I only listed 6 issues, and 6 is a small number of issues in my opinion.
I clearly state in the review that I enjoy my time in VRChat because of who I'm playing the game with and what was created in the engine. I specifically stated that the game was not worth putting any money into because the software is lackadaisical. I also included a warning against people considering to develop content for the game, which—as a content creator myself—factored into my consideration.
Many games on Steam are rolling-release, and change rapidly over time. What if a game out for years suddenly included MTX? Would a thumbs down review be invalid there? If so, VRChat+ is a very recent addition to the game. Is my review still invalid?
And finally, I did state that our group is planning to switch to NEOS. Since a high majority of the players in our group are on desktop, we're waiting for the NEOS desktop overhaul scheduled to release soon. In the meantime, VRChat is our best option for these community game nights. Are we not entitled to continue using poor software if we still enjoy our time in it, and have made a pact to not spend money on it to bolster its development? Especially if we're actively telling people not to use the software or service?
VRChat has been in "Early Access" for 4 years now, and Early Access is nothing more than a tag a developer can place on their product, rather than a real moniker with any meaning. Additionally, in this review I'm comparing VRChat with NEOS and High Fidelity (and have, on certain occasions with friends, compared it to Tower Unite) which are all (except Second Life) also Early Access games—some of which are younger and less popular than VRChat. This is specifically why I include these comparisons—to prove that these issues should be unacceptable, since other services do not have them.
How are comparisons to Second Life incompatible? They're both virtual shared space applications with an enormous focus on user generated content and social interaction. Second life has been out for so long and has been so successful that it should be considered a springboard for designing a VSS app. If VRChat is floundering somewhere that SL succeeds, it proves a deficient in researching and developing VRC.
The biggest issue with 25 avatars is not only that it was previously 16 (they only upped the count recently to tease people for VRC+) but that the limit is purely artificial. All avatars aren't stored as all data for that avatar, but instead as a UUID to access that avatar from the VRC servers. It takes less than kilobytes worth of data to store 100 avatars for a single account.
The landing world includes ads for VRC+. When the service launched, the menu included 2 ads for VRC+. VRC+ is advertised every time you open the dashboard. It's pretty intrusive.
Once again, this could be solved easily with robust permissions and security settings. The VRC devs just don't seem to be willing to try for one reason or another. Perhaps when joining a locally-hosted instance, the app could display a warning like it already does for Labs worlds? Or perhaps you could only join local instances from people you trust? Again, the fact that this exists in other apps (NEOS, OpenSim, etc) means it's entirely possible to create a secure sandboxed environment for this kind of thing—even Garry's Mod is capable of such a feat.
I'll believe it when I see it.