​The Exhaustion system in D&D 5th Edition is terrible.

That's not a hot take, and is in fact a pretty generally accepted opinion, but today I thought I'd take a look at why it's so bad, and what could be done to fix it. So: what's wrong with exhaustion?

1. The penalties are weird
As much as I love raw mechanical systems, flavour tie-in and things making intuitive sense is essential, particularly in a RP game like D&D. The penalties levied at different exhaustion levels aren't really related to each other. Levels 2 & 5 are movement-related, 4 & 6 are HP-related, and 1 & 3 only seem similar to each other until you note that ability checks are mostly non-combat while attacks and saves are almost exclusively combat-related. These penalties being so distinct means that at any point, one more level of exhaustion could be either crippling or irrelevant, dependent on the task at hand.

2. Penalties scale differently for each class
The penalties affect each class very differently. For example, most wizards really won't care at all about the first two exhaustion levels—ability check disadvantage & halved movement—and can work around the third without too much trouble by using spells that target an enemy's saves rather than requiring an attack roll. Meanwhile, one exhaustion level can seriously affect a skill monkey's ability to do things outside of combat, or a grappler's ability to do things in combat, and two exhaustion levels may cripple a Barbarian or Paladin, classes that are only effective at close range.

3. It's part of the very half-baked mechanical support for exploration
On page 8 of the Player's Handbook, the three pillars of D&D are listed as Exploration, Social Interaction & Combat. Of those three, the mechanics of 5e are very heavily combat-centric, and while there is some mechanical framework for social interaction, exploration for some reason got completely shafted. Now, my problems with this are numerous, and I'm sure I'll have a proper gripe about 5e's lack of exploration support at some point, but for now my point is that exhaustion is a mechanic largely built for exploration/survival, but without any actual exploration/survival systems it just sits there unused, orphaned by pretty much the entire rest of the book.

Between the PHB and the Dungeon Master's Guide, exhaustion is given as a penalty for extreme temperatures, frigid water, starvation, dehydration, and travelling at a forced march. Xanathar's Guide to Everything adds exhaustion from long periods without a long rest, and temporary exhaustion from the spell Sickening Radiance. There's a single enemy from Princes of the Apocalypse, and to my knowledge that's pretty much it. No other spells, no monsters, no abilities... except for the PHB's first Barbarian subclass: the Berserker.

Berserkers can take a level of exhaustion to make attacks with their bonus action for the duration of a single combat. Cool, fitting, powerful. However, in additon to the fact that the other subclass is the famed Totem Warrior, who can extend your resistance from physical to all damage, the Berserker really isn't an appealing subclass because exhaustion is so punishing. If you're gaining a level of exhaustion each time you use that ability in combat... that leads us to Number Four.

4. Exhaustion can only be removed by a long rest, and only one level at a time
(or by the 5th-level spell Greater Restoration, which you can't get until 9th level, is only available to three classes, and is a pretty expensive trade for the Berserker ability). That's it. Those are the only ways to remove exhaustion, which means for practical purposes a Berserker can only gain their extra attack in one combat per long rest, is a mediocre grappler at best, and usually has disadvantage on ability checks, discouraging the player from participating in half the game.

So, how would I change exhaustion?
Exhaustion is one of the most frequently house-ruled or hand-waved parts of 5e. I've seen a number of house-rules, most of which seem to focus on adding things that cause exhaustion or changing the Berserker penalty. Here's my rework, redefining exhaustion with hit dice—another underused mechanic in 5e, directly tied to a character's endurance.

These changes still function in line with the original design goals, creating pressure to rest, while impeding a character's abilities in a clear, understandable way that affects all classes in a similar manner.

Tank: a character whose primary role is to absorb damage as a means of protecting allies
5e classes can tank damage in a few different ways. Paladins and Fighters can be ‘AC tanks’, who reduce incoming damage by being difficult to hit and having solid saving throws. Rogues and Monks, while they don’t usually serve as a blocker to protect allies, avoid damage through ‘dodge tanking’ – using abilities such as Uncanny Dodge, Evasion, and Deflect Missiles to nullify or reduce damage when hit. Moon Druids can be ‘sponge tanks,’ using their Wild Shape as a disposable HP pool to soak up huge amounts of damage. Barbarians are also ‘sponge tanks’ of a form, using damage resistance to halve incoming damage and simply, well, ‘tanking it’ with their massive HP pool.
A cynic might say that if these methods were perfectly balanced, it really doesn’t matter which method you use – all will work out equally well in the long run – and the difference is more for the sake of appearance than anything else. Naturally, my strawman cynic is wrong, and for a number of reasons, but the one I’m looking at today is the impact of variance on planning.
Let’s consider a Paladin and a Barbarian, in a heavily abstracted scenario. A monster hits for 50 damage. The Paladin has a 50% chance to avoid being hit, while the Barbarian will definitely get hit but only takes 50% of the damage. Both start with the same HP. Who dies first?
At first glance, this might seem to be a draw. Both have the same expected (average) damage—25—so over an infinite number of attacks, we’d expect them to lose HP at the same rate. As they do not have infinite HP, however, the threshold becomes quite important. If they have 40 HP, the Paladin could die in one hit, while the Barbarian will not. If they have 20, the Barbarian is faced with certain death, while the Paladin might get lucky.

​Maybe this doesn’t appear that useful. If you’re really low, you want to be the Paladin; if you’re doing ok maybe you’d prefer to be the Barbarian – so what? Given you can’t just switch classes on the fly is it really true to say that one is better off in general? Yes. Yes it is.
Imagine they have exactly 100 HP. Neither is at immediate risk of death, and on average we’d expect both to die on turn 4. How should each approach the encounter?
For the Paladin, that’s a complex question. If they’re unlucky, they could die in two hits. They know they’re safe until they take damage the first time, but at that point all bets are off. Can the Paladin defeat their opponent quickly, or are they willing to gamble with their life?
For the Barbarian, planning is much easier. They will die on the fourth turn. Can they defeat their opponent before then? If so, the combat is safe. If not, they should flee.
In actual games, we have to account for variance from the damage dice as well. The monster could roll higher or lower than expected. Here, too, a Barbarian’s damage resistance reduces the variance: by halving the incoming damage, the difference between minimum and maximum damage rolls is less for the Barbarian than other classes. Less variance again means a better ability to predict outcomes, and more options in combat.

​The fundamental difference between these two modes of tanking is in outcome variance, and that affects their ability to plan an encounter and take calculated risks. With less uncertainty to account for, a Barbarian is better able to plan their actions in combat, and can expect a greater degree of success.