Good evening, I am one of the major collaborators to the effort, and directly responsible for implementing the filtering system onto the server. I am here to clear any misinformation about this and how this works.
To begin with, there are 2 ways the game does filter chats.
- Client sided filtering
- Server sided filtering.
Client Sided Filtering:
- To save on processing and data usage, Jagex in 2004 would automatically filter chats that YOU type into your own game client. This is because the client is already aware of what you typed and can just already do the filtering. Because of this, the filtering code is written as-is from Jagex themselves since this code is packaged along with the client, which came directly from them. So basically what I am trying to say is, everything you type on your end, is filtered by their own code. If you were to disable the filtering code, only you would be able to see the unfiltered text on your screen.
Server sided filtering:
- This is basically just the same as #1 but we copied and pasted the exact same code from the client, over to the server. This is because we need a way to filter your chat to all of the other players around you. This point is not very important but figured it was worth mentioning.
To be completely transparent, we have not really done comparisons of filtering to any screenshots, because there isn’t any reason to, since it is already just their code. However I can confirm I have yet to see any inauthentic filters from ones that I have randomly compared to, or that players “swear this is not right”.
You can view how the code works, on both the server & client. Warning, it is pretty complicated.
WordEnc
WordEncBadWords
(there is more)
And yes it has changed a many times over the years. More words got added between 2004-2006. Around 2007 a bunch of it got reworked when they started compressing the chats with a Huffman tree… and it just kind of goes from there.
As for the list of words and such, all of this comes from the game cache, which also comes from Jagex. And it is completely unmodified, because modifying it would completely void the entire point to 2004Scape. We have CRC checks to make sure inauthentic changes do not make it through, and I personally review pretty much every single PR that gets opened on GitHub. I can guarantee to you that if you were to follow our exact steps, starting with the original cache, you will end up with the exact same list of words/filters.
TLDR:
Yes it is just that bad.