R3C0D3R
Before stating my opinion, I want to note that I've met new users for years and poured my effort into making people get used to VRChat. Since 2020, I've been holding a user event that guides new Korean users every other week. Also, I'm the creator of the world named "VRChat 한국어 튜토리얼," which is a community tutorial map for Korean speakers since VRChat doesn't support localization on tutorials.
While explaining this game, the most complex concept to explain is user rank and safety settings. For new users who do not know game development, it is hard to explain what audio, animation, and the shader is. Since "Safety setting" is one of the most critical settings to be acknowledged to every user to avoid malicious content, the safety setting is hard to understand and does not work well.
First, a safety setting based on the user trust rank system does not act as a "safety" feature. Most malicious users who used crashers / NSFW / offensive avatars while on user-guiding events were either "Trusted users" or "Known users." The default value for those two ranked users' avatars can harm others if they want. Since user rank is assumed to be determined by content creation and friends, we don't know if this user created malicious content. Also, on my friend list, more than 60% of users are trusted users(even if I add more new users from the user guide event); I wonder if the trust rank is effective.
Two. While using the current system, it is impossible to protect non-adult users (13~19) from adult / offensive content. There is a user report system, but it doesn't protect non-adult users directly from NSFW content. Maximizing shields also does not work since disabling shaders and only showing avatar mesh can have NSFW content. There is an option to set one's avatar as "NSFW" when uploading, which is quite a useless option since it doesn't change anything in-game (and is not forced), so nobody cares about that option.
I suggest more simple safety settings that protect users. I'll give a quick example. Please see the included image. (This is only an example; I think devs can find a better way to express this.)
Settings to protect oneself should be easy to use and easy to understand.
TL;DR
  1. Current "shield" system doesn't work properly.
  2. "Trusted Users" and "Known Users" can harm other users, but the current shield system can't protect them.
  3. Protection for non-adult users is required.