More thoughts on the ONE D&D test material...

Crit chance in general is technically bad game design. It fails the test of "if this is used against the player, is it still fun?"

Variance is a problem in high level games...

Variance is fun but it's incredibly frustrating at the same time... and sometimes, variance is not on your side:

I recently came off a one-shot where I was crit 5 times (we rolled it back to 4 after we realized an enemy was supposed to straight roll instead of with advantage) in one session. In fact, the boss crit me 3 times in 4 attacks across 2 turns, the non-crit was still a hit. If I was playing any class besides a Moon Druid, I'm fairly certain my character would have gone down -- instead I was down both uses of Wild Shape with 3 remaining hit points.

The math on this gets weird because we'll assume damage happens 1/3 of the time but a crit means I took twice as much damage 3 times... Let's assume the enemy was rolling a 1d12 for their attack, that's 6.5 damage per successful attack and at 65%, I should expect ~4.25 per attack or about ~8.5 per turn or 17 damage in two turns. In this example, I effectively took 7 damage rolls of 6.5 damage for a total of 45.5 average damage... 2.5x the expected damage of those turns. I didn't use any damage bonuses there, but it's fairly safe to assume it was at least +3 but I omitted that because the house rules in this particular game is double damage, not double damage dice (so it would be more damage than normal rules). In a turn based game, you can't expect to heal though this because healing doesn't have crit rules... I can't cast Healing Word or Cure Wounds and have a chance for double healing.

You can't really balance for multiple crits in a round of combat. The EV of a crit basically means 5% more damage in the long run but it's 100% in the short term. A crit functionally means 2x action economy for that particular action. In this example,

Monsters shouldn't Crit

I agree with this rule for a few reasons.

1. It's impossible to balance variance. The long-term expected damage increase of critical strikes is little less than 5% using standard rules but since combat lasts ~3 turns, a 100% increase in damage is effectively a free turn.

On top of that, combat snowballs. The difference between an ally dropping in round 1 vs round 3 is drastic and the game can't account for that.

2. Critical Strikes are fun when players give them but feel really bad when they receive them

3. There's a silent nerf to controversial effects and spells like Silvery Barbs

One of the first homebrew campaigns I ran eventually ran into a small problem -- in order to threaten the players, I had to rely on critical attacks to make combat interesting. If I would have statically increased encounter damage by 20-30%, a crit would have outright killed players instead of knocking them out.

4. It's easy to house rule monster crits into the game, or more importantly, add the critical hit feature to unique monsters

If you want deadlier combat in your game, let monsters crit. I'd prefer to run a game where I can make an encounter exceptionally difficult but not risk a TPK due to luck. We're already running off a game where successful 2/3 of the time and a monster with 3 attacks can down an enemy in one turn by hitting 3 times... it just gets worse if they effectively hit 4 or 5 times instead.

Nerfs to Critical Attacks on Sneak Attack, Smites, and Spell Attacks)

Critical Hits
Weapons and Unarmed Strikes have a special feature for player characters: Critical Hits. If a player character rolls a 20 for an attack roll with a Weapon or an Unarmed Strike, the attack is also a Critical Hit, which means it deals extra damage to the target; you roll the damage dice of the Weapon or Unarmed Strike a second time and add the second roll as extra damage to the target.

I dislike this. If we're going to allow critical strikes in the game, we shouldn't nerf features that allow classes to keep up in damage. Rogues trade-off multiple attacks with Sneak Attack damage, it's their thing -- their damage is all or nothing (no partial damage). On the other hand, fighters get 2-4 attacks and can critical hit each of those. I understand the balance issue between a rogue making 1 attack and getting a crit vs a fighter making 4 attacks and getting 1 crit; there's a lot of math going on here but critical strikes scale disproportionally well with Sneak Attack. But there's a trade-off, rogues get to use Sneak Attack once per turn and they can't spread that damage (but fighters can). Paladins need to spend spell slots to do damage, so they aren't damage batteries like fighters and rogues. The fighter's multiattack feature scales extremely well with magic weapons like Flame Tongue or spells like Elemental Weapon. The point here is that each type of damage has trade-offs and different styles.

The math on Sneak Attack damage scales similar to the expected damage from a fighter (without Flame Tongue). This means a rogue's damage with one attack can keep up with a fighter's 3 or 4 attacks later on. There are two critical assumptions here -- (1) the rogue can proc sneak attack on their turn and (2) the rogue only procs sneak attack once per round. Rogues benefit from things like Commanding Strike and attacks of opportunity way more than fighters do. It makes rogues unique and incredibly powerful in certain situations -- and that should be fun for players who want to be rewarded by making good decisions.

Limiting critical hits to weapon attacks (basically removing the chance to crit with spell attacks) is confusing. Why would spell attacks like Eldritch Blast not be able to crit? Most damage cantrips don't get to add an ability score modifier to the attack (and if they do, it's a class feature that typically comes on at mid-levels). It's not even really clear if "damage dice of the weapon" should include Flame Tongue's bonus damage or just the base weapon; I assume it's just the base weapon because of how the character sheet displays damage (and that DnD Beyond displays the bonus fire damage in the notes section instead of the damage box but without that example, I would have thought the fire damage is from the weapon, not a tertiary effect).