. But having a panoply of magic items just to do the day to day is strictly forbidden as a concept. Fantasy Hack already does this; in fact, I only added magic items in as an afterthought in Appendix I anyway.
Second, they sat down and decided that the total flat bonus a player could receive on a check could not exceed the value of one whole die. (Anything more than that, and you just have an eternal arbitrary arms race of increasing values; the "treadmill" of the past editions) In other words, +20 is the theoretical desired limit of all combined bonuses to an attack roll. There is some debate, but it appears as though, by core rules only, the highest check result possible is 47, a bonus of +27, and it requires a lot of fiddly build options which the developers probably hadn't anticipated, plus a good circumstantial situation, and is not applicable to attack rolls. In other words: they did a good job of staying in that limit. Generally, nobody will ever be able to roll higher than 31 for an attack, check, or save.
I didn't specifically plan for this, and I only achieve it, I believe, because I halved the number of levels. Let's see for the sake of argument, how high my bonuses for To Hit can get with a character in Fantasy Hack?
If I create an orc fighter with maximum STR, he'll have +6 for his ability score, but an additional +3 can be added as he goes through the level progression to make for a total of +9 (!). He also gets +1 as a Fighter bonus, which increases to +3. He gets up to +10 as a bonus for his level at the top of the progression. And if you customized the Fighter class to give you the Outdoorsman's weapon bonus, but applied to whatever weapon you use, you'd have an additional +1. And let's say, just for the sake of argument, that he's also got a +3 weapon, which is the maximum bonus allowed by the magic item rules.
That gives him a to hit bonus that is higher than 20; 26 to be specific. Now, granted, that's a character designed around one thing: getting the maximum to hit bonus. It also requires a +3 magic item, and there's no guarantee that such a thing even exists in the Timischburg or DH5 settings anyway.
I think I squeak in under the definition, although not entirely by design. I'm actually a little bit surprised how high you can get that bonus if you really trick your character out to get it, but then again, even if you don't, your 10th level human fighter is going to be; what, probably only about +5 less than that, especially if magic items aren't readily available. Just about any fighter-style class will be. And even non-fighter style classes will be probably over +10 at 10th level, unless they have a really weak STR score.
Third, they divided how that maximum bonus would be proportioned between standardized sources. In general, these sources are the only sources of bonuses to an attack, check, or save. The sources are: ability score modifier, (maximum of +5, attainable even from first level) proficiency bonus, (minimum of +2, maximum of +6, grows slowly with level) and magic gear bonus, (max of +3, but it's unlikely you'll ever even see +1). This gives, under optimal conditions, without feature intervention, a maximum roll result of 34 with magic.
Yeah, I blow that away, I'm afraid. Theoretically there could be conditions that give situational bonuses already on top of my 10th level orc fighter, but ignoring those, I've got a maximum roll of 46 based on my calculations above, and heck; the average roll is still 37 at 10th level with that build. If I'd use the optional wose race, I've be even a point higher.
I think I may have been too generous in allowing bonuses to stack without really seeing how far they could be pushed to potentially break the bounded accuracy paradigm. Which, granted, I didn't specifically attempt to create a bounded accuracy system by the strict definitions of the 5e design team, but I did want to do so at a looser high level approach.
Fourth, they made sure that PC ACs could not exceed 21, and monster ACs do not exceed 31. (See how that 31 lines up perfectly with the maximum possible roll result without magic? Notice that the maximum PC AC is 10 less than the maximum monster AC, a full half-die lower.)
If I wanted to maximize AC, with the races I currently have, I could get a maximum ability bonus to AC from DEX of +8 (I could increase that maybe a point or two with a custom race). +5 for level bonus, +6 from armor, an additional +3 for magic armor, and +2 for a shield. Total AC would be 10 +8+5+6+3+2, or 33. That might have another point or two of wiggle room, but again—that's higher than I really envisioned.
I think that my problem was that even with only 10 levels instead of 20, I didn't really give a lot of thought to the concept of playing all the way to 10th level.
Sigh. I'll probably have to do some work to keep those bonuses a bit flatter. The highest AC any of my monsters has, Cthulhu, is only 35, and the players can theoretically hit that range pretty comfortably if they really trick out their character specifically to do so. My top ranges, compared to my most powerful monsters, are too high. And too many foes would literally be unable to ever hit the characters at all.
I had thought that I had higher armor values impose a penalty on how much DEX bonus you could apply, but I guess I decided that that was too fiddly to work with and did away with it. But even then, it gets out of control. I actually think a big part of my problem is that ability scores can get too high. I should probably put caps on those too.
Notice that most of this doesn't actually put limits on players. It actually puts limits on the developers when designing content the players can use. The standardization of player attack bonuses allows them to anticipate the bonus range any character can put out at a given level, regardless of class. This allows them to design monsters which have ACs which alter the probability of a hit based on PC level. Rather than probability being rapidly pushed to 0% or 100%, the monster becomes viable for use against a much wider range of PCs. By having limits to player AC that are not tied to level, they can change the hit rate for monsters by adjusting only their attack bonuses. Because the two things are no longer tied together, it is now possible to have monsters that always hit and always get hit, always hit but rarely get hit, rarely hit but always get hit, or rarely hit or get hit, as well as anything within those four extremes. Finally, the whole point of all of this was to make lesser enemies still useful in larger numbers at higher levels, and powerful enemies still survivable at lower levels. (Survivable is not the same as defeat-able. TPKs still happen.) That means you no longer need to have special tier-balanced versions of each monsters, or special minion monsters, you can just use a higher CR monster to present extra challenges, or throw a whole bunch of lower CR monsters to make up a total CR equal to one big monster.
This is what I need to actually spend a little bit of time on proper math to get accomplished for my m20 games.
The simplest way to address the higher To Hit and AC class scores are to do the following two things:
--Cap ability score bonuses. You can only ever get a maximum of +4 plus your racial bonus to your ability score, which you can get at 1st level. Either eliminate the ability score increases as you level up, or say that if you're already at your max ability score in one ability, you have to apply it to another. If you somehow have maxed ability scores in all abilities, you are not only one lucky bastard with your ability score rolls, but you lose your ability to increase them when you level up. It's hardly like you need it.
--For your AC, you can take either your DEX bonus or your Armor and shield bonuses, but not both. Wearing armor makes you unable to dodge blows as effectively, and the whole point of armor anyway is to block blows that you can't get out of the way of. This isn't simulationist realistic, but it does reduce the maximum AC that a player character can get, if tricked out, by about 8 (technically a little bit more if you consider the possibility of magical armor.)
That's still a little bit higher than the 5e bounded accuracy ideal, but it's now within a couple points of that ideal, and I did so with an elegant solution.
If I felt like more was still needed, I could halve the rates at which skill and To hit bonuses apply to characters; instead of adding your level, you add half your level rounded down. I probably won't do this, but that would be the next step. AC progression is already halved, but I'd make it apply only ever third level instead of every other level, and it would also reduce the maximum by a few more points. It's not perfect; low level opponents like a goblin or human commoner still can't hit PCs past a certain level because they can't get an attack roll higher than a 21 or 20 respectively, but that only becomes an issue at very high levels (in the m20 scheme of things, where 10 is the cap.) Any solution that addresses that needs to do so in a way that's elegant enough to not be fiddly and annoying; it may not be worth addressing at that point. Either that or eliminate the 10 baseline, or change it to a lower number. I'll have to think on how feasible that option may or may not be.
UPDATE AGAIN: I've rethought the AC problem. Rather than making you pick Armor Bonus or DEX bonus, how about we just get rid of the level based AC progression altogether? Although I actually like the level-based AC progression in many ways, it was specifically implemented in the d20 games from which I borrowed it as an alternative solution to the magic item economy to keep your AC progressing as it needs to. Therefore, my solution is becoming a new problem in the absence of the problem that it was originally designed to solve. How ridiculous is it that I've still got it in there? Sigh. Now I've got to modify the documents again. I also stuck a max DEX bonus on when wearing armor, which is a relatively simple solution, and already familiar from most D&D games anyway.
With these changes, my orc fighter would have a max To Hit bonus of 23, but realistically 20 or 21 because I'm not in the habit of giving out +3 swords. My super hard to hit halfling (or anyone else, for that matter) can't seem to get his AC over 19 without magical armor no matter what he does.
I think that solves my bounded accuracy problems.
For that matter, although I didn't discuss it, it keeps my skill and save checks down too; with a max bonus of +4 plus racial bonus (effective max +3) and level check, it never gets above 17 without some kind of magical or circumstantial bonus either. This means that no matter your experience level, a 15 DC is difficult for novices but not experts, and a 30 DC is always fairly difficult for even the most advanced expert known to mankind; a 35 DC is an effective cap on difficulty for most intents and purposes.