News: Mistakes found in main series Pokémon game coding

[bobandbill]

Mistakes found in main series Pok?mon game coding



Those gender ratios for Pok?mon you believed to hold true? They?ve been wrong for years. The games could be running faster too.

Read Here

 

[User19sq]

The title is slightly misleading; this article only covers mistakes from the last two generations, rather than the implied all generations. Please don't tell me PC Daily's becoming a site committed to creating click-bait. XPPPPP

 

[Melody]

This isnt even new news. It's just been difficult to prove.

Many people have observed these skewed percentages over many hours of gameplay. It's just that nobody who knows this has the programming chops SciresM has and thus saying anything but what the Major sources said got you treated like you know nothing.

Honestly I've noticed this since generation 5. Female Pokemon are truly scarce in most species because of it. If the rates were what they're *supposed to be* then female Pokemon wouldn't be so hard to come by.

I found it especially annoying that Popplio didn't have a 50/50 gender rate, especially considering how it's got a specific evolution that is very feminine looking.

 

[bobandbill]

The title is slightly misleading; this article only covers mistakes from the last two generations, rather than the implied all generations. Please don't tell me PC Daily's becoming a site committed to creating click-bait. XPPPPP

I really have to disagree with you here.

The title is "Mistakes found in main series Pok?mon game coding". The excerpt says "Those gender ratios for Pok?mon you believed to hold true? They?ve been wrong for years. The games could be running faster too." Nothing there says it's been wrong for all generations to me. The excerpt states "years", and I feel it is a stretch to assume I meant to imply every Pokemon game had these mistakes. (That said every game technically has had their fair share of errors. Especially RBY, good grief.)

Also, one of the tweets included in the article states that the gender ratio has had errors in it since gen 4 (IV).

"This goes back to Gen IV -- the off-by-one was still there, but not the wrong-divisor error. Wrong divisor error seems introduced by XY?

For Gen IV(/V? need to check) 12.5% female = 12.1%, 25% = 25%, 50% = 49.6%, 75% = 74.6%. Pretty damn amusing."

Please don't accuse Daily of clickbait when it's not happening. Point it out when it does happen, sure, but I struggle to see where this is happening here, and furthermore you claim it's an error in only the last two generations is, to be blunt, wrong.

This isnt even new news. It's just been difficult to prove.

Many people have observed these skewed percentages over many hours of gameplay. It's just that nobody who knows this has the programming chops SciresM has and thus saying anything but what the Major sources said got you treated like you know nothing.

That's news to me that people observed these percentages. Do you mean personal 'man I can't find a female starter for ages' stories, or actual trials gaining statistics of Pokemon caught and their gender and showing that they don't hold by the law of big numbers or standard statistical probability or somesuch? I know people have tried that with other things (e.g. hidden abilities for Pokemon in hordes). But detecting a difference between e.g. 12.1 and 12.5% is not that large, so I am surprised people would have noticed it with certainty without looking at the code like SciresM has.

 

[Melody]

That's news to me that people observed these percentages. Do you mean personal 'man I can't find a female starter for ages' stories, or actual trials gaining statistics of Pokemon caught and their gender and showing that they don't hold by the law of big numbers or standard statistical probability or somesuch? I know people have tried that with other things (e.g. hidden abilities for Pokemon in hordes). But detecting a difference between e.g. 12.1 and 12.5% is not that large, so I am surprised people would have noticed it with certainty without looking at the code like SciresM has.

Why yes I am referring to anecdotal evidence. It's evidence, just not the kind people will take at face value.

Few people if any have time to run statistical analysis on the games they're playing. That and you'd have to extract pokemon generation code and run it some 1,000 or 10,000 times to confirm it. (Lets assume a typical rom hacker doesn't decompile code and thus wouldn't notice the code error.)

There's not any easy way statistics can prove such a small percentage either, it will always be a matter of the error ratio being too large to detect the difference.

Hence why I said it was so hard to prove. From what we know about the PRNG in Pokemon, we know a number of values such as Trainer ID and Trainer Secret ID, as well as the Pokemon Secret Value ,Shiny Values, and Individual Values. These all in some way fall into the equations sometimes.

I can say anecdotally, that sometimes you just get an entire Save File Configuration that simply has what I consider to be a very stingy set of random seeding values; even across multiple dates, hours and times. I think the clock plays a role in number generation too, and starting on a "bad time" is also possible.

Additionally the most common PRNG, the Mersenne Twister has a few weaknesses, such as beginning it's generation with quite a few lower ranged numbers. Get a bad enough seed value and you may not get larger, "Lucky" numbers without spending lots of time abusing the RNG. Worse is how the game will reseed the RNG with saved ids, which can reset the spiral. This does make the game MORE random yes, but you can end up hugging the metaphorical floor with your rolls.