I think most game styles do not fit the "open source" development model. Mainly because games often need a "surprise effect" when released and that is a poorly fit for open development and also because it is easier to contribute code than convince an artist to contribute art tailored to a very specific style.
RTS and games like Wesnoth are an exception: they are born a certain way and slowly evolve through contributions to a point where they sometimes finally became very interesting. I have high hopes for the future of 0ad.
I think highly variable, replayable, emergent games in general are a good fit. A random aspect (complex interacting systems, or rng that doesn’t railroad you) is a key component. That’s where some games get the surprise you mention. I think of paradox games, 4x, Minecraft, even Skyrim as possible good fits for open development (notably good for modding too). I don’t believe platformers, farcry type story/action, and multiplayer CoD types would make sense because the game is the same every run. Ie open development is viable when the team can be unopinionated in the way each playthrough will look.
Agree that games where art direction is strong and an important differentiator are also not a good fit (thus making a Skyrim and its AAA style probably hard to replicate in an open team)
I've been thinking about this recently -- there's a healthy mod community around the tactics game Battletech (Roguetech being the most expansive mod, but there are many others) and the modders all seem to agree that the base game is quite janky under the covers; it's not a deep simulation nor is it graphically advanced, yet it suffers from poor performance, and it's also awkward to mod in some places, since there are some things that you just can't change with mods.
I do wonder how far we are from having the default "community driven game" be implemented from the ground up in open source instead of being implemented as a mod on top of a closed-source commercial game.
One issue is that most OSS games I've seen build an engine from scratch, which is fun but a massive time sink; I wonder how much of this is due to professional engines having OSS-incompatible licenses? I noticed that Unreal Engine 4 was made open-source (royalty-based licensing, not FOSS) which might be compatible with an OSS free game?
Another aspect is that it's just a lot of work to build an A-tier game (not even considering AAA) but with a good engine this floor is being reduced over time; Battletech for example was I believe on the order of 100 developer-years of effort, which is huge for an open source project. But could you get 80% of the way there in 10 developer-years if you strip out some of the features the mods eschew like story content? Plus, with Patreon, these days it seems viable for a small number of developers to work full-time on niche content like this.
In summary, I'm surprised at how few OSS games there are considering how much effort goes into modding. Does anyone in the industry (or otherwise) have insight into what's holding us back here?
> In summary, I'm surprised at how few OSS games there are considering how much effort goes into modding. Does anyone in the industry (or otherwise) have insight into what's holding us back here?
I'm not in the industry, but I'm a programmer who's tried and abandoned various game projects.
For me, it always comes down to the fact that game design is a deep and totally different skill set that I don't possess. You can't really wing it, or you can but you're unlikely to end up with a game that's fun and balanced.
Same thing can be said for why so few OSS projects have solid documentation or excellent graphic/UI design.
EDIT: I'd also guess that's why the few OSS games you see are usually clones of more popular ones, as is the case with Widelands. By cloning and iterating on a proven game, you don't have to playtest, refine mechanics, etc.
I can heartily recommend BattleTech Advanced if you don't want to go as far as Roguetech. It's definitely tougher, but not as impenetrable as Roguetech is. (Plus there's lots of difficulty sliders so you can probably tweak it to be about as forgiving as vanilla).
I really enjoy the MWO-style mech customization that both of these add (the https://github.com/BattletechModders/MechEngineer mod); one critique I have of vanilla Battletech is that they simplified the mech construction mechanics a lot, and stripped out some of the tradeoffs that make mech design interesting in the canon. In vanilla, you just optimize for free tonnage, and can put jump jets on everything, so there are just a few chassis in each weight class that are strictly optimal for most roles; this makes it quite boring as there's little reason to pick other chassis. In BTA/Roguetech you can take an XL engine to shave off a lot of weight, but that means your mech will die if you lose your side torso. Lots of meaningful trade-offs, with no "best solution".
I will admit that my eyes glossed over when I first started the game and looked at all the new part types I had to learn though...
I quite like Endless Sky (link: https://endless-sky.github.io/ ), an open source Escape Velocity-style game, and Supertuxcart. But I think in general you're right, the vast majority of finished, fun, playable games are going to be commercially developed.
Not only they need a "surprise effect" at release, but once you played it enough you get bored and move to something else. Harder to keep contributors engaged.
Unlike tools (i.e. text editors) that you can keep using for 20 years+ and never want to stop using it.
What would be great is if they'd become open source after a certain amount of time. Then there could be a nice release, with the surprise element and everything, and later when the steam runs out, the community would pick it up if it still has potential, and do something else with it.
In 2010 Notch said he'd probably release Minecraft as open source within a few years. I was rather looking forward to that.
Jason Rohrer releases all his games as open source. So there are mobile ports, spinoffs, etc. With the latest game you basically are just paying for a multiplayer account, the whole game is on GitHub.
While it's not quite Minecraft, there is an open source game (or rather, game engine) in a very similar style named Minetest[0] with a diverse community and a variety of game designs and mods to choose from.
There is no scarcity of open source Minecraft imitators/clones, but Minetest seems to be among the most thoroughly developed. Veloren has some interesting demo videos, for example, but its modding interface is barely a sketch at this point. (The plugin API documentation covers interaction with the chat window, and events for players joining or leaving… and that's about it. There is no way to plugins to actually interact with the game.) A closer match for Minetest would be Terasology[0], which has a modular design based on Java rather than Lua.
Yeah true - in particular for new / innovative games.
Whereas open source versions of old popular games seem to do very well, because it's less about the surprise and more about the nostalgia. See openra for example.
> it is easier to contribute code than convince an artist to contribute art tailored to a very specific style
I sometimes wonder if modding scenes draw most of the volunteer artistic talent that might otherwise be available to open-source projects. There have been some pretty impressive assets created for Skyrim or GTA V mods, for instance.
I think the bigger pull is because of the bigger audience. Without the modding scene, I don't think that would necessarily make open source games multiple times more popular, even if they had some percentage of the artists making them content, because popularity seems to be highly influenced by current trends, and slowly advancing open source games don't seem to fit in with the trend model vary well.
No matter what the next Elder Scrolls or Fallout game is like, as long as it supports modding (of course it sill), there will be a bunch of people playing it (even if it's widely panned), and there will be a bunch of mods and talent making art for them, because that's how you get tens or hundreds of thousands (maybe even millions if everything aligns) of people using what you make.
I do agree on the need for a surprise effect to make it fun to play, but surely the number of developers who are in on the surprise is so vastly outweighed by the potential number of players that the open development model should still work for the 99+% of the audience who aren’t part of the development?
If I find a simple clear bug in code I can easily fix it. If I feel something odd on a painting it is not easy to even tell IF something is wrong, then what is wrong, then how to fix it.
RTS and games like Wesnoth are an exception: they are born a certain way and slowly evolve through contributions to a point where they sometimes finally became very interesting. I have high hopes for the future of 0ad.