Lorem ipsum dolor sit amet, elit eget consectetuer adipiscing aenean dolor

# Random and AI

Please correct the bias (in favor of the AI) in the game’s random.
Clearly this is done to compensate the weakness of the AI.
This is not a good idea. It only gives the impression that you can play as good as you want, you will lose because the developpers decided you have to lose. That’s not fair.
So, correct the random (and improve the AI…)

I know this look like a 2 years ago topic. But it seems you (developpers) keep on thinking that a good idea (hello Lyrasza !). That’s not, it’s ruining the fun.

1 Like

I really think that this has everything to do with confirmation bias and nothing to do with fudging gem drops.

up !

8 respawns in row: with a normal probability of 30%, it should be a 7/10 000 chance.
Still no problem with the randomness of the game?

3 Likes

I’ve lost track of how many tens of thousands of battles I’ve fought in this game, so you’d kind of expect it to happen a number of times.

Also, I believe the probability for Backup to trigger is 35%, not 30%.

The AI doesn’t have an advantage with these percentage chances.

That may be so that the AI has no advantage. However, having fought incredibly unlikely numbers of Bandit and Giant Spider summons today in GW battles, and having done similar things way too many times, there’s absolutely no way that the purported percentage chances are both accurate and working correctly.

1 Like

Maybe the chance is still 30%, but we roll a d100, the AI a d40

Or maybe the AI rolls 4 d100s in parallel (one per troop) and takes the most favorable result… Who knows, the possibilities are endless

I don’t care btw, the AI is so bad at playing the game that the only way to steal a win from time to time is to get massively lucky.

Edit: to clarify my examples with dices, oc I’m assuming a raw trait chance value (like 30) is directly compared to a rolled value (that would normally be between 1 and 100), without adjusting percentages based on the dice faces.

Probably not. Going by your numbers it’s something that happens roughly every 1000 battles, so dozens of players should encounter it every day.

All probability percentage goes out of the window according to difficulty. I’ve had amarok miss devour 3 times in 1 game with 95% and then been devoured by a 20pc troop. Percentages aren’t real in this game, the playing field is askew and it is the player that takes the hit

1 Like

.05^3= .000125

So, there’s about a hundreth-of-a-percent chance of that happening. About the same as pulling a mythic from a gem key, I think, if I recall correctly that the odds if that are about 1 in 8000 (I might just be thinking of shiny-pokemon occurence rates — my bad, if so).

Unlikely, and sucks for sure, but not so unlikely that there’s necessarily anything wrong.

Now, if he missed 10 times in a row in the same battle, that would be interesting, as that might suggest the seeding is wrong on a per-battle basis or something.

10 times in a row across 10 battles would also be interesting because if we were to say there was only 1 attempt each fight, all of which missed, and only in the 11th battle on the second attempt did one finally work — that would maybe suggest that the first attempt is “doomed to fail” no matter what the percentages say.

But—since one anecdote does not constitute either—all we have is an outlier here (and also I picked 10 out if thin air — I’m not sure what the length of the “streak” would actually have to be to have a confident reason to doubt its fairness/randomness).

Statistics don’t work if one cherry-picks

You’re right the spawn probability is not 30% it is 35%.
So in my case it was not a 7/100 000 chance (I made a mistake in the result reading) but a 2/10 000 chance.

And you’re right an example is not a generality (I know my statistics very well, thank you).

The problem is that it’s happening quite everyday in GvG (1/5 chance…). And I have played (and seen played) enough game in GvG to make that a generality (more than 1 year in rank 5 or less, playing and looking my mate playing).

It is just that I’m now so pissed off that I finally took a screen shot.

This is not a confirmation bias but a confirmed bias.