I once ran a large online forum and set up a word filter to replace certain phrases with "fluffy bunnies" and other absurdities. It did a remarkable job of improving discourse once the foul-mouthed types realized their hateful messages were being massaged.
Microsoft could use some kind of super-cool AI-drived ML-based image recognition to replace the flag with a rainbow flag or something.
Really? When I've seen this done, all that happens is you end up the 'approved' word. Everyone knows what it means though. So I guess if all you care about is the word itself, and not the meaning of it, it does work great. In reality if every instance of 'shit' is replaced with 'barnacle', if I call you a 'barnacle head' on said forum, we both know what I'm saying.
Almost like a dog whistle, that only some can hear...
That you don't think you can communicate without swearing and slurs says more about you and gaming culture than it does about censorship.
Turns out it's not the word, it's the intention behind the word. And who gives intention? People. Not words.
In the end, you have to decide what outcome you want - using half effective measures hoping that the racists will keep it down, or kicking them out and making it clear that they aren't welcome? You can't have it both ways.
[0] This comic is the original source of the word: https://pbfcomics.com/comics/weeaboo
The goal is not to censor the image, it is to change the behavior of society. The goal is to punish bad behavior and set a good example.
The reason we are in this mess in the first place is that people turn the other way and sweep bad behavior under the rug instead of facing it head on and addressing it.
If we hypothesize that it would take X level of effort for MS to centrally detect and alter the graphics of every car, we can conclude it would be at most X/accounts effort (way easier) for each human player to detect and remove the same graphic from each of their own cars.
This is a silly complaint IMO. If you break the rules, whatever they are, a ban is a completely reasonable punishment. A 1- or 2- strike policy gets the same result you are talking about with no "super-cool AI-driven ML-based image recognition"
Don't forget that the replacement filter is just for banned words. Toxic/abusive behavior would be covered by the usual manual moderation. Though I don't recall ever seeing someone banned for working around the filters on the rare instances that they did.
EDIT: This is in stark contrast to normal filters which just remove or blank out words. Those were _always_ worked around. For whatever reason people feel affronted by those kinds of filters, but the silly replacement filters were "fun".
It didn't happen. The system worked fine for the 15ish years I was running it, with about 30,000 daily active users.
The offenders just got bored and moved on.