…so long as the owner and their friendgroup doesn’t dislike you.
Because it’s a discord with a solo owner, it’s prone to irrational dictation like any other group. This would not be the case in a proper professionally handled setting.
1 Like
These requirements seem a little crazy, but I guess I’ll try to apply anyways.
People can be creators but don’t have avi worlds, comm sheet, or store.
Slack doesn’t keep archives on free accounts, and commercial accounts are paid per-user something like $7/user which adds up quickly.
I will revert to the sentiment that many others have brought up; it’s a very big deal and important for technical help and information on such topics to be available in a search engine viable location; either discords that get archive copied to comfortable to read forms; or just focus more on the forum-like structure, and getting as many people as possible to see and participate in the perpetuation of the platform.
Forums usually die not because they’re bad but because nobody is made aware of their existence. People are told about discord servers on the regular, constantly reminded of them, but nobody is reminding anyone about forums like these ask forums, so nobody is here.
Discords are resulting in the burying and destruction of a lot of information, making it very inadvertently gated and ephemeral.
2 Likes
I know saying is not doing or action, but seeing all this discourse at least around resources makes me even more motivated to at least get my VRC Resources website to a point that information can begin to be aggregated and it can be somewhat useful.
I’ll try to prioritize it over the next week and hopefully I can get a working beta.
Having a help wiki is one thing; but a help forum is one in which problems are all timestamped and collectively accumulate over time as new problems are discovered, without having to wait on someone to document them all. The documentation is automatic and realtime.
(edit: also having custodians that will sometimes go around and move solved or update-made-irrelevant issues to archives or labelled as obsolete)
As some people joke: “Ben Golus posts on unity forums are the real unity documentation”
3 Likes
In that case, then yeah, I think going for a StackExchange style website would work best in general. Or at least if this forum can be modified to support that style of information aggregation/indexing. We need a place that is easy to access, known to the general VRChat audience, and is a low barrier to entry, while also verifying whether or not someone has at least attempted to create something.
The number of people that do not know about this forum, the dev updates, or even the discord announcements is baffling. A significant majority of people don’t even know they exist.
I think maybe having a forum that has comment tags where people can put stuff like the community notes twitter has to fact-check or post-mortem address things.
Can’t think of how many times where it would have been nice to be able to go and add a comment to an old dead and/or archived thread on a technical topic that i solved that threads i was looking at didn’t.
Having comments below each post the way stack exchange has, or maybe a live chat on the side for each thread where people can have their arguments/disagreements or whatever on the topics. …I don’t know if any prebuilt system like this exists might have to build one if wanting it to be ideal.
Lack of good indexing makes it easier and easier for Google to search for irrelevant or low-relevance. Moreover, even people who don’t know the relevant language well know that Google’s translation is a translation error, which makes people doubt the level of their AI. …(Google is the world’s number one search engine)
A good search system can definitely solve most problems, or a more advanced recognition system such as AI can search and solve questions to avoid unnecessary repeated questions.
Many times there are not many things that people need to do. For example, the organization and version changes of blender and art knowledge are often very fragmented. It takes a lot of time to find the problem, but the solution is very simple. Personally, it feels like more than a lot of time spent on Retrieve rather than make things.
There are also many Udon or some prefabricated functions that if they can be easily retrieved instead of going out of their way to find people who know them, it would be very tiring to rely entirely on relationships.
Although the naming of some things is interesting, it will make it difficult to find. In addition, the content introduction cannot be retrieved on search engines at all, causing people to waste a lot of time looking for it.
I’ve got a number of ideas that would help enormously. They take some effort but if done correctly would (or could) provide huge benefits.
Another discord and/or messaging system is fine but most of them are repositories of questions and answers and that isn’t typically what a person wants. We know this because people continue to post the same questions over and over again. It can be a source of input but there should be a “wikipedia” of sorts. I like wikis as a general rule but it can be any system that culls a generalized question from those asked and provides a official definitive answer. The “I tried”, “I heard” stuff from a year ago isn’t applicable in most cases. Keep the official answers up-to-date and others can simply refer the people repeating the question to the official answer.
If you want an example of incredible documentation visit .NET documentation | Microsoft Learn. I’m not aware of anything else so complete. It won’t happen overnight but it can evolve over time.
Now the big one. VRChat really needs to produce tutorial videos. From getting oriented (users including me) often walk away when trying VRC and being met with a bunch or rude kids and I don’t mean “some” I mean lots and very rude. Then getting started with world development and avatar development. Very simple “this will work” videos. Again I know that there are some but 3rd party tutorials are a) not official and b) often out-of-date and c) sometimes wrong.
Then there can be detailed videos on various subjects. Udon for instance, tooling, etc. VRChat corporate doesn’t have to produce them all but it should offer a format to follow and should review them for content. Unity doesn’t do a great job but they used to do a pretty good job offering tutorials.
And lastly, VRChat should consider in-game support/orientation and weekly video streaming. NEOS (they’re gone now) used to do it and Resonite (their replacement) is doing it now. It is a chance to get face-to-face with the VRC community and provides immediate feedback on topics of interest. It can be a challenge with a lot of attendees but there should be ways around trying to handle 1000 users (some just trying to cause trouble).
Oh… and give us access to actual REST services 
Wikis require an active maintainer, making them limited or delayed. A forum-like platform (forums, stack exchange, even reddit) is an informational database that updates itself as people inquire issues and solve them independently. A curated database is going to be less useful for unexpected new issues or innovative design.
Maybe better than a wiki or something similar, is have an AI that trains on help forums (whatever format they may be, such as traditional, stack exchange-style or reddit-like etc) so people can ask the bot about known problems and solutions. If their problem is not known the by ai, or it didn’t help them, they can then make a thread.
The bad thing is that LLM doesn’t know ‘don’t know’ Often it’s hard to consistently answer ‘don’t know’ or ‘refuse’, it’s all artificially added to limit detection.
It can be bypassed or made to produce answers that appear to be correct but are in fact problematic.
We’re talking about the latter problem, and to completely circumvent it feels like it’s enough to create a new AI market, so I’m more interested in whether AI can be utilized to make more efficient searches for existing answers, rather than letting it try to answer too much unless it’s simple enough.
Above a certain level of complexity the AI will choose to search for existing answers rather than try to answer the question (I don’t know how to effectively and consistently analyze the complexity of any question…)
I believe you are mischaracterizing wiki and wiki-like sources. They are no more “limited” than the documentation pages of any tool or library and are in fact an excellent source of information. They typically maintain a record of changes making it easy to tell what has been updated. Pages can be spawned providing more details as necessary and of course links to related topics means not having to repeat stuff over and over.
There are no unexpected new issues or innovative designs that can’t be accommodated but the important thing is nobody suggested doing away with Reddit (which I never check) or Discord. Anybody can check those sources anytime they want.
I don’t know Ben Golus and have managed to learn a lot about Unity and have answered a good number of questions (again didn’t see BG anywhere in the threads). But Unity, VRC, .Net and other platforms cannot rely on whomever sees fit to answer something.
That is certainly the worst of the worst solutions. Scan the .Net documentation and tell us how speaking to an AI model is better.
In any case my suggests are aimed at VRC management and ways that they can improve the situation. Official tutorials and guidance is a start.
I think we need to differentiate between “answering a question” and “learning a skill”. Documentation and tutorials are about learning and in that way a user can answer many of their own questions. The AI tools do not educate and anyone can (and I have) simply ask the popular AI bots a question about UdonSharp.
Both Bing and Bard produced code that would not compile and that benefits nobody.
I reply to you but that’s more targeted about the indexation problem.
It is true that using discord as a forum would make it harder to index. And that we already have this discourse forum. But why not imagine something hybrid where we could mirror the content of such discord forums as threads in discourse. That way we can link both services through a bot or something, and replicate the content of the discord forums on the website.
By so they would get indexed and accessible through any search engine.
I’m not the ultimate authority on knowledge-based systems but I’ve been developing commercial software for more than 40 years so I have some experience. So you can answer some of your questions by popping onto the VRChat discord server and reading the messages.
my friend invited me
i can’t open menu
whatcha world supposed to be?
and 1000’s of others including people calling other people stupid. No you are, no you are, etc. This is not “information” of value to a person looking for an orientation or specific answers. Search the udon channels looking for “networking” (as a new developer might do). What do you find? Not (I guarantee it) how to properly handle networking in VRC.
If you post your non-working code you will get commentary and probably get a fix for what’s wrong but again this isn’t something people search for. Their code is not the same as the person who had the problem. And what about all the screenshot images of Udon graph? Do you think that it “explains” anything to anybody? They mostly don’t work which is why it is posted.
A VRC-supported “learning channel” would answer the basic questions. Maybe a few people would discover what a variable is or how to properly check for null references to players if it wasn’t scattered over the Internet and lived instead in a format that was a) consistent, b) correct, c) translated into more than one written language, etc.
Why don’t schools and universities do the same thing? Have a bunch of students post questions and guess at the answers they heard from somewhere. Why have textbooks and professional teachers? I’ll answer that, because it has been proven to work.
Nothing stops you from writing a bot to grab all the output from all the sources and order it how you please. If I want to know the signature for the C# String.Substring method we don’t need to ask a friend (who doesn’t know all the details) but we do look it up in the official documentation. It is even versioned for us, it shows us the overloads, the return values, it shows us an example and it offers other links for additional information. This does not happen on Discord.
It’s too stationary and provides a prescribed set of information (which gets curated by someone). It’s not ideal for conversational help on arbitrary and evolving topics. People need to be able to comment and give input and feedback that is visible.
Again, they’re too slow. Someone ahs a new problem that came about with a just released patch, they need to be able to ask about it, and someone needs to be able to respond. You have to consider how many random obscure problems people run into constantly, and ones that are unique to their particular setup or situation. A wiki would be extremely unhelpful for anything other than common knowledge and common issues; which defeats the entire purpose for this.
The whole point this topic was brought up at an Ask thread was because vrchat wants to have creators able to communicate with eachother and the developers on currently developing features, not stuff that is already established. They suggested a discord, because discord is the most dynamic and flexible medium that is accessible and familiar.
The Discord is a nightmare and it’s phone gated, the reddit is also a mess. Discord isn’t searchable via google, so it’s bad for information getting buried.
Then you probably haven’t been learning unity properly. Ben Golus (bgolus) is the most common responder in the unity forums that provides unparallelled technical insight to even the most obscure of issues or design attempts. He is well heralded by unity developers far and wide. …The heehee-hawhaw is because his responses are magnitudes more productive, constructive, and beneficial than unity’s own documentation, which leaves out absurd amounts of information that many people need.
Have you never used windows forums to solve the endless slew of problems, limitations, and beneficial configurations of various facets of the os? I would never have learned my way around the policy editor or registry without the windows forums… nor troubleshooting harddrives, drivers, or many other things. Over many years, they literally brought my ability to rectify computer issues from drooling monkey to the level i have the skill to run a computer service shop (yes i’ve worked at one without any formal education; this is how a lot of these places gain their best employees and how a lot of owners started). Many of the useful technical responses are by users or power users, NOT by microsoft employees. So no, that is not correct.
Also, VRChat staff does not know everything about vrchat; the users constantly discover crazy stuff that the vrchat team never imagined or thought of or knew was feasible. Most features and solutions started as quirky hacks that caught interest. You cannot rely on them to know how to solve everything, nor expect them to have the spare bodies to do so for everyone and everything. Again that’s the entire reason vrchat wants to have a dedicated creator conversation space.
Authoritative documentation is only valid if you are talking about a coding language and every facet of it has been documented, then after that point it’s up to your own trial and tribulations; though even that is no help if you are trying to achieve a specific emergent function and seek advice for how to achieve it within the language’s syntax.
No? Having an ai trained on active support forums, documentation, and so on, would be the surefire way to give many peoples’ small common problems a quick and easy response, especially if the problem was had before, it would be able to just get it to you from a simple question or description (which it would be able to interpret). If you have more nuanced discussion, you’d partake in conversation about it, make a thread or whatever. And once the solution has been found, the ai can quickly and easily feed that information to new people.
Keep in mind language ai’s are the future of internet search, and are already starting to take over, because they’re vastly more efficient and productive than abused and antiquated algorithmic database indexed search.
Look at a wiki for something like transformers or star wars or doctor who.
A wiki is what people put into it, possibly limited by some rules.
A list of obscure problems and solutions would be helpful to anyone stuck down a well.
1 Like