The game that invented epic loot: Angband (1990)

Angband is a 1990 computer game in the subgenre of turn-based permadeath dungeon-crawling RPGs, also known as roguelikes. Very few people play Angband today, and most PC gamers have likely never heard of it - even when they've likely heard of NetHack and Dwarf Fortress. But did you know that the randomized equipment or "loot" so common in video games today was originally an innovation of Angband? 

Today, these "loot lotteries" go far beyond just RPGs: the same blue "rares" and purple "epics" can be seen in the mainstream from mobile games to Fortnite, Destiny, and Counter-Strike: Global Offensive. We tend to talk a lot about the principles that make these systems compelling and even addictive. But what is less often discussed is how the concept's introduction into the mainstream of commercial games came largely through the vision of a single passionate Angband fan.

Angband's items aren't entirely colour-coded by their rarity, just by category...

 

...However, artifacts are set apart by their names that glow in light blue. Might this choice be inspired by Bilbo's glowing sword, Sting?

The roots of loot

Like its predecessor Moria (1983), Angband is an unoffical Tolkien fangame, pitting the player against Sauron and Morgoth. (It got away with this by being free and open-source; don't try doing this in a for-profit game.)

[EDIT 2024-07-09] This article originally failed to name the authors of Angband, a glaring omission for which I apologize. Angband was originally created by Alex Cutler and Andy Astrand. Their work was continued by Sean Marsh and Geoff Hill, who released the first public release of the game. More information can be found here: https://rephial.org/release/2.4.fk. I thank co-creator Andy Astrand for commenting below, which prompted me to realize the omission and very belatedly rectify it. [/EDIT]

The Monsters & Treasure booklet.

In the procedurally generated halls of Morgoth's fortress, the player can chance upon a variety of items on the floor or dropped by slain enemies. The item drops are randomly determined. There's nothing unique about this in itself, of course - the same could be said for Rogue (1980) and, in fact, pen-and-paper roleplaying games starting with the very first publication of Dungeons & Dragons (1974).

From the very beginning, the methods of generating loot were quite complex. In the original D&D game's second booklet, Monsters & Treasure, each monster is given a treasure type, for example: Giant, 5,000 GP + Type E. This indicates a giant lair will contain at least five thousand gold coins worth of treasure, plus potentially other things determined by a series of dice rolls on row E of the treasure types table. The table lists chances for several categories of treasure, each of which is rolled separately: separate chances for each type of coin, gems, and, finally, "Maps or Magic". For Type E, the entry for this last category reads:

30%: any 3 + 1 Scroll

(Gygax & Arneson. Vol 2. p. 22, Dungeons & Dragons. TSR, 1974.)

This means that the dungeon master rolls a d100, and on a result of 1-30, magic items are present. If so, the DM then proceeds to roll on the magic item lists: once on the Scrolls table to determine the type of the scroll, and three rolls for the other items of randomly determined categories (swords, potions, wands, etc.).

Treasure Type E, used for giants, elves, and some of the more dangerous undead.

 
So this all adds up to the dungeon master making up to 15 rolls generate the treasure in a giants' lair. That's the worst case scenario, of course - but, on top of those rolls, some items have further details to be determined: if the scroll is a spell scroll, then the spells written on it will need to be determined, and so on...

It's quite an involved process, and has not grown particularly more complex in later games - at least as long as a human brain and a pencil are used to do the generating. (Of course, a game master is not beholden to these guidelines, and may choose to instead place items using their own judgment.)

When computers began to be used for generating dungeons, they had no such limitations: a computer can make thousands of simulated die rolls a second. Now, the greater challenge lies in designing the rules in such a way as to generate results that are interesting, varied, and balanced, even after countless repetitions. On the other hand, the computer does not have the benefit of using common sense or imagination. It cannot choose to occasionally ignore a rule here and there, scrap a nonsensical result, or suddenly create an entirely novel type of item. The designer must specify all the rules and parameters exactly.

In the same D&D booklet, rules for magic swords were described. Swords were unique among magic weapons in that they could have personalities: the sword's alignment was rolled, as well as values for its intelligence and egoism. These were numeric statistics similar to those of player characters. If the sword's intelligence was sufficiently high, it was sentient and communicated telepathically, and potentially had a number of randomly chosen special powers such as detecting magic, flight, or telekinesis.

Sentient swords with a high egoism score were able to exert their influence on the wielder, initiating a battle of wills. Mechanically, this was a contested roll using the sword's egoism statistic against the wielder's statistics, with the winner gaining control over the other. These rules for mental conflict were designed to emulate the cursed swords in literature - specifically, Elric of Mélnibone's struggles with his runeblade Stormbringer in the many stories written by Michael Moorcock.

 

Randarts: Stormbringer goes electric

So, if loot has been randomly scattered throughout the dungeon in RPGs ever since RPGs existed, what made Angband so unique in its day? How did it impact the history of video games far beyond the roguelike genre, starting with Diablo (1997) and World of Warcraft (2004), like I claimed earlier?

The key innovation of Angband was randomly generated artifacts or "randarts". This feature is arguably what made Diablo such an addictive game - and a game that printed money for Blizzard Entertainment and has been a part of their winning formula ever since.

Angband's randarts can get quite elaborate.

A weapon of Epic rarity in World of Warcraft.

 

 

 

 

 

 

Randarts took a base item, such as a dagger or a tower shield, and modified it by adding multiple properties or "intrinsics", often with further randomized numeric values, some activated powers, and so on. By laying these randomized properties on top of each other in different combinations, an effectively endless variety of equipment is generated. For example:

the Mace 'Taratol' (3d4) (+12,+12)

Slays dragons (powerfully). Branded with lightning. Provides immunity to lightning. Cannot be harmed by acid, fire. When activated, it hastens you for d20+20 turns.

These randomly named artifacts were supplemented by fixed artifacts from the Tolkien legendarium, such as the Phial of Galadriel. Additionally, there were randomly-created magic items with no proper names but which appended a suffix  "...of X" to the item's name:

Iron Helm of Seeing
Chain Mail of Resistance
Mithril Arrows of Frost

The "X of Y" naming is immediately recognizable to anyone who has played an MMORPG. Of course, many of these properties were directly copied from D&D as well. In the Angband fandom, and later roguelikes such as Dungeon Crawl: Stone Soup (2006), these properties are called "brands" (cf. "frost brand sword") for weapons, or more generally, "egos". 

I was unable to find any direct source stating that the "ego" terminology in these games has its origins in the "egoism" statistic  (in later editions, "Ego" score) of D&D's sentient weapons. But I think the lineage is pretty evident here, and this connection can probably be assumed.

Nevertheless, compared to pen-and-paper games, the huge amount of combinations that was possible through computer algorithms took the loot lottery to a whole new level, and is the secret sauce that made Diablo so popular and influential.

The devil in the design

"Having realtime combat was an unusual thing." -David Brevik

           (GI Show. Game Informer.  YouTube. May 2019.)

The other unique feature of Diablo behind its popularity is its real-time gameplay. This was relatively rare in computer RPGs of the 90s, and the hack-and-slash games that did exist were often first-person affairs such as Might and Magic VI (1998) or The Elder Scrolls II: Daggerfall (1996). However, the real-time aspect of Diablo was not present during much of the game's development: in a design document from 1994, the game is described as operating "on a turn-based system", and a complex action-points system is alluded to. The same document reinforces the randomly created dungeons as "the heart" of the game. The roguelike origins of Diablo are immediately obvious.

While the creator of the game, David Brevik, was not afraid of taking some innovative departures from those roguelike roots (such as network-based multiplayer, a simple mouse-based interface, and a business model inspired by Magic the Gathering), there was one leap he was strongly opposed to taking: Diablo would never be a real-time game. In an interview with Ars Technica, he calls it his "his line in the sand" for what the game should be. It was turn-based or bust. It took a very long time for his colleagues and publishers (Blizzard, fresh off the success of the first Warcraft game) to get Brevik to agree to even test the idea.

"I said [to Blizzard], 'What are you guys talking about, no no no, this isn't one of your strategy games. We are really commited to making a turn-based game.' I love the sweat in a turn based game, especially in a roguelike, when you've made a turn, and you're down to one hit point, and you're frantically searching through your inventory ... I really loved that tension. And they said, 'Yeah, but, you know... real-time will be better.' ... There were two or three of us that held out 'til the bitter end."

(Diablo: A Classic Game Postmortem. GDC. YouTube. May 2016.)

When he finally implemented real-time gameplay into the Diablo prototype, Brevik immediately fell in love with it. Development continued on the now-realtime game with renewed excitement, and of course players would come to love the gameplay as well. Still, it is interesting to note that unlike many other roguelike trappings with which Brevik disposed of early on, the turn-based framework was one that he was not willing to let go of.

Diablo.

In an interview with Game Informer (YouTube, around 2h37m), Brevik is asked about the colour-coded loot which has been copied into so many other games. He directly credits Angband as the inspiration:

"What about loot rarity colours? Did you guys take that chart from anywhere, or was that invented in Diablo?"

"That came from a game called Angband. Back in college, I played a lot of roguelike games - including Rogue, which is where the term roguelike comes from. There were kind of different versions of this and there was Tolkien-themed ones ... they had a version called Moria, and somebody else made a version of it called uMoria, and then that changed into this game called Angband...

Anyway, so Angband was a game I played thousands, literally thousands of hours of. It was really the game that Diablo was based on. I wanted to take Angband, that original game, and make a modern version of it.

Because this was an ASCII game, you were the '@' symbol, and you were attacking the letter 'k' ... that was what the game looked like, it was just all letters. But there were different random items that you could get in the game and they had different colours. So if you found a rare one, the text wasn't just grey, it was blue text, and that meant that it was magic! So there were some colour variations to the text - yes, we had colour text back then, this new-fangled colour text stuff... So the idea came from that, and then we kind of expanded on it, but originally the idea came from Angband."   

(GI Show. Game Informer.  YouTube. May 2019.)

These days, Angband is also available with graphics.

Blizzard had a smash hit with Diablo and its sequel, and went on to make World of Warcraft, which was played by basically everyone. The influence of those games is visible in most games made today.

So, there you have it. Way back in the '90s, before the roguelites boom of the 2010s, roguelikes were silently changing the face of mainstream game design. And it was about much more than just randomly generated levels. This example shows that there are many valuable ideas and design patterns that originate in the roguelike genre. And, there may yet be many more that could be borrowed into other genres which the mainstream doesn't know about. In the niches beneath a craggy ASCII exterior, there might still be nuggets of gold, just waiting to be mined.

The Way to Do Ascending Armor Class

Why Ascending?

As a DM who loves the OSR, you probably want to get as many people into it as possible. Thus, if there is something you can do to make the game more accessible without any functional changes, you should probably do it. For example: using ascending armor class, which the vast majority of players are more familiar with than descending armor class.

Of course, old adventure modules you will use at the table descending armor class. In this case, it's easy for you, as the DM, to look at the page and convert the number in your head. Far easier and faster, at least, than having to explain attack tables or THAC0 to each new player.

The Full Score

Conversion is as simple as subtracting the old AC or THAC0 value from your conversion base, which is a fixed number. So what do you choose as this base? If the choice can make the mental arithmetic even a tiny sliver easier for you, then it will save a lot of time in the long run. I find that a round number like 20 is slightly easier to subtract from.

So your conversion formula would look like this:

Desceding AC = 20 - Ascending AC
Attack Bonus = 20 - THAC0

Thus, if you're reading an old module and see a monster with 5 AC, you tell the players that it has 15 AC. Monster attack bonus is an even easier case: you don't even need to check the attack table if you can memorize how monster HD relates to attack bonus. In our case, that is exceptionally easy - monsters have an attack bonus equal to their HD! (At least up to 8 HD - the B/X attack progression slows down at very high HDs.) That means that monsters with less than 1 HD have +1, and so on. "Normal men", which I like to think of as "level 0 types", have an attack bonus of +0.

Very clean, right? It's kind of intuitive that all 1st level PCs attack at +1, and thus all "normal humans" - those who haven't gained a class - attack at +0. Anyone with a class attacks at +1 (you can think of 'monster' as a class).

If you chose another number as your base, say 19, then level 0 types would attack at -1? That's weird.

So if 20 is such a good base, surely most modern AAC-using retroclones use it? As it turns out, no.

The Big Comparison

Swords & Wizardry offers both descending and ascending AC. Old-school descending AC and attack tables are the same as B/X - unarmored AC is 9. In ascending AC, unarmored is 10. This means that to hit an opponent with 20 AC, a 1st level Fighter needs to roll a 20 - as if her attack bonus was +0. So does that mean normal humans have an attack bonus of -1? The book doesn't say. Is the authors's sole reason for choosing 19 as the conversion base that an "unarmored base AC" of 10 looks pretty? If so, I don't think that's a very good reason.

B/X Essentials only includes descending armor class, but its upcoming refinement Old-School Essentials adds an option for ascending armor class. It sets unarmored AC at 10, taking its cue from Swords & Wizardry.

Lamentations of the Flame Princess goes in a completely different direction. There, unarmored characters have 12 AC (ascending). In other words, the conversion base is 21 (21-9 = 12). To match, Fighters attack at +2. But all other PC classes only attack at +1, so they've been demoted to "normal humans" - while NPCs themselves attack at +0! Overall, everything is harder to hit, to suit the game's greater emphasis on horror. I actually think this is a bit of a trap for new players, who get recommended LotFP as the "default, generic" OSR game, but are faced with a lot more frustration than anticipated. Buyer beware!

Basic Fantasy RPG, for one, does it right. PCs have an attack bonus of +1, normal men of +0, and unarmored AC is 11! Hooray! For everything BFRPG does that I disagree with, here is one part where they knock it out of the park: designing with the actual use case (on-the-fly conversion) in mind, rather than for a surface appearance of elegance.

I don't think anyone cares if unarmored AC is a nice round number - when it goes on the sheet, it gets modified by their equipment and Dexterity anyway. Basic Fantasy: 1, Swords & Wizardry: 0.

Of course, if you use the above system and play a S&W module that lists both DAC and AAC, you will need to either use the DAC for conversion, raise the quoted AAC by 1, or use the AAC as-is and be happy with the fact that monsters are nerfed by 1 AC, which is probably perfectly fine.

(Another tiny caveat: the above only talks about converting B/X and OD&D materials, where unarmored is 9 and Fighters have THAC0 19. AD&D actually changes these, moving unarmored to 10 and giving Fighters THAC0 20 to match. If you use the above to convert the AC values from an AD&D module, you will end up nerfing the monsters by 1 point. And that's probably perfectly fine too.)

Making the case for limited ability modifiers in OSR

D&D B/X is, I would wager, the most popular basis for OSR games. Therefore, B/X style ability modifiers are the most common. In case you didn't know, ability modifiers in B/X range from -3 to +3 and are determined like this:

Score	Mod
3	-3
4-5	-2
6-8	-1
9-12	+0
13-15	+1
16-17	+2
18	+3

Nice, unified and symmetric. Except of course when it's not - only positive modifiers affect languages known, and reaction adjustments from Charisma only go from -2 to +2, and ditto for initiative adjustments from Dexterity. But we'll get back to that.

While this is probably the smoothest system in any official edition - more unified than Original and Advanced D&D's jumble of adjustments, and less game-deciding than WotC D&D's gigantic -5 to +5 range, they're not the best choice for everything.

Negative modifiers can be problematic when you want to modify numbers that are naturally quite small. Rulesets will often include things like “you can hold your breath for a number of minutes equal to 1+CON”, which will then require special stipulations and minimums for negative modifiers.

Also, I have witnessed applying modifiers to dice rolls cause headaches to OSR hackers creating their own games using B/X as a starting point (often via Lamentations of the Flame Princess). ‘Esoteric Enterprises’ and other games by Emmy Allen use X-in-6 skills like LotFP, except ability modifiers are applied to skills - though they can never go below 0-in-6 (rolled as a 1-in-36 chance). Since skills start at 1-in-6, this means that a modifier of -3 or -2 is the same as -1, not to mention that a +3 is pretty massive. The WIP post-apoc game Ruinations by Brent Ault has gone through several iterations with its skill system, likewise attempting to include ability modifiers into skills, but trying to dampen their effect. At one iteration of the ruleset, skills were moved up to the d12, starting at 2-in-12, plus ability mods. Once again, anything below a -1 is not accounted for. The skill system was changed in a later version to a d100 where start skills at 20%, and have each point of modifier count for 5%, so a -3 modifier would give a 5% success rate. Which, you'll notice, is mathematically the same as a d20 roll with the modifiers applying in their usual way.

Basically, the d6 skill system is liked (by me and many others) for its chunkiness - adding a pip to a d6 feels much better than adding a handful of points to a d20 - but big modifiers and big chunks don't mix.

I'm going to suggest something to all hackers, tinkerers and homebrewers right now:

Ditch the negative ability modifiers from your game. Completely.

It's okay. Just because you're using B/X as your engine doesn't mean you NEED to have the same ability modifiers. It doesn't break compatibility. You still have the same scores, in the same range of 3-18, for when you take ability damage or whatever. You can still run Keep on the Borderlands even if there isn't some unlucky geezer running around with a -2 DEX. You still have HP, and XP, and AC. You can still use all the great TSR and OSR content out there exactly the same.

Then, squish down the modifier ceiling to one that you think won't break your maths too much.

Yes, it may be somewhat more fun for players to have wide variation in characters' abilities, and amusing to laugh at the one chump with a big negative modifier. But reducing modifiers to a range of, say, +0 to +2, opens up a lot more design space for a homebrewer. Never again will you have to worry about special stipulations when applying modifiers to a base number of 1.

(Btw, this pairs quite well with the static Health/Wounds mechanic I talked about in an earlier post - I'll just have CON modifier increase your starting Health at level 1, but more on that in a later post.)

As for precedents, there are already places where B/X et al. restrict adjustments from abilities to -2 to +2, like reactions, initiative, and XP adjustments - because a +3 would be far too large a modifier on a 2d6 roll, for example. Why not expand these limits to everything, thus truly unifying ability adjustments? A smaller range of ability modifiers that excludes negatives means doing less maths, a larger design space, and fewer special cases.

I propose the following modifiers, and will use them in my next game:

Score	Mod
3-12	+0
13-15	+1
16-18	+2

Essentially, it's the B/X range but with negatives completely removed, and +3 squished into +2. You still have some characters (21.3%) who are very good at a given thing, and a few (4.6%) who are exceptional at it. The rest (74.1%) are just average. And that's okay. Now, your base Bushcraft (or whatever) chance will be either 1-in-6, 2-in-6, or 3-in-6. No special stipulations.

By the way: in the oft-referenced late Gygaxian houserules for OD&D, abilities modify things by +1 or not at all. And they modify very few select things. Constitution of 15 or more gives +1 HP per HD, and so on. In discussions of these rules it is often pointed out that they were made for convention games, and therefore do not represent how Gygax ran the game at home. It is true that many of the changes there improve PC survivability - which makes sense when running a quick convention game. However, the streamlining and restricting the effect of abilities does not necessarily improve PC capability - and in fact set a lower ceiling for it than those in B/X and AD&D. I'm not going to say "if it's good enough for Gygax...", I'm just including it for completeness and to show that it's okay to do things differently to B/X - as long as compatibility is preserved. Compatibility must always be preserved.

D&D Without Damage Dice

A contest on G+ to come up with a new way to speed up combat in D&D caught my interest, two responses in particular.

The first was Cavegirl’s one roll fights. It reminds me of the War Machine mechanic from Rules Cyclopedia, and really drives home the idea that battle is a risky proposition that you should not undertake without heavily stacking the odds first. Personally, if I were to hack it, I might turn it into two rolls instead. Players could see the result of the first roll, and then get a chance to affect their fates with spells and such, or simply run away. Then, winning two rolls would mean an overwhelming victory, winning only one roll a victory at a cost, and losing both rolls an escape-or-die situation.

The second interesting idea in the thread was replacing HP with HD, that is, you can take a number of hits equal to your number of hit dice. One hit from any weapon is... one hit. There would be no damage rolls, which certainly should speed up the game and reduce calculations and bookkeeping... I couldn’t stop thinking about this idea. So I grabbed my trusty spreadsheets and tried to hash it out in detail.

Note that this is not playtested yet, just theorycrafting so far. I would quite like to playtest this at some point if I can iron out some of the issues, but that may require additional hacking of ability scores.

Working out the number of hits

First, let’s look at how many hits a PC can take under current rules (using the gold standard of B/X). On average, a 1st level Fighter, having d8 HP can take 1.29 hits from a d6 weapon before going down. Magic-Users can take 0.71 hits, and clerics 1 hit. These averages increase linearly by level.

Here’s how many d6 hits a B/X character can take on average under the default rules, rounded up.

Level	MU d6s	CLE d6s	FTR d6s
1	1	1	2
2	2	2	3
3	3	3	4
4	3	4	6
5	4	5	7
6	5	6	8
7	5	7	9
8	6	8	11
9	7	9	12
10	8	10	13
11	8	11	15
12	9	12	16
13	10	13	17
14	10	14	18

So that is the rough guide we should look at when designing this new damage-roll-less system, in order to approximate the same level of PC survivability. Throughout this article, I will refer to these new survivability points as Health, to avoid confusion with the normal Hit Points. Instead of damage, attacks will deal Wounds. Taking one Wound is the same as losing one Health. A d6 of old HP damage translates to one Wound in the new system.

So why are we looking at d6 damage? d8 damage, after all, is quite common for a one-handed weapon of war - there’s not much reason for a Fighter to use a handaxe instead of a battleaxe. But remember that variable weapon damage is an optional rule in B/X. We are looking at the d6 because that’s how much damage all weapons originally did. As for monster damage, looking quickly through the list, the d6 seems to be most common die, although there are a lot of d8s (and “as weapon”) too.

We are faced with a choice of two options for this new system. Either we base player Health off the d6, which is in line with non-variable weapon damage. This makes monsters die more easily (29% more), since we’re pretending they used a d6 hit die before when in actuality it was a d8 (well, at least after variable weapon damage was introduced - before that it was a d6). The other option is to base everything off the d8, which makes everyone about equal in relation to B/X, but greatly boosts d4 weapons and smaller. Perhaps we should err in favour of the players here.

So, we have all weapons dealing 1 Wound on a hit, and monsters having Health equal to number of HD. The straightforward way to work out player health would be to use that d6 table above straight. Essentially, players would have 1 Health per d6 (3.5) hit points, rounded up. Fighters would start with 2 Health, while Clerics and Magic-Users would start with 1 Health.

However, looking at only averages results in a very poor fit for low levels. While it works out to the same results on a high number of repetitions, at low levels repetitions are low and you’re cheating players out of a lot of chances to survive. For example, when taking a hit from a d6 weapon, a 1st level cleric (d6 HP) has a 42% chance of surviving. Even a Magic-User has a 25% chance of surviving, while the Fighter has 56%. Generally, you have slightly worse than even odds to survive a weapon with a damage die equal to your hit die at 1st level (because a tied roll still brings you to zero). See the chart below.



Apparently, the original reason for the creation of the granular hit point system in Arneson's games in lieu of the "one hit = one kill" system from Chainmail is that players didn't like it when their characters died too suddenly:

"It meant that players had a chance to live longer and do more. ... They didn't care if they could kill a monster in one blow, but they didn't want the monster to kill them in one blow."

(So, it might be that we're moving backwards with this experiment and there are very good reasons damage rolls are standard... but let's ignore all that for now.)

The simplest solution to the “level 1 problem” would be to have all classes start at their Health values of level 2. They would then not gain an additional Health at 2nd level, but follow the progression according to the above table thereon.

However, since we will have to make a whole new Health table anyway, this is a good opportunity to tweak the progression. We can use an entirely new equation, unshackled from the idea of one die size and one die per level. This way, we know where our numbers are coming from, and we’re not just blindly reimplementing artifacts that arose from the restrictions of the original system (i.e. the shapes of dice that were available to the original designers) while compounding them with new restrictions. And gaining 2nd level is the first time players get to experience leveling up, so it should feel like something! So, here’s an alternate progression you can use, based on a linear factor multiplied by level.

Alternate progression

Health = 1+YxLvl, rounded up. Y factor: M-Us 0.55, Clerics 0.82, Fighters 1.12. These progressions follow the original progressions the closest at mid-levels. They start out better at low levels, and fall off at high levels.

Level	M-U Health 1+0.55xLvl	CLE Health 1+0.82xLvl	FTR Health 1+1.12xLvl
1	2	2	3
2	3	3	4
3	3	4	5
4	4	5	6
5	4	6	7
6	5	6	8
7	5	7	9
8	6	8	10
9	6	9	11
10	7	10	13
11	8	11	14
12	8	11	15
13	9	12	16
14	9	13	17

Comparing survivability with the old HP system

Fighter:

Cleric:
 
Magic-User:


I think it looks quite good. A bit more survivability at low levels is not a terrible thing, in my opinion. Some interesting artifacts: Magic-Users do not gain any Health on levels 3, 5, 7 and 9, when they gain powerful new spell levels. Between levels 5 and 6, which is when Clerics suddenly gain two new spell levels in B/X (though not in Labyrinth Lord), they do not gain any Health. There is also a remarkable resemblance to the hit dice progression from the OD&D (before the Greyhawk supplement), where every class (and monstes) had a d6 hit die but gained them at different rates - Fighting-Men gained a die per level, Clerics gained one on almost every level, and Magic-Users gained one every other level.

Constitution modifiers

Now that we have our Health progression, we simply need to modify by Constitution. We can say that each +4 HP from Constitution (round up) is about equivalent to one additional hit die (and, consequently, Health). Thus, a +1 modifier should grant an additional Health every 4 levels, a +2 grants an additional Health every two levels, and so on. A +2 modifier should grant an additional Health every two out of three levels. This means that for a max level Cleric the gain from Constitution is actually slightly lower than the original HP system (1.77x vs. 1.86x). Once again, this formula improves things for players at low levels but makes it slightly tougher at high levels.

	Constitution Modifier / Health Adjustment
Lvl	-3	-2	-1	+0	+1	+2	+3
1	-1	-1	-1	+0	+1	+1	+1
2	-2	-1	-1	+0	+1	+1	+2
3	-3	-2	-1	+0	+1	+2	+3
4	-3	-2	-1	+0	+1	+2	+3
5	-4	-3	-2	+0	+2	+3	+4
6	-5	-3	-2	+0	+2	+3	+5
7	-6	-4	-2	+0	+2	+4	+6
8	-6	-4	-2	+0	+2	+4	+6
9	-7	-5	-3	+0	+3	+5	+7
10	-8	-5	-3	+0	+3	+5	+8
11	-9	-6	-3	+0	+3	+6	+9
12	-9	-6	-3	+0	+3	+6	+9
13	-10	-7	-4	+0	+4	+7	+10
14	-11	-7	-4	+0	+4	+7	+11

The downside is that on each level up, players will have to reference both their class’s Health chart and the Constitution Health adjustment chart and sum the values. Then, because they progress at different rates, it is actually possible for them to lose Health on level up with a -1 or -2 modifier. That's obviously a huge flaw in these steps. So there needs to be some stipulation that your maximum Health can never decrease on level up. It gets rather messy and inelegant.

Constitution is the biggest obstacle I currently see to the new system. All this gives me half a mind to just get rid of the -3/+3 modifier range and have them go from -2/+2. Abilities shouldn’t matter overmuch, after all. Alternatively, one could have reduced Health gains from Constitution (and penalties which never overtake Health progression) and make Constitution important for other things, like dying/bleeding out.

Monsters and spells

Monsters will have one Health per HD. Thus a PC hitting a goblin or skeleton always kills it, no damage roll required. Functionally, this makes them similar to the famous Minions from 4th edition. Monsters with a plus on their HD will be treated as one HD higher for purposes of Health.

All sources of damage do one Wound for each damage die (d4 to d12). Each +4 counts as one additional die. For example, a trap that does 4+2d4 damage will deal three Wounds, and an attack that does 1d10+1 damage will deal one Wound. Spells which do half damage on a successful save will only do half Wounds. For example, a fireball spell has 8 damage dice so it will do 8 Wounds on a failed save, or 4 Wounds on a successful save.

The one exception is when halving the damage dice (e.g. due to resistances) would leave half a damage die, or when an attack would deal damage in an odd number of d3s. These are called “half-dice”. A half-die will have a 50% chance of doing no Wounds and a 50% of doing one Wound. For example, an attack that dealt 3d3 damage will now deal one Wound, and have a 50% chance of dealing a second Wound. Or, you hit a creature with a torch for 1d4 fire damage, which would usually be 1 Wound, but if the creature is resistant to fire, you only have 50% odds of dealing a Wound on a successful hit. On-hit effects are still applied regardless of damage. For example, a ghoul’s claw is d3+paralysis, so the paralysis is applied on every hit, even though damage is only applied 50% of the time.

(Also, to prevent everyone just using torches as weapons instead of swords, you may want to give them a reduced damage chance too, or say they’re only an effective weapon vs. specific enemies such as trolls and mummies).

Weapons

What about two-handed weapons? Differentiating them will be dealt entirely with Fighter abilities. When using shields, Fighters can sunder them to negate damage. When using two-handed melee weapons, Fighters can attack again each time they score a kill. Greatweapons might also be allowed to overcome the slashing/piercing resistance of skeletons due to their large mass. Additionally, Fighters will deal an additional Wound on a natural 20.

All this leaves crossbows in a somewhat strange position - they shoot more slowly but deal the same damage as everything else. I may reinstate the armor piercing from LotFP there. Crossbows will fire every other round and pierce 2 AC for light crossbows or 4 AC heavy crossbows.

Unarmed fighting: If an attack roll with an unarmed strike succeeds, players need to roll to see if it deals a Wound. Since unarmed strikes had a d2 damage die before, their odds of dealing a Wound are now 2 in 6, plus 1 in 6 for each point of Strength modifier. Perceptive players will notice that this is the same chance as Open Doors rolls in B/X - perhaps we can rename Open Doors to “Smash” and use it for both. We could even use this chance for improvised weapons (e.g. torches) and treat them the same as unarmed strikes.

To Recap

• Health will replace HP. Player characters gain Health according to new class tables. Monsters will have one Health per HD, plus one Health if they have any pluses on their HD.
• Damage is not rolled, instead damage sources will deal one Wound per die of damage, plus one Wound for every 4 fixed damage. 
• d3 damage sources and odd dice halved due to resistance will have a 3-in-6 chance of dealing one Wound. 
• Open Doors chance equals the chance to deal damage on a hit using an unarmed strike or improvised weapon.
• Effects that refer to the target’s Hit Dice (such as Sleep) will use levels instead when the target is a PC.

Advantages:

• No rolling monster HP.
• Less tracking/HP maths on the fly - Health values are in the single digits for most of the game. 
• No rolling damage. 
• Fewer dice required (players only need d20 and d6). 
• Increased survivability at level 1. (YMMV whether this is a good thing.) 
• Doesn’t break compatibility with existing material.

Disadvantages:

• Less granular damage modifications: STR, magic weapons…
• Smaller design space for abilities. 
• Some players may find less rolling less fun. 
• Things become more static and predictable. It’s easier for players to tell when they should retreat. Players usually want to retreat when they have 1 HP, but feel like they have a decent chance at 4 HP.
• Applying Constitution modifiers becomes more complex.