Once upon a time. It made sense. Now the defenses are tough enough opponents that they don’t need increase luck when there’s one troop left on the AI side.
This week I had at least 10 losses where I had 3 troops left and the AI only had one troop that was almost dead. Each time the match went in the favor of the AI. I doubt I speak alone about this pattern.
Also the “win streak” counter should be removed since it’s impossible to win more than 50 matches in a row now. And it seems like it’s only here now to increase the difficulty of the AI if you win more than 10 matches in a row.
Sorry, which “increased luck factor” for the AI? As far as I know, the existence of such a thing has never been confirmed, and has in fact been denied by the devs.
The closest thing that actually has been confirmed is that they reduce the likelihood for generated 4-matches for generators only for both teams after multiple Extra Turns.
Denied you say? Besides the words. It’s very apparent by just playing more than 100 matches in PvP to see the AI get 3x luckier at certain times in the match.
It’s been a while since I read the article. It was back in July.
I assume that AI in PVP is the same as in Guild Wars. If that’s true. @Saltypatra, does a player that’s tier 1 in PvP face a different kind of luck factor in GW than a player at tier 5?
She stated that the AI should never get more than 3x the player’s number of “lucky events.” This sounds like a limiter to me, not a sign that the AI is luckier than the player. Removing this limiter would allow the AI to get even luckier.
I read the opposite. I read it as the 3x kicks in during certain events during the battle. I’m totally fine with removing the luck factor on both sides.
I suppose she’ll just need to clarify, but while the phrasing was a bit ambiguous, it seemed fairly clear that she was discussing a “luck limit,” not a “luck creator.” Nevertheless, my personal experience doesn’t match yours, but if it did, I too would be frustrated by a luck disparity over time.
Right, they’ve reduced the amount of artificial luck the player was getting beyond the expected 50/50 split, by decreasing the luck limiting on the AI. In that case, the thread request is more like, “Restore Increased Luck Factor for the Player,” which likely has a similar psychological effect.
I think the only way to remove “luck” is to prevent either side from ever Mana Surging, getting a chained skydrop ever, or having spawned gems create a 4-match. The feature that’s in place is intended to prevent the AI’s luck from ever exceeding the player’s by a huge amount.
As an example, let’s say you are rolling 6-sided dice against the AI. Whoever gets a “6” gets one point, and you’re competing for score. Without a limiter, the AI could get extremely lucky (at a low chance) and could get a streak of 10 “6s” before you get even one. The limiter prevents the AI from getting, say, 4 6’s when you haven’t gotten any. Effectively, the game is cheating in your favor, though it doesn’t prevent the AI from getting lucky – it simply prevents it from getting wildly more lucky than the player.
Salty’s post was that they discovered a bug wherein that limiter was overly severely cheating in the player’s favor – say, by never letting the AI get two “6s,” in my above oversimplification – and they fixed it to properly make things more even – but not exactly even, and the game still cheats in the player’s favor, just less so.
Im not sure if this make sense or just makes things even less clear
Lol… All I know is that RNG favors the AI when close to defeat. Whatever makes it that way…I ask that it doesn’t anymore.
Same RNG regardless of how many troops the AI has. That’s all I want. Consistency.
Either it doesn’t work as intended, or it doesn’t matter in any event because once the ai goes on a rip the match is over before the limiter will have any effect whatsoever. I miss the days of Puzzle Quest 1&2 when the ai could get several turns in a row, and it didn’t end the match. Now it’s just spam mana creation until one side wins. And that’s not even mentioning the insane skull drops and superhuman ‘one troop left’ phenomena.
Sorry, it does not make any sense and I also think it is somewhat … oversimplified and overcomplicated at the same time as well as somewhat misinterpreted. The “luck” you are talking about does not work that way IMHO in GoW at least in current meta and with current pRNG generator. Of course, you cannot be sure what is blocked by the code and how it specifically works. Otherwise, please enlighten us on the matters.
The main issue I have is that how specifically the game determines what is “lucky” and what is “unlucky”. It does not seem to be evident even to quite experienced players unless the AI “counter-cheats” (makes a move that might benefit the players instead of itself, which I have not noticed much) deliberately during its turn which is the only easy way to code it. In another words, I don’t think that definition of “luck” is correctly coded at the moment in this current unity build from the point of human common sense.
Salty would need to address what specifically is considered a “lucky event,” but I understood it to mean “a 4-match is created as a result of spawned/falling gems,” in which case it is most decidedly possible for the game to adjust the luck of the AI to avoid this situation.
So, what happens in this case? IDK, just curious? Is one of the gems replaced? Two? All rerolled? It is very obscure. Now, what about a situation of extra turn? Like creating a 4+ match, for example, for some goblin card after the action is taken by AI? Or cascading? Does that count? Sorry, too many questions. How does this work in case of respawning board, like casting exploders? I’m sure everyone understand the implications.
Let me try to briefly expand. Let’s say it is a replacing drop of 4 tiles and all replacing tiles are of the same color. Probability of this is about 0.4%, so we are talking at least about 400-sided dice and a chance of 1:400. What about chance of 1:399 or 1:398? Is there any rounding involved? I’ll have to stop, this list of obscurities is virtually endless here.
Human perception including statistical criteria is built around 95-99% chances. This is due to difficulty that human brain has processing visually or cognitively larger numbers exceeding more or less normal perception threshold of 100. So, anything that regularly falls out of this range is considered improbable or having negligible probability. It will be extremely lucky for AI to roll something that has smaller chance than 1 in 100. What is the threshold in GoW? Is it 1:400, 1:1000, 1:10000? Is there any filtering in other aspects like devour, resurrection, etc etc? There are just too many silly questions here.
I understand your points, and I agree that calculating the resultant shift in probability is a nuanced and difficult process. My point is that it would not be hard to write code like this:
while (new_gemfall_would_result_in_fourmatch())
{
reroll_all_new_gems();
}
But again, it’s all speculation until/unless Salty decides to enlighten us.
I have actually seen this several times just today… On multiple occasions forest troll has cast immediately following my removal of green gems using shiny. A f. Troll cast with 3 green gems on the board that on one occasion even set me up with a 5 match. Also in one battle today apothecary cast twice choosing a color that gave the ai zero brown matches and setting me up with several 4 or 5 matches.
Does it mean anything? I have no idea. But it seemed to me that the ai was trying to help me most of the day today (what little i played before i got sidelined by a crippling migrane) lol
