Open Simulator wasn't something I expected to see on HN.
Open Simulator checks off almost all of the boxes for the "metaverse". Multi-user 3D world, check. User-created content, check. Open system, check. Interoperability between servers run by different parties, check. Anybody can run a server, check. Compatibility with external asset storage systems, check. Compatibility with external payment systems, check. Second Life itself isn't as open.
So why isn't it the metaverse?
Well, there's a problem. It's not "fun".
If you start up an instance of Open Simulator, you get a blank piece of terrain. You can log in and visit it, and so can others, if you let them. Now you can build things. You can import assets and place them in world. You can sell or give parcels to others so you can have neighbors, who also can build.
This appeals to the fraction of the population which needs an outlet for their creativity. People who probably have a workshop or a drawing pad or a software development environment or a CAD program. It's not a passive form of entertainment. This makes it niche.
That may be a generic problem with the "metaverse". People don't actually go to any of the NFT worlds much. Decentraland runs about 200-300 concurrent users. (They hit 350 today!) Sominium Space, maybe 25. Most of the rest are vaporware at this point.
Attempts to make metaverses fun mostly so far involve bringing in DJs, or adding gambling.
You need something fun for new non-creator users to do the first time they log in.
You can put games inside a metaverse. That's what Roblox is. Roblox is Youtube for junior game developers. Upload something, compete for attention and revenue. Open Simulator and Second Life are a bit too sluggish for that. Mostly because of dated implementations.
Open Simulator only has one real maintainer at this point. Linden Lab has a new VP of engineering who's trying to increase performance, but he's facing a long-standing corporate culture of "can't do".
(I've been working on an SL/Open Simulator client that uses Rust/Rend3/WGPU/Vulkan, and can get consistent 60FPS frame rates. But it's nowhere near feature complete. The performance problem is solveable through parallelism and some algorithms that turn O(n) into O(n log n). The fun problem, I don't have an answer for.)
- The viewer is fricking hard to use by today's standards. I regularly introduce people to Opensim/SL and they can't figure out what the hell is going on. IT doesnt help that SL has a bonkers ancient avatar system, where people wear 2 avatars, use an alpha to hide the first and then wear a second as attachments. Good luck explaining this to people whose computer skills end at scrolling.
- People use underpowered laptops that crash all the time. Sadly it s the reality of having $1000 phones and $200 laptops/desktops
- It's buggy as hell. Hypergrid is a hack, nothing more. It's not supposed to be there, but thanks to the patience of some viewer developers they make it barely work. Unfortunately opensimulator does not have enough developers (basically only one person, Ubit) , and almost nobody is working on making a viewer exclusively for it.
- Opensimulator is a complex system, hampered by the need to emulate the SL protocol in a distributed world. The developers try hard but it's so hard to improve without a complete rewrite and a team of devleopers that can actually cooperate (there are like 100 forks, lack of leadership, and developers generally dont get along)
- There are more than enough regions -- but most are empty because of the chicken-and-egg issue (not enough users means people get bored and leave) and because its so buggy that people crash while trying to teleport.
- There is a lot of content pirated from SL that keeps people busy in opensim. Unfortunate but also true.
- There is a popular farming system , sailing system, almost every kind of animated roleplay that you can do in second life, there is the ability to script elaborate scenarios with NPCs (that dont exist in SL) . All those keep a lot of people busy (shameless plug you can find them at https://opensimworld.com). AFAIK there is nothing that it is missing from Second Life at the moment except the users. If you think something is missing, i d be happy to build it
The metaverse doesn't NEED to be for everyone, but it can certainly have wide appeal if it is easy to access and people can express themselves . In a social environment the games etc are mostly a distraction, what matters is finding users to interact with. Non-creators spend most of their time modifying their avatar.
And BTW people don't need to run a region to log in to the hypergrid. You can instantly create an account in one of the open grids like https://osgrid.org and go online, and then you can jump to any world.
> The metaverse doesn't NEED to be for everyone, but it can certainly have wide appeal if it is easy to access and people can express themselves.
If there are still technical problems both about and inside the "metaverse", why not focus on making it appealing to people with technical chops first? It surely isn't, otherwise this wouldn't be the first time I hear of OpenSimulator.
GP says it's not fun, and that attempts to making it fun involve bringing some lowest-common-denominator forms of entertainment. But what is it about it that makes it not fun for techies? Or for creatives with enough patience to pick up the necessary tech skills along the way? These people should be able to provide content, activity and root out technical problems. It's how many technological products in the past (including the Web, arguably) became popular: first becoming popular with tech-savvy people, and building up widespread appeal on that base.
It IS appealling to those people. This is a 15 year old project with only 1 active developer yet it still has thousands of users despite needing to set up a bunch of ini files and NAT loopback to run it. The problem is that it is hard even just to visit. Most people who run grids and regions are quite technical people who were in second life around the time it was hot , which is a long time ago, thats why you dont hear a lot about it. During the VR hype mania years many people left to try the new platforms until they failed. And in general people don't seem to reach outside their social media bubbles much.
Afaik there are no casinos in opensim btw. It s usually parties with wannabe DJs, real DJs and often live music
It's not appealing enough then, and I think figuring out why might be an angle worth pursuing. "Setting up a bunch of ini files and NAT loopback" does sound like accidental complexity that could be eliminated. It's the kind of stuff I did when I was a kid, and I no longer have patience to bother with; I imagine that, for the younger generation of software developers, it's way beyond their comfort zone.
Since I haven't actually tested Open Simulator yet, I hesitate to speculate on what the actual problems might be. But, for example, I've been looking for an easy way to have a persistent 3D world with optional VR support, that I could easily script to my heart's content. Something closer to Cube/Sauerbraten[0] or Minecraft, with an embedded code editor, than to Unity or Unreal Engine. Something that I could drop into, play around a bit, tweak a little bit, and come back later. I had an idea that we could make a kind of "digital twin" of our local Hackerspace, to prototype some ideas faster than in our real one. Open Simulator looks like it could be just it - so every piece of friction that makes it hard for me to set up the persistent world, build the rooms, build the doodads, script them, and get other HS members to join - I'd consider these to be adoption/interest limiters for the tech-inclined crowd.
You can do that with second life / opensim, but there is no VR viewer and the FPS is low because models are not optimized.
That said, you realize that opensim is very complex. it involves assets, streaming, simulating physics (the physics happen on the server) with multiple physics engines, managing users and groups, networking, scripting engines (it has 2 now), and is constantly trying to reverse engineer the second life protocol in order to work with he viewer which is not fully open source. It's not appealing to developers because it's a lot of work.
When it started the project received backing from IBM who developed a large part of it. Nowadays there's no corporate interest probably because it's considered unmonetizeable.
> The problem is that it is hard even just to visit.
Yes. The hard part isn't installing the viewer and logging in. That works fairly well. So does the initial tutorial on how to move and view. Then it gets hard.
There are two main new user questions in Second Life and Open Simulator: "What do I do now", and "How do I fix this %(^$* clothing problem"? It's quite possible to end up with body parts missing or invisible. The clothing system was created by users on top of the built-in avatar system, so it's mostly workarounds atop workarounds. People who master the system have the best looking avatars of any virtual world.
> FPS is low because models are not optimized.
That's a common comment, but can be overcome. The client is doing almost everything in one overloaded thread, and uses OpenGL. The code is two decades old. (I'm working on a new prototype viewer, using Rust and Vulkan. Easier than trying to work on the legacy code.)
> It's not appealing to developers because it's a lot of work.
Right. That applies both to Open Simulator, which is volunteer, and Second Life, which has paid developers.
The SL developers have a "can't do" corporate culture. (Yes, some of you will read this, and you know this is correct.) Worse, they convinced top management years ago that nothing could be done. This resulted in top management diverting resources into other projects, all of which failed after losing tens of millions of dollars.
It's also hard to get people who can work on this. The people who can do this kind of work can do other, more profitable things. The primary architect of the SL/Open simulator technology went on to do Facebook's mobile app, became VP of mobile at Facebook, and cashed out.
I would argue that some of the larger minecraft servers are much closer to the metaverse than this. They often have 'hub' servers where you can teleport to different worlds with different rulesets. Creative servers for artistic expression, survival servers for gameplay, and any number of other more niche rulesets. It's not unusual for servers to host several hundred people.
You're not wrong, but I will point out that Dreamgrid has an option for people to import pre-made enviroments (OAR files) from the outworldz server so you don't completely need to be a builder -but as far as I know, regions top out at 100 users (not that there's probably even 100 unique people on the hypergrid anyway -but I digress) so that goes back to your point regarding concurrent users.
High Fidelity got around that, but I'm not even sure if they're still around any more?
Certainly you can buy pre-made environments. There are some nice ones available.
Lots of concurrent users in one place is a tough technical problem. Improbable claimed to have solved it with Spatial OS, but it only scaled fully for limited situations, like large numbers of spaceships, and it cost too much to operate. Roblox is working on it. Their goal is a stadium of 50,000 people where you can wave to your friend on the far side and they can see you and wave back. I admire their technical ambition.
High Fidelity pivoted to spatial audio. They didn't do a big, seamless world, just something where you could have your own island. Sinespace and Sansar are other examples of that niche, which I call "game level loaders". Long loading delays of mostly static worlds, mostly running locally. Sansar did a lot of that. Ready Player One prop museum, Star Wars prop museum, etc. Turned out to be a very tiny niche. Nobody in that space got concurrent user counts past 2 digits. Second Life runs 30,000 to 50,000 concurrent users. Open Simulator, nobody knows, because it's so distributed.
I think you could take the Second Life / Open Simulator architecture, which is good for maybe 40 users/region, and get it up to 200 or so by multithreading. But to get to a thousand would need a new approach.
This is a real problem, because social 3D worlds need crowds. You can build a city, but it feels empty.
We're a couple of decades away from having compute and network infrastructure to actually have large, high fidelity crowds of real users.
That said I don't believe the statement that social, 3D worlds need crowds is true. The abundance of them without crowds speaks to that. You just don't build cities. A lot of the problem is that metaverse thinking tends to be about replicating the real world in a virtual one. Rather than thinking about how to make something better, or more exciting, or more empowering than the real world. A problem games have been solving for a long time.
They are real users. Regions with alts are removed . Most users are alone in their regions though. And these are just the regions that use this directory. You can also check the stats of popular grids like osgrid https://www.osgrid.org/infos_grid_result.php
Open Simulator checks off almost all of the boxes for the "metaverse". Multi-user 3D world, check. User-created content, check. Open system, check. Interoperability between servers run by different parties, check. Anybody can run a server, check. Compatibility with external asset storage systems, check. Compatibility with external payment systems, check. Second Life itself isn't as open.
So why isn't it the metaverse?
Well, there's a problem. It's not "fun".
If you start up an instance of Open Simulator, you get a blank piece of terrain. You can log in and visit it, and so can others, if you let them. Now you can build things. You can import assets and place them in world. You can sell or give parcels to others so you can have neighbors, who also can build.
This appeals to the fraction of the population which needs an outlet for their creativity. People who probably have a workshop or a drawing pad or a software development environment or a CAD program. It's not a passive form of entertainment. This makes it niche.
That may be a generic problem with the "metaverse". People don't actually go to any of the NFT worlds much. Decentraland runs about 200-300 concurrent users. (They hit 350 today!) Sominium Space, maybe 25. Most of the rest are vaporware at this point.
Attempts to make metaverses fun mostly so far involve bringing in DJs, or adding gambling. You need something fun for new non-creator users to do the first time they log in.
You can put games inside a metaverse. That's what Roblox is. Roblox is Youtube for junior game developers. Upload something, compete for attention and revenue. Open Simulator and Second Life are a bit too sluggish for that. Mostly because of dated implementations. Open Simulator only has one real maintainer at this point. Linden Lab has a new VP of engineering who's trying to increase performance, but he's facing a long-standing corporate culture of "can't do".
(I've been working on an SL/Open Simulator client that uses Rust/Rend3/WGPU/Vulkan, and can get consistent 60FPS frame rates. But it's nowhere near feature complete. The performance problem is solveable through parallelism and some algorithms that turn O(n) into O(n log n). The fun problem, I don't have an answer for.)