Ability Score Improvements

ASI with no balance issues whatsoever

It is my uncontroversial opinion that as a character gains levels, that character should improve. This is already the case in AD&D for saving throws, attack rolls, proficiencies, and spells, but it excludes Ability Scores and the values derived from them. PCs can become better combatants mechanically, but they never become stronger or more agile. They can improve in the mysteries of their religion or the finer points of etiquette, but never become wiser or more charismatic.

But, do they need to? In modern D&D, the ability score is the number from which you derive all other modifiers, so of course it must improve. This is not the case for early D&D, wherein the scores were descriptive markers, something that let you know what kind of class you should play based on the gated prime requisites. Any bonuses for high scores provided only moderate benefits, if any at all. Do characters really need to improve their base stats when increasing Dexterity from 13 to 14 does absolutely nothing?

Well, yes. I think so. Even if a game like AD&D doesn’t have a mechanical need for Ability Score Improvements, or ASI, it certainly has a psychological one. These are improvements to numbers created by the player, rather than the tabled stats that they’ve been handed, and therefore provides a stronger sense of advancement when the player gets to make ‘their’ numbers better, especially since those values are the most prominent and primary entries on their sheets. Of course, it can have mechanical impact, too. If the DM is making good use of ability checks, a single point to any score is a 5% increase toward success. If the DM has been creative with house rules and put some bite into the ability score bonuses, there’s even more of a reason to allow ASI.

However, if you’re running any version strictly by the book, then the answer is no.  As a result, searching for options to implement ASI within TSR’s published rules unearths very little. There’s the option for adjusting scores during generation found in OD&D and B/X, the age adjustment table in AD&D, wishes (with a ruleset that shows how much Gary’s players were abusing wishes), and magical items. Aside from chargen options, these are all DM-granted improvements.

The only time you see active improvements chosen by the players themselves is in Unearthed Arcana, where the Cavalier class can invest into scores in a similar manner to the Exceptional Strength feature for Warriors. Given my distaste for that edge-case percentile barnacle, I’ll reject that as an improvement mechanic and keep it far away from the other ability scores.

Get out of here, you weirdo.

 

I even went to the Player’s Option books to see what crazy ideas the TSR crew might have been cooking up in the heady days of ’95, but there are no ASI rules found there. However, they did add sub-ability scores in the effort to make the game that much more cluttered.

So we’ll have to house rule it. There are a couple of options.

You could implement structured improvements with a basic +1 bonus at specific levels, such as is done in modern D&D. Players pick two scores and bump them every four levels, or whatever schedule you think best, perhaps even based on class. This is reliable and easy, though these bumps are all but guaranteed to get dumped into prime requisites. PCs under this improvement method will suffer from a curse of sameness as they all invest into the ‘good’ abilities that their class (or your game) needs. Mythlands does a good write-up on this.

You could instead make the players roll for it. At scheduled levels, the player can roll a d20 against one or two scores in the manner of an ability check, but this time attempting to roll at or above their score. If they succeed, the score goes up by one. If you wanted to make the increase of bad scores easier and high scores more difficult, you could use 3d6 against the ability, which also has the added benefit of aligning with the dice used to originally create the score. With either option, this pulls back on the power creep and incentivizes players to improve poor scores rather than hyperfocus on minmaxing their elf.

Of the choice between static and random, I’m clearly a fan of the latter. Still, the auto-improvement (even with a chance of failure) isn’t great. PCs in AD&D should not and do not ‘ding’ upon level up. Therefore, let’s add in a feature of standard leveling: training.

 

When leveling a PC, the DM is already provided roleplay tools (mandated in 1e, optional in 2e) to ensure that both time and treasure are spent in pursuit of improvement in their class. This can be extended to ASI. If a players wants to increase their poor stats, they need to find and pay for a teacher, perform a task or quest, or generally come up with some good description for the way in which they’re investing time into that ability score. In the same way that high level characters require high level trainers, the effort to boost Charisma from good to great means that PCs will be hunting for the most charming person they can find and shelling out precious resources to do so. Another lever is thereby granted to the DM in the pursuit of keeping players engaged and PCs poor, while the players themselves are given agency in the upgrade of their characters.

ASIs are not a slippery slope to overpowered or boring PCs. They’re an improvement path that can reward roleplay while giving players a chance to build up their most personal scores. As evidenced by Gary’s very reactionary rules regarding the use of wishes to increase ability scores, it’s clear that players have wanted to increase these scores since the earliest days of the game. I’m with them on this one.

Ability Scores and Checks – Part 2

I3-5: Desert of Desolation, 1987

In Part 1, I did my best to explain why the ‘push-button’ argument is crap, but it’s not the only concern people have with the Ability Check mechanic. Three other objections were mentioned.

1. Characters with very high scores reap all the benefits, while mid-tier scores see increased failures.

Yes, that’s the design. A higher ability score means a higher chance for success. If the PC has a 10 in any ability score, then they fall in the middle of the range, equivalent to a normal person. They should only have a 50/50 shot at difficult tasks that require that ability. If the task isn’t something a normal person could fail at half the time, then the check needs modifiers, the ability score target needs adjustment, or the roll doesn’t need to occur.

By providing a -5 to +5 modifier to the roll, you can swing that chance by 25%. As stated previously, this provides a chance to ask the player how their character is accomplishing the task, and to then reward that. It grants a lever to increase intelligent play. If you really need to ramp up the difficulty, you can halve or quarter the score. This drops the chance precipitously, but still recognizes the impact of the PC’s score on the task at hand. And you can still modify the roll! With these adjustments, you can reach a wide range of chance, and the character’s score serves as the primary source of that final value.

Once you’re in need of chances between 95% and 100%, it’s time to just let the PC have the win; their score itself serves as the evidence of their success. If you’re looking for chances below 5%, you’re again being too granular, and are best served by granting the character a minimum floor of their 1 in 20 chance. It’s simple and allows the surprising to happen more regularly, which is always a good time. After all, you asked for the roll, so there’s a chance, right?

Your players when you tell them how the modifiers affect the roll.

 

Speaking of that.

2. A score of 3-18 only spans 16 values on a d20, PCs always have a 15% chance of success, while a PC with a score of 18 always has a 10% chance of failure.

I think we have the solution in the text above. Modifiers and score target adjustments allow us to circumvent these minor caps. Still, if you’re calling for a check, it implies a chance of failure; a 1 in 10 chance of failure at the peak of one’s ability would appear to me appropriate when fully testing that ability. I see it as a feature rather than a bug.

Mentzer suggests an alternative approach, Basic 1983

 

So I’m not swayed by the prior points, but the last one has teeth:

3. Ability scores never improve, which locks in those mid-tier failures for the life of the PC.

Yeah, this is absolutely fair. In AD&D, the PC’s ability to hit and their saving throw values increase as they level up based on their class. Their core ability scores, on the other hand, are permanent outside of a magical item or a wish. As a result, the scores never provide a sense of progress in the same way that other rollable stats do.

A common solution is to just use saving throws instead of ability checks. I see the cleanliness of using a pre-existing mechanic to solve the issue, but I think it’s a poor substitute. First, that’s not what saving throws are for. Saving throws are meant to be reactionary. Using them proactively to succeed goes against their design of avoiding disaster, which can work but muddies the waters. Secondly, they often don’t match to the desired action. If our chandelier-swinging thief uses a saving throw to determine whether they look awesome or fall hysterically, what’s the check? Death magic? Breath weapon? Petrification? It’s not intuitive.

We’ll look at solutions in the next post, and figure out if we even need one.

Ability Scores and Checks - Part 1

AD&D 2e Glossary

The ability scores of D&D have remained consistent from ’74 onward, albeit the final reordering only occurred in 2e. How one uses these scores for adjudication has evolved with every iteration, and I won’t waste time relating a history that has already been well researched by others.

Ability Checks, wherein you attempt to roll under your ability score with a d20, were still considered nonstandard by 1989. 2e shoved them into the glossary and begrudgingly noted that you can use them as saves. Yet, the Nonweapon Proficiency system of 2e uses ability checks as the base for that mechanic; when you’re rolling a Navigation check, you’re just rolling an Intelligence ability check with a -2 penalty to the ability score. It’s as messy as its ludicrous name suggests.

Chatter on the old forums makes it clear that ability checks were common at many tables due to its utility and simplicity: roll equal to or less than the number on your sheet, and we’re done here. No math on the player side beyond a bonus or penalty given by the DM (or for particularly difficult tasks, halving the ability score). Both players and the DM can immediately calculate their chances of success; the DC is right there. It’s easy to call for when needed and easy to categorize most actions into one of the six abilities.

That same chatter evidences that others most certainly did not use this semi-rule. They note that it is overly beneficial to high stat characters (giving an 80% chance of success to a stat with 16), while being functionally useless to anyone with an average score (50% chance for a stat of 10, aka: a coin flip). Conversely, it creates a ceiling of a 90% chance, where even a PC with a max score of 18 will fail 1 out of 10 times.  Since ability scores do not increase with level, there’s no improving this percentage, leaving a level 14 character exactly as good or bad at any given ability as they were at level 1.

These objections are worth discussing, so we’ll come back to them later.

The main objection of the grognards, as with any skill-like system introduced to ‘old-school’ play, is that it replaces descriptive problem-solving in favor of push-button solutions. Why attempt to offer any details on how the PC talks to the guard or searches for the hidden clues when players can just ask to make a check? Why think about solutions when you have a big button that solves any problem with a 75% or whatever chance of success? This mechanic will be the death of thoughtful and creative play!

Pictured here: Moldvay ruining D&D as early as the '81 Basic Set

 

I find this last argument to be nonsense for a few reasons.

First, push-button solutions are everywhere in AD&D. Need to lift this portcullis? Bend bars/lift gates. Open a stuck door? Open door. Oh, hey look, they’re just modified Strength checks. Then there are thieving skills which are explicitly push-button; for the thief, they’ve got a Hide In Shadows hammer and everything looks like a percentile nail. If we dig into the spells, aren’t they a bit push-button? After all, you don’t have the wizard describe the intricate details of Magic Missile; it’s fire and forget, literally. Hell, attack rolls fit this definition.

Second, what’s the alternative mechanic for determining success for an activity that has some chance of failure? A common answer suggests eyeballing the chance of success, then calling for a roll with a d6 (à la secret doors, and don’t get me started on that vestigial search mechanic) or using a percentile roll. Okay, what are you basing that chance on? If the fighter is attempting to push a boulder, you would take the PC’s Strength into account. If the thief is attempting to swing off the chandelier, you’d consider their Dexterity. If you’re making a call on their chances without considering the ability score, that’s just arbitrary and it invalidates a core part of what makes that PC who they are. If you’re making that call with the ability score in mind, why not just use the damn score and apply a modifier? It’s right there.

Third, if exploration and social interactions are spoiled because players are using ability checks like they’re Hit X to Skip, that isn’t their fault; it’s a DM issue. This is AD&D we’re talking about, right? It’s the DM’s game. As DM, you choose the mechanics of how your players interact with the world. You’re the one calling for the checks! There is nothing stopping you from using ability checks to reinforce world-building and emergent play. Ask the player how the character searches/talks/whatever, then let them know how their decisions impacted the modifier you give to the ability check. If they’re asking for the check themselves, judo them into explaining how they want to use it, and reward clever responses with good bonuses. If the PC has a good enough stat that modifiers obviate the chance of failure, then don’t call for the roll; instead, congratulate them on their ingenuity and let them do the thing they were trying to do.

Sure, but how do you pay respects?

 

The argument that ability checks (and skill checks) make players dumber does not fly with me. Players aren’t pushing a “solve problem for me” button, they’re pushing a “have my character interact with the world” button. They’re directly interfacing with the game through the lens of the PC they’ve created, and using the ability score – the first thing they generate when making a character – to do so. Honestly, what else are ability scores for? They should have more meaning than just providing two dozen small modifiers to varied subsystems.

You don’t have to pick between smart play and simple, sheet-based mechanics. You can have both. Anyway, it lets the players roll the dice more and they like that.

We’ll look at the other problems in Part 2.

To Hit Descending AC

In the last entry, I discussed why I like descending AC and some of its advantages. I left the disadvantages for later, and so we have arrived.

My argument is simple: the bad parts of descending AC are the complicated ways used to determine how one beats the numbers, and ascending AC is therefore identified as the superior method only because it resolves the annoying math. However, I do not believe that one necessarily follows from the other; you can have descending AC without the complications. Before we get to that, let’s run through some of these methods and see where they cause friction, starting with the two methods that were used during AD&D’s run: To-Hit tables and THAC0. 

To-Hit tables were there from the start; the linked blogs concerning its genesis (found in the last post) are a recommended source to understand why the tables were built this way. The use is straightforward: find the monster’s AC and the PC’s level in the table headers, then cross-index to get the required value to hit. The 5th level Fighter is attacking an enemy with AC 5? Look at table I.B. on page 74 of the DMG and find that an 11 is needed to hit. The player rolls, adds any bonuses, and says, “I rolled a 12.” The DM says, “You hit.”

This is dead easy for the player, but compared to modern approaches it’s work for the DM. Instead of having to keep just a single number (AC) in their head and compare it against the player’s results, the DM has to reference a page of tables (or jot down the required values, for each level and class in the party, prior to play). Whichever way you look at it, having to reference a table each time a player smacks a goblin isn’t what I would call an elegant solution, though it was a very familiar state of play for early designers and DMs.

The state of play in the early '70s (Chainmail)

 

THAC0, or To-Hit Armor Class Zero, is a house-rule and cheat for resolving this lookup issue, made official in 2nd Edition. Since the original tables don’t change, their math can be worked backwards to give players a single number with which they can determine hits. Take the THAC0, subtract the AC of the monster, and ostensibly you’ve got your To-Hit number without any table lookups or fuss.

But of course, there is fuss. First, who does the math? If it’s the DM, then we’re just substituting one table for another, now with subtraction! If it’s the players, then are we giving out the monster’s AC? Online discussion suggests that it was common practice for players to subtract their attack roll from their THAC0 and announce which AC had been hit, though the original books don’t provide this particular formula. Even this is a kludge: you’re now asking players to roll high, add modifiers, and subtract two-digit numbers (which are easy to accidentally transpose) to reach a low target. Regardless of how much you can get used to it, this is why people talk about using THAC0 in the same way you’d describe using a slide rule.

Well, this is from 1e, but you get the idea.

 

Okay, those are the official ways to hit descending AC. What has the OSR developed? As noted in the prior post, most of them ditch it entirely, but for the few that keep the numbers heading toward zero, there’s little change. Games which hold to the original definition of OSR (i.e., cloning and cleaning up old-school D&D and AD&D) still stick to the tables or THAC0, while the “do whatever feels old-school, man” side of the OSR isn’t participating.