This post is a response to the "Does the AI / Game Cheat in the battle frontier" motivated by the comments section of the following video:
(Regulars can obviously ignore this post. This post is also true of the gen 3 frontier)
As an appeal to the good faith of the reader; it is
hard to change your mind about something. The longer you've thought something is true and the more emotional your memories attached to that thing are; the harder it becomes. I realise this and its why I'm going to give a proper response instead of dismissing it. These "cheating" rumours are very old and surprisingly well ingrained into the community. If you are a long-time fan, you probably also had some form of confirmation bias when your Battle Tower team got hit by 3 Sheer colds back in 2010 or whatever. But I think its our job to share the truth, so feel free to link this post if you see the topic come up anywhere online.
I will be dividing this post up into sections which you can read depending on what exactly you want to know or why you think you believe what you do.
1. We literally have the source code of the games
https://github.com/pret/pokeplatinum
If you don't have the time to read the rest of this post; the knockout punch is that we have decompiled the exact source code of the game, and there is no evidence of any functions that manipulate rng, check the players streak (to alter what the enemy ai does).
Similarly there is no evidence of the AI cheating by "reacting" to what the player chooses. In fact, the way that the AI chooses its moves in gen 4 (and gen 3) is well known and documented, and none of it involves reading the players inputs. I do not really play difficulty hacks of these games (I just play the frontier); but those who do should be familiar with exactly what i mean here.
This point speaks for itself. Annoyingly, its difficult for me to "prove" this to you. If you don't trust me (or anyone else) then you'd have to read through and finish decompiling the entire source code for the game, which is an activity that would take weeks if not months - assuming you already have significant coding experience. So, you'll either have to trust the entire decomp community on that; or you'll have to do the work yourself.
Needless to say, if we discovered that the game was in some way "rigged" - that would be huge news. Knowing the people active on these threads, it would also immediately trigger a "ok so its rigged, lets find a way to exploit that".
2. The streaks achieved by users here are way too long to be possible with the claimed level of cheating by the AI
Consider just how hard the game would be if the enemy genuinely had a 30%+ chance to crit, and would double or triple the chance of secondary effects like flinching or freezing. It is not an exaggeration to say that under these conditions, we would have no streaks past 500 and probably no streaks past 200.
Something else I will add: some of the mechanics prevalent in the Frontier (Quick Claw, Brightpowder, Sheer Cold, etc) are things players have no experience with elsewhere. These are banned (or not used for their variance) in 1v1 games. They are not in the mainline story/game on regular NPCs. These (and a few other) mechanics feel like bullshit; and that's because they
are bullshit. It doesnt mean the frontier cheats with them; just that something which happens to feel like cheating is most prevalent here.
In years past, a determined counterargument may be "ok but maybe the streaks here are cheated/hacked. Well as a shameless plug, I am currently on a mission to get the record on each category, livestreamed, with all footage available (
https://www.youtube.com/channel/UC0bfFoSbJanL-9ryWWr9Y3Q/ )
A streak of 1000+ is honestly unthinkable under these conditions. It is true that the best strategies are built off 2 main ideas: 1) abusing the enemy AI and its tendencies to use certain moves/ strategies and 2) trying to absolutely minimise the impact of RNG. This RNG idea isnt about the game being rigged; its just a fact that if i play 1000 games of pokemon, one of those games is going to be the "most unlucky" game. I should expect a 0.1% chance to occur (e.g. ice beam freezing 3 times in a row is 0.1%). Therefore, a good team should have protect or counterplay against this.
3. Our (the wider community's knowledge) of RNG specifically is extensive at this point
I'm not an expert on RNG but I sometimes watch speedruns and a few videos from youtubers like Im a Blissy
https://www.youtube.com/c/imablisy
The knowledge and way we can manipulate these games is basically as deep as we want it to go with effort. But this level of rng manipulation (including RNG manipulations used for speedruns of the gen 4 frontier! -
) have never revealed secret forced-crits or forced-secondary effects or forced-enemy counterteam choices.
If the game was acting against the player in the frontier then we would obviously expect things to breakdown in the frontier; this either means different RNG results, different seeding, differences in how RNG advances in longer streaks, or whatever else. The fact is: none of this exists or has been found.
4. "It is well documented/ proven that the AI cheats in the frontier"
I wanted to start the discussion here (I didnt since the better arguments need to be nearer the top for people who dont have time to read it all) - because the most reasonable null hypothesis for the question "does the AI cheat" is "no". In other words, I want to reframe the discussion as "if a person knew nothing about the game, that person should probably assume that RNG/ luck is not selectively changed to beat the player".
But to rip the bandaid off;
there is absolutely no evidence that proves the AI cheats in the frontier. And no, being hit by 3 sheer colds in a row (something which is only 0.3x0.3x0.3 =~ 2.7% (aka not even that rare, particularly when lumped in with all the other various ways you can just get unlucky) is not evidence for rigging or cheating.
The reasons for this (the reason why we should assume the game isnt cheating) are simple;
- If you are not a programmer, it is a lot harder than you assume to "rig" the RNG in an early game like pokemon, particularly in a dynamic way that would somehow be checking the exact winstreak of the player. Lets forget that our knowledge of how RNG operates is already detailed enough to manipulate shinys and allow TAS runs to get almost whatever result they want to occur in-battle. What exactly does "rigging" mean? If I have a guts Machamp, i may prefer that the opponent burns me. If I have an Overgrow Sceptile, I may rather be hit by a crit so I can activate my Overgrow boost. This goes deeper; there are times I would rather miss a move, or rather be hit. You may think "so what, rigging rng like this should still be effective"; and yes, but just take a step back and consider the sheer scale of the number of things affected by RNG in the game which all have to be accounted for by this supposed magical "Rng rigging" code. Because if there are defects or oversights here, a dumb version of this could actually make the game
easier and not harder.
- Pokemon (when you remove the advantages the player has in-game, like items (although enemy NPCs can use items, they do not have near-bottomless revives), EVs (the gen 3 and gen 4 mainline pokemon games have 0 opponents with EV training - the player's pokemon will almost always have an advantage here even with random EVs unless you just caught them), overleveling (there really is nothing to stop the player having such a big number advantage here that it trivialises almost all strategy))
is actually quite difficult. 3v3 even moreso. It's actually pretty easy to lose a 3v3 battle, even against a kinda dumb opponent. A human should be better than the in-game enemy AI, but good enough to win all games; including the "unluckiest" game out of 100 you play? Gamefreak simply do not need to make the frontier harder than it is (the frontier features perfect IV pokemon with full EVs, bans items in battle, and prevents you from overleveling).
- Gamefreak themselves do not have a reason to rig the RNG in the frontier. Ask yourself if there is a genuine reason why this would make the game better. I would argue that the frontier is already too hard for most (particularly people not able to rng-manip IVs or who know how to breed near-perfect pokemon) and the silver/gold symbols being pretty low raw winstreaks support that gamefreak was also aware of this in internal testing. Wouldn't the battle tower be more fun if a kinda-good team could get 200 wins? Wouldn't this actually make the game better? Why would Gamefreak want to make "very long" streaks almost impossible? What does this achieve? If someone gets a 10,000 winstreak, it doesnt hurt anyone else.
I don't know what else to put in this section. If there was any kind of real analysis or study on how the AI gets more lucky than it should, I could counter it i guess. But that doesnt exist. This is entirely a rumour that outgrew itself.
5. Other communities have experience with what "real" rigging/ RNG cheating looks like
I'm guessing we have a non-zero number of Fire Emblem fans in here given the huge overlap in careful and deep strategy/ turn based gameplay.
Earlier fire emblem games have famously
rigged RNG in the favor of the player and against the enemy. (This in in the way the hit chance mechanic is implemented for those curious) This is a fact a lot of people are actually unaware of while playing (another huge point of evidence in the "humans suck at intuitive statistics" and also a huge point in the "you dont properly notice your own good luck, only the bad").
Fire Emblem games with "true" hitchance are actually hated by the community and are regularly reported as being rigged or unfair to the player. This is how we feel about what the randomness and no rigging is like to play!
6. There is a perfectly reasonable explanation for the true source of this rumour
I've put this last although I've alluded to it above. Without meaning to cause offense to players (particularly those from like 15 years ago) - a very obvious reason for these rumours are simply that the playerbase has a skill issue.
If the level of strategy someone is using does not go deeper than "use the strongest/ super effective move" and/or the deepest and most detailed part of teambuilding is "well X is weak to fire but Y resists fire" - this player is simply going to be stuck at winstreaks under 100 almost no matter what.
Combine this with a known fact of human psychology (humans are famously shit at intuitive statistical reasoning, including people who have Maths/Statistics PhDs (there is a reason why you always do things properly)) and its not at all surprising that some hypothetical player (lets call him Jim) lost at battle 80 with his team of Infernape/ Salamence/ Suicune, after his pokemon got fully paralysed twice (25% chance - 6.25% to happen on consecutive turns), thought "wow that sure does feel like bullshit, i think the game must have cheated" - and ended his frontier career there. Jim might have easily beaten the game, including the Elite 4 with almost no real hurdles. Maybe all his friends did too. but none of them get a good battle tower streak, they realise that all of their losses included some bad luck (statistically, its likely your loss will be in a game with below-average RNG) and decided that 5 separate points all reporting similar feelings is conclusive evidence.
This post isnt meant to embarrass or disparage these kind of players though. Everyone's "first experience" with the frontier is going to include getting destroyed, and if you play enough - getting destroyed by RNG that seems deeply unfair. And to be honest, between things like Band Dragonite, Scarf Rampardos, Horn drill brightpowder rapidash, Quick Claw Sheer cold Lapras (etc) - the game is already very hard the second it gets slightly luckier than you in a single game.
Something people are famously blind to its their own sources of good luck. If we play 5 hands of poker and you get lucky in 4 of them but i get lucky in one, guess which one sits most firmly in your memory and feels the most unlikely? You (people have in past studies) could even come out of that game thinking that I was the luckier player but you played better. This isnt anyone's "fault" - its just the way human brains are wired.
You don't have to like the frontier or battle tower (frankly, given how ruthless it is, how many hours it takes to create and train a team, and how many hours it takes to win enough battles - its an understandable grievance). But lets not mistake something being difficult (and perhaps overly difficult) for something being rigged or cheating the player.
