Making a turnbased tactical dungeon-crawler/roguelike.

Some people do not like RNG stuff like critical damage. It can be frustrating and also seem less skill is required as you can be helped/killed "by the RNG".

Im using a similar setup to XCOM when it comes to crits. You also control a group of PCs and fight against groups of enemies. You can not heal between all combat situations.

Typically a average goblin enemy takes 3 hits from your heroes to die, while your heroes take maybe 10 hits to die. There is many special attacks that deal extra damge or do other stuff of course.

Any attack/ability inflicts 80-120% base damage. It may crit (right now: 20% chance to crit, which doubles the damage). Unlike games like XCOM, WOW or Diablo, you can not increase crit-chance by items/leveling up. (If each game agent/item has base damage, crit chance, and crit damage modifiers it gets annoying to compare them in my experience).

Ideas to limit annoyment of crits:

-dont allow it if it will kill the PC. PCs can only be killed with normal damage (non-crits). Disable this rule on higher difficulty.

<<>>

What is your thoughts on this? The idea is that crits makes fight more unpredictable and unsafe. If you have 7 hp remaining and the goblin deals 3-5 hp damage you are safe for one turn more, but not if it can crit! But you dont want the player to think the game cheats. Many players complained about missed shots and unfair crits in XCOM as far as I remembered...

Randomness is fine. It never stopped people enjoying Magic: The Gathering or poker. Just try to reduce long streaks of unexpected results.

Either is fine as long as you balance the game well. Plenty of games have it, plenty of games lack it.

If a player is attacked by few foes between turns and it takes many hits to get them to 0 hp, crits don't have a big impact. Being within a crit of death is already a failure case.

If a couple bad rolls kill a player, it's a much more important part of the system. That can be fine, some players really like the mechanic of nobody being completely safe in battle, but it's something that has to be executed more carefully. For example, that works better in an x-com type game where attrition in your party is expected.

Yes attrition is present (you heal between dungeons and each dungeon may have many fights).

So some weapons deal less base damage but have more crit chance? Some games do this but the logic is sometimes a bit strange. Why do shotguns have more crits than sniper rifles in XCOM for example? I would be the other way around since shotguns spray an area but high precision rifles could score "headshots" etc.

In a medieval setting which weapons would have higher crits? Spears more than swords? More than axes?
Im more leaning towards weapon types dealing different dmg depending on enemy/armour type so it might be too much with different crit chances as well.

Warcraft 3 does this well. It uses seeded values and tries to skew everything towards an average ammount.

So if you have a 10% chance to crit, and haven't critted in a while, the game will start increasing your odds behind the scenes. If you crit a lot, you'll start decreasing chances behind the scenes.

Funny enough this makes (in custom maps/mods) it extremely worth it getting very low crit chance passives, as you can take note of when you haven't critted in a while, and time your crit as your odds increase.

How much extra do your crits do? I think maybe that could be used as a mitigating factor. Player crits could do double damage, and I don't think the player would complain, but if a player gets one-shotted by a goblin crit, your probably going to hear complaints.

I like randomness within statistics curves.

Dice-based games like D&D do this naturally.

Consider something that rolls 3d6 (three six-sided dice). It forms a curve from 3 to 18. Your worst odds (1/216) are on the extreme values of 3 and 18, your best odds (27/216) are the central values of 10 and 11. The same number range can be skewed right with the 4d6 dropping the lowest value, used for rolling certain stats. Less of a curve are two dice, say 2d10 (two ten-sided dice): it is more like two straight lines, 2 to 20. The worst odds (1/100) on the extremes of 2 and 20, best odds (10/100) for rolling an 11. The game designers carefully designed different items with different curves, sometimes favoring many small dice (e.g. 6d4) for a deep distribution curve strongly favoring the middle, sometimes few dice or a single die like 1d20 for a flat distribution, and sometimes mixing many different dice to get a distribution that works just right for their design.

This type of distribution curve works really well in video games. You want an occasional critical miss or critical fail, but mostly you want values in a central hill. However, sometimes you may want something that has a very different curve. You may have a special axe that deal enormous damage but has a high probability of failure, so build a curve with a large hump near zero and a small hump at the high value. Repeat with whatever specific designs you need.

The same is true for loot drops. Generally the stats skew left, you have a higher probability of getting bad stuff than anything else, but sometimes you want to get something amazing.

Statistics tables are generally a designer's close friend. Even if you need to work with pictures of the curves and then ask the programmers to make the desired curve, understanding distributions is important.

If you are making a "tactical" game, you might want to be careful with RNG. It's usually not a good thing because it's literally the opposite of what "strategy" gameplay is all about.

Gambling is fun, but not when playing a strategy/tactic game. In this kind of games it feels terrible because a lucky hit will always feel "deserved" while an unlucky one will always feel unfair. You may have 50/50 odds but people will only focus on the negative side of rng, that is just how it works. They won't think "oh I got a lucky shot there, nice", they will think it's just normal, and they will be frustrated when the opposite happens because it will disrupt the strategy they designed and were putting in place.

Some rng can be useful : it can be a good thing to make your game non deterministic so you can surprise the player and force him to adapt his strategy constantly, making the game more dynamic and interesting and also faster paced (because you can't effectively predict a lot of moves when rng is involved). The key is to balance it right, so that good strategy can always overcome bad luck and bad strategy can't be saved just by good luck.

I would advise you to stay away from "crit type" rng though. Involving rng in a strategy/tactics game can be good, but you need to hide it. Having a crit randomly kill one of your character who, you thought, could still endure 2 shots is a terrible feeling. The enjoyment of strategy is about making the right calls, if you make the right call and then rng decides that your shot misses, it sucks. The best way in my opinion to involve rng is within the AI. Don't use rng on the numbers, or on stuff that is clearly visible, don't advertise it. Use it to determine which pattern your AI will follow this turn, or which target to attack when there are multiple good possibilities. It ends up achieving the same result, your game is unpredictable, but without people realizing that rng is there and without the usual negative feelings associated with that. It also makes your AI look smarter than it probably is.

The more rng there is, the less tactical your game is. It's your call to decide where you want your game to be on the strategy<--->gambling spectrum. If you really want your game to appeal to player looking for a strategy/tactical game, don't put rng in for the sake of it or because it's what other games do. You are making a rogue like so there is probably already a lot of rng involved in generating the levels, do you really need to add more ? Rng doesn't just feel bad because of bad luck, it takes control away from the player, makes your victories feel undeserved and your defeats feel bitter. It can really turn off people and even kill games. Look at Darkest Dungeon, it's a great game in many aspects, the art is awesome, some of the mechanics are very original and yet it got some truly terrible reviews from people expecting a strategy game and finding out it's in fact just a RNG fest. Now some people loved the game, it was a hit, but not the people looking for a strategy/tactic experience. Those people didn't enjoy it solely because of the rng.

It's not quite as simple as strategy/tactics vs. luck. Backgammon and Magic are both very luck oriented games while still demanding skill and strategy. It's a topic of some contention in the wargaming world: Is the ideal a platonic perfect information battle of the wits, or is battle inherently chaotic, and the winner should be the one who can flexibly ride the ebbs and flows of a battle?

In some sense, it tends to be a trade-off between depth of reading and breadth of reading. With perfect information, you're trying to think further ahead then your opponent. If there's a convoluted path just on the edge of disaster to checkmate, then that's a win. With randomness, you can't prune out paths in the same way: Your flank might be routed, your feignt might actually break through, your infantry might be delayed and fail to take up position on the hill. So instead of "what's the best move 10 turns from now", it's "what 20 situations might I be in in 3 turns?" If you're in a situation where a bad crit loses you the game, that raises questions of how your strategy got inflexible enough that one roll could decide the outcome.

There's obviously an extreme where strategic considerations no longer matter, but there's think there's a surprisingly wide range of randomization usage that can all be considered equally strategic, if strategic in different ways.