What is the line between engines and games exactly?

[Jastiv]

I know a lot of developers say they want to make games, and end up making engines instead, or vice versa. So, I've been reading the minetest forums (and not playing the game) and I think about engines, and how they are designed to be used by multiple games but if the engine is too inflexible, then you can't make the type of game you want on it, on the other hand, if it is to flexible, then it is hard to mod, and newbie users don't like struggling to make games for it.

I'm also not exactly sure how engines are different from straight up forking, ie you take the code, and take it entirely in a different direction ie, an easy game for newbies with cute graphics turned into a hardcore game with realistic graphics (or vice versa)

[dulsi]

I don't think there is a hard line between engines and games. Any game with sufficient configuration in data files could be used as an engine to build another game. How much of the configuration is in the data files would determine how different the game could be. Some people even use game without separate data files as an engine by modifying the executables but that is usually rather limited.

Engines can be designed to run multiple games. Like how ScummVM can run a variety of graphical point and click games. Or you could expect people to take the engine and perhaps customize the code itself essentially forking the project. I'm generally not a fan of forking a project. I'd rather try to work with the project to add anything you need but I understand that is not always possible.

[anon666]

some games are engine-less
the main difference is that the engine is reusable and it can be somewhat separated from the game itself, for instance you can test it with other dataset/testdata or even run other games on it
really universal engines are frameworks

[Lyberta]

The fact that people still write new engines shows how bad current engines and programming languages really are.

[Danimal]

I differ Lybertas, normally people still write their own engines because they want to study how such a complex program is created and works, that people normally are computer engineers (or related students). But they are as well the ones that push new technologies and languages around, the ones that evolve current status quo of things, so in that respect you have reason.
There are plenty of awesome and relatively free game engines around that give people with not much knowledge the chance to create their own games; Godot, Unity, Flare... you just need the will to persevere to make something.

As for the main topic, the line is making something out of that engine, along the years in the libre community i have seen countless persons start making engines from zero. But out of all of them, only one reached the goal and delivered a game that went commercial and became an indie success. Im talking about Kenshi, i saw the creator first post nearly 15 years ago on the Ogre3D forums. So, if you start fussing how this or that could be done, you are just creating another useless engine or wont do nothing at all. A game is just an engine with beautiful furniture and a story to go along.

[Lyberta]

only one reached the goal and delivered a game that went commercial and became an indie success.

If you mean that the game is proprietary, that's an epic fail, not a success.

I differ Lybertas, normally people still write their own engines because they want to study how such a complex program is created and works, that people normally are computer engineers (or related students). But they are as well the ones that push new technologies and languages around, the ones that evolve current status quo of things, so in that respect you have reason.

I started my own engine because I wanted to test the network protocol for a mod I was working on. But as original mod died I kept developing the engine to see if I can make an engine that fits my criteria of "good enough". In particular:

1. It must be free software.
2. It must be written in a sane native language.
3. It must allow developing game in a sane native language.
4. It must follow "every player is a game developer" philosophy. So that every server can run mods, including running code on the client, and these mods are downloaded automatically.

There were/are no engines that satisfy that. Cube 2/Red Eclipse came close but failed 2 and 4. DarkPlaces/Xonotic also came close but failed 2, 3 and (to an extent) 4. So for the last 10 years I'm trying to get something that is good enough but now I'm bumping against the limitations of C++ itself. Maybe migrating to Rust is a solution but I will have to spend another 10 years rewriting all my code which is not fun. And maybe Rust has fatal flaws too, so I'm not in the mood to explore it right now.