Both 1- and 2- averaged really suck, and I resent them.
Two averaged numbers (rand1 + rand2 / 2) creates far too much stratification on a lot of numbers. Displayed 25 and 75 being bumped to 12.5 and 87.5 is just too dramatic in a lot of cases, and dupes players into thinking that low odds don’t happen. This was attempted to be fixed in Fates and Shadows of Valentia with its sine wave based formula for the higher numbers.
Whereas 1-100 allows the player to get destroyed because statistics are not intuitive to most people, because a normal child will see a 90+ and assume it’ll happen almost always and strategize around it being consistent and then get totally destroyed when it doesn’t, creating incredible frustration.
And in my opinion, the sine formula inherits the worst of both of these. It still stratifies the range very strongly, so players don’t eventually learn that statistics are unreliable (a displayed 75 is now… 85! still a huge jump), while also leaving things incredibly frustrating (die to a 10… 10% of the time!)
I babbled too much. A *lot* too much.
The thing I personally really like about a manipulated randomness is that it makes specialists better than generalists, something that the game’s other formulas don’t do as well. This would be important if enemy quality was higher, but in most of the actual games, enemy quality is well below the player’s stats. That’s the thing that, in my opinion, makes the overall system terrible. In most games enemies hitting is almost guaranteed on the most evasive, or becomes unrealistic on even the slowest.
For example, in Geneaology of the Holy War, a lot of enemies just can’t offer big damage against your better fighters. Lots of player units are pretty bulky in G1 even without holy weapons and this also makes ie. Arden not very useful, because having an option for dying in five hits instead of four is incredibly marginal - you can just position units while baiting to avoid these issues.
But, unlike in Thracia 776 or Binding Blade or Blazing Blade or Sacred Stones; there isn’t an enemy archetype running around in large quantities that does 8 damage with 20 hit. Enemies have better hit rates or damage, making them maintain some scariness despite your units’ available health – and there’s the times when the game forces you to charge into range of 8 or more units at the same time to get a good fight going.
And yet, your own units can keep missing these foes, which makes things take longer and just get tedious, and isn’t something that high skill units don’t suffer from.
Whereas in the GBA games, we get both! Enemies that have horrible hit rate and evasion, making them totally free kills for units that have high speed stats. But at least they’re not obnoxious to deal with, we just watch them march into their death - just boring and not offering a challenge.
So, for me, if you’re going to leave enemies as having low hit rates, then you should let them hit, which means using 1 RN. If you’re going to improve their abilities to be actually a threat, then either the increased consistency and average-player game-feel of 2 RN or the statistic-friendly 1 RN are both quite appropriate.
Despite all that, I prefer the idea represented in a stratified randomness system, because quite simply, I’m also a human who is bad at statistics. I’ve played so much Angband (100 floors of dungeon!) and it still hasn’t really rubbed into my intuition that, no, a 1% chance to fail is going to happen, you can’t rely on it in any capacity, and I’ve lost at least a dozen characters due to them failing their 99% chance to cast a spell twice in a row.
But god, ((A+B)/2) is terrible. Like, let’s look at some ridiculous things, okay?
Disp | 2RN | Sine
65 | 75.85 | 72.01
70 | 82.30 | 78.87
75 | 87.75 | 85.00
80 | 92.20 | 90.14
85 | 95.65 | 94.16
90 | 98.10 | 97.05
A unit with 75 displayed hit actually connects with the target a whopping 17% more often than they “should” in 2rn, decreasing to “only” a 13.3% boost in the sine formula. I call this “relative” change; in contrast to their “absolute” changes being 12.5% and 10% respectively.
That’s a very substantial increase and it just makes me hate them both, because it stratifies the values so far from their originals that I know even kids noticed the manipulation was happening.
I think, overall, having the randomness be manipulated is good, but that nobody’s actually made a system that only covers what we really care about – the actual tips. What people truly despise is missing a 97, getting hit by a 6.
Tear Ring Saga did something similar to this, with a weird table. Unfortunately it’s still almost comedically dramatic, but it does a slightly more reasonable job.
Dsp | TRS
10 | 3
15 | 7
20 | 12
25 | 20
30 | 26
...
70 | 74
75 | 80
80 | 87
85 | 92
90 | 97
But you can see an important seed here - By homing our manipulations to only have those hugely dramatic effects on the edge, you get the upside of not lying to the player as much in the middle, which lets someone learn how probability “should” work, but still be leaning into how humans look at the edges of chances and incorrectly dismiss the unlikelier result too soon to avoid frustration.
We could do this with breakpoints Hit <= 5: actual 0; Hit >= 95: actual 100
; possibly with curving instead of abrupt jumping, or entirely with a formula like True = Disp*1.1-5
. We don’t… actually have to have our curve map directly in the range, if we’re having one.
I’ve done lots of things like this with the overall point being that, and the one I like the most is 9x/8 - 6
and at its maximum swing it only pushes RNs by 6% - A displayed 94 becomes a 99.75; a displayed 6 becomes 0.25. At the previous extremes, 25 and 75, it’s only pushing a 3% absolute and 5% relative change (25 → 22.13).
Of course, all of that is pointless, because it starts to miss the point of why you would want to do 2RN: To help people who are bad with statistics. Budging probabilities by small amounts doesn’t really help them.
So it depends on what you want. In a game where enemy stats are actually competent, I think it’s better to have randomness be modified such that it accentuates the consistencies of those who specialise in consistency, while not making units that lack that consistency also become reliable; which is directly why 2rn (itself) sucks - If I give you a 0 skill unit in 2rn they’re actually usable, but if I give you a 0 speed unit they’re awful – because the basic formula at hand doesn’t do good enough a job at making different stats worth even remotely the same amount.
You might think that “speed is the best stat” has no relation to the rn formula but they’re the same statement, because one of the reasons speed is the best is because base weapon hit is so high that against enemies without huge speed stats, you already have consistent hit from 2rn fixing you up.
All of these things really interplay with one another, and looking at each one in isolation is useful but also can trip you up because they all in the end comprise one actual thing –
2rn is about making the game consistent and reducing player frustration from randomness, ie. allowing the player to feel both the joy of making a tactical decision, but also to reduce the frustration of low success rates.
You know what else does that? Units with high skill stats. But it feels bad to use only the highly consistent units instead of playing with the more dangerous ones.
I’m babbling on and on too much, agh. The point is that each option has good and bad points about it, and that someone will feel betrayed no matter what you do.
If your game has enemies with actually competitive stats compared to player units, then it comes down to if you want more consistency on units whacking one another, or if you like having to manage the chaos of misses and needing multi-layered plans.
If your game has fodder enemies with strength stats, then you risk 2rn becoming tedious juggernauting, and 1rn being that but also with “oops, you died!” sprinkled in when getting whacked by 10%s.
If your game has units with low skill stats, there’s a good chance they’ll feel bad without a corrective factor, and a lot of those have enough influence that makes units with high skill stats feel out of place.
Overall, I like modified randomness, because missing 95s and getting whacked by 5s more than once a map is why I’ve got so few completed runs of Angband, and they’re incredibly frustrating moments that make it feel like the strategy I crafted was perfect if-only-my-units-didn’t-suck, instead of like I made an actual error in counting on them, because I am one of the many people who are bad at statistics.