AI % vs Player % - RNG Hates Me or It's Rigged

Zekir

As the title says... Why does it feel like the AI's % to trigger is entirely in favor of the AI, and the Players % is also in favor of the AI.

Start Spidey with his 3% chance to dodge feels like it's really saying AI has a 3% chance to fail to dodge, meanwhile it never seems to trigger under player control. I've seen enough Starkie Gameplay to notice when it never triggers for players, but can evade up to 10 hits without fail with AI, (No, I'm not talking about his Special Dex ability).

Or Glancing on Storm X, Yeah it's a 60%, but that doesn't mean 1 hit out of 20 actually connect. I never see glancing when using my Storm X, but I see it on EVERY hit with AI controlled Storm X.

I get it's suppose to be challenging content, but that doesn't mean fix the ratios behind the scenes to make stuff more annoying.

(Also, can we talk about when Glancing is gonna actually be affected by ability accuracy of SOME kind? Falcon glancing when locked on is really dumb. Glancing is a Defensive Ability, and yet, it ignores modification if every kind.)

If you read all this, thanks for taking the time.
TLDR: Feels like Rigged %s for AI, but I know someone is gonna say RNGesus hates me.

DNA3000

As the title says... Why does it feel like the AI's % to trigger is entirely in favor of the AI, and the Players % is also in favor of the AI.

There are many people who live stream content. Ever wonder why these kinds of problems never seem to show up when players are recording content?

It is a combination of three things. First of all, players are not always aware of how effects work or what the percentages affect. For example, when it comes to evade, the percentage chance listed is the chance to trigger evade. But evade is a kind of status effect: when a champion triggers evade they will often be able to evade multiple hits that come in rapid succession, intentionally so that a single evade isn't worthless when a multi-hit attack is used against an evading champion. But that means a player might see "evade ... evade... evade" in rapid succession as the champion evades multiple hits and thinks the champion is getting extraordinarily lucky, when in fact they just triggered evade once, and all the evade call outs are just the champion dodging multiple swings as part of a single evade trigger.

Second, players tend to remember low probability failures more than low probability successes. If a champion dodges 3% of the time, a player is not going to count every single swing in that fight. They will swing forty times in that fight and notice a single evade, but then notice that evades are happening in many fights in a row. That's going to seem like it is happening more frequently than it actually is happening.

And finally, humans suck at noticing how often things happen over long stretches. When a player is told something has a 3% chance to happen, they think "one in thirty" and then when they see two evades happen ten swings apart they think "aha, its broken." But for random events, a 3% chance doesn't mean once every thirty swings. It means each swing has a separate independent 3% chance to trigger. Sometimes you will get three in a row. Sometimes you will see a hundred swings without a single trigger. That's not just likely to happen, for genuinely random events that will essentially always happen. Part of being random is being unpredictable. If 3% happened regularly every 30 or so times, that wouldn't be random. That would be regularly scheduled.

The combination of all of these is why most people think random number generators are broken. It is their own sense of and recollection of what is supposed to happen that is broken. But if these random events were as broken as players claimed to have observed, this would be trivially obvious to see and document in gameplay recordings, over long periods of time. But that has basically never happened.