Tell me the ai it's not rigged!

[Flint]

@Amadan

@Flint

We have been told that “difficulty” scales beyond the player’s control in such a situation

o.O

Um…we were kiiiiiinda told the exact opposite? That there isn’t secret, invisible difficulty scaling? Not sure where you’re getting this from.

https://community.gemsofwar.com/t/concerned-for-the-future-of-the-game/26591/95

Specifically, the following from Salty:

@Saltypatra

You may not be aware of this, but a combo breaker does exist in game. On console the AI is much fairer, and has a much better chance of getting lucky than the PC AI. The AI (this is for PVP specifically) is meant to scale as the player goes up in rank, until it matches the players luck at rank 1. However, as we released the Unity version for PC, we realised that there was a bug in this code, causing the AI to only be allowed a small percentage of the player’s luck, unlike the console build. This has now been fixed, and Tier 1 PVP players should be seeing the AI (in very extreme cases where the RNG is exceedingly lucky) only getting 3 times the amount of lucky events that the player does in a single match. They cannot possibly get more since the fix.

There are many other examples. It’s been acknowledged that the “luck” scales as you increase in PvP rank, and this was assumed to be modifying the combo breaker or gem drops. The developers have acknowledged many, many times that they tweak these things. Here’s an example where it’s acknowledged that the AI can be three times as lucky as the player.

I personally do not notice anything major with cascades that can’t be attributed to recall bias. What I have noticed is that things that are supposedly based on a percentage are not exactly accurate. When the Kraken was dominant in the meta, I remembered thinking that the devour sure seemed to activate more than 35% of the time. So I started tracking, and after two weeks I was at 65 observed devours out of 84 casts for the AI. The chances of this occurring given a 35% expected rate was something like 1 in 1 quintillion, which curiously also meant that devour activating 0 or 1 times out of 84 (given a 35% expected rate) was more likely than 65 out of 84.

Others that I did not track as accurately included Fizzbang’s explode, Nobend’s explode, Infernal King’s respawn, Dragon Soul’s respawn, and Gorgotha’s Ancient Horror. All were off by a significant factor. (Dragon Soul’s was acknowledged and changed.)

I simply do not believe that Tier 1 ranked PvP (to be fair, this is exclusively what I am referring to) adheres to the published chances because I have not observed anything close to them for what I have tracked. Basically, if we were told to collect data on these things and determine probabilities (assuming they were unknown), we wouldn’t come up with what’s listed on the card.

Is it possible I’m just really unlucky? Sure. Like a 1 in a quintillion chance.