Is the AI cheating? I'm pretty sure it's cheating.
I've seen the hard data showing that the game skews coinflips in the AI's favor roughly 60-40.
I have less hard data on it getting preset opening hands, but unless my perception bias is even worse than I thought, the AI opens with double-pokeball-into-Oak or play-entire-hand-then-Copycat-a-new-one far more regularly than it statistically should, to say nothing of how frequently it gets Rare Candy on turn 3 or 4.
Personally, I could live with this - the goal is to provide a challenge (presumably), and opponent AI is notoriously hard to do.
However, there are two problems.
First, the cheating seems to have gotten noticeably worse after a reward was set for winning 5/10 matches in a row in the Random Battle mode, either because the numbers were tweaked or because having something on the line made it a lot easier to notice how consistent the AI was at getting its win-state out within a few turns. Setting this challenge and giving the AI a Millenium Puzzle makes for a very unpleasant experience.
Second, I'm playing on an older phone - every time I concede to the AI because it got a perfect opening and I didn't draw the cards I needed to counter it, the game takes roughly 10 to 20 seconds to ask me if I want to use one of the rental decks. This is especially obnoxious when I'm explicitly trying to complete the type-restriction challenge.
It's gotten to the point where I've grown to hate every single one of the Expert deck archetypes with a passion. Playing ladder matches is less of a frustrating Sisyphean nightmare, even with losses deleting progress.
Comments
-
It's not the AI cheating… it's just how they poorly implemented their random algorithm.
3 -
"The AI isn't cheating, the game just has a poorly implemented random algorithm" - that's HEAVILY skewed in it's favor. Noticeably even more so ever since a reward was introduced that requires winning 5 times in a row.
3 -
Yes, the coin flips are reflecting how the algorithm works… if you go vs AI it's obvious it gets way more heads than real players... but again, that's just their algorithm.
3 -
It takes like a dozen matches to see the computer get an opening hand that wasn't either already perfect, Oak + Pokeballs into perfect or Copycat into perfect.
I finally got a match where this didn't happen and instead I got six tails in a row against Igglybuff and Milotic EX.
Random Match is a great idea in concept, but when the cheating on the Expert decks requires a custom-built counter strategy to win consistently (and the pool is nothing but Expert decks) it's really just a recipe for misery.
1 -
Indeed, that's the state of the games today... since they can't make a decent enough "AI", virtually every game (especially in higher difficulty modes) boosts the AI with as many perks as possible to keep it competitive... in TCG Pocket they manipulate the coin flip outcome, great starting hands and overall favorable card draw.
2 -
If you were running a game where a coin-flip isn't 50/50 all the way through, that's cheating. Straight up. If you were at any other sort of gambling scenario, there would be people being arrested over it. Don't kid yourself.
But my observation of this problem is worse the coin flip rates. It's the way the deck is stacked. It's not shuffled. It flat out prevents a player at the "highest difficulty" from getting the right next cards while the AI consistently gets exactly what is needed to win before you even get to evolve to the next stage.
I play this game mostly as a more interesting version of solitaire. If I want higher difficulty, I want it to be in the form of deck selection, not ANYTHING related to randomness not being actually random. Anything other than honest shuffling is, by definition, cheating. Again, if you did any such things in a REAL TCG game? It wouldn't be pretty.
The real confusion is when game developers don't know the difference between "harder" and impossible.
Playing against other people is almost as bad, actually. The game matches players in order to prevent people from winning long streaks or to guarantee wins to someone who otherwise wouldn't depending on his deck selection. DEI is about determining outcomes and that's a pretty reasonable definition of what's actually going on here as well. They are attempting too much to determine outcomes. The problem? It's obvious.
If you're just wanting to enjoy playing the game and completing all of the various types of missions, you will run into 10 straight losses before you are allowed to win. The short way to win (if you can call it that) is to quit when you get a bad hand. Eventually, it will give you one because it doesn't want "too many losses" or people will quit.
But what you can't do is "10 wins in a row" at the highest level without some amazing persistence. And if all you want to do is play a relaxing game? That's not what you're going to get.1 -
This is where I am. The game is cheating because they don't think about what this game actually is. If a coin flip isn't 50/50 all the way through or if the shuffle isn't a real shuffle it's rigged. And if you want to complete levels of solo play, it means they are wasting your time intentionally. There is no higher degree of disrespect of a player's time.
1 -
If you pay attention to what your opponent does in a series of matches you'll see that it isn't really rigged. The computer can definitely get draws that brick and bad coin flips. A lot of it comes down to perception and personal bias. Like anything RNG.
1 -
I have been paying attention. And the AI does brick occasionally.
It also does very silly/annoying things like Copycat against opponents with zero cards in hand, or repeatedly using Fragrant Forest when it isn't running any Grass Pokemon, or activating Victini's effect on Typhlosion EX when its target has no energy on it and will die from the damage regardless.If anything, that just makes the cheating more blatant - the AI appears to just play whatever on the board happens to be playable, and the game's way to balance this is, on expert, to deliberately feed it cards that it cannot misplay.
I'm also obligated to point out that it has been measured elsewhere: the coin flips are skewed in the AI's favor.
1 -
I have now been trying for two weeks to get five wins in a row against the Expert Random Deck CPU.
It's gotten to the point where I built a deck specifically to be as consistent as I can figure out how to make it, by using a combination of Quick-Growth, Auto-evolve Caterpie, Copy and Fragrant Forest.
The result still isn't as consistent as the AI's openings.
1
