Use an engine that works in the way that you want and that you are familiar with. Some people work better with higher level engines like Unity or Unreal or Gamemaker and some people prefer working at a lower level and coding from almost scratch. I prefer somewhere in the middle such as XNA, Gameplay3D or SceneKit.

You may have your own engine / framework that you have written yourself. If this is the case make sure you focus on creating your game and not refactoring / improving your engine.

Work on a game that isn't restricted by your talents (If you can't draw then work on a game that uses abstract graphics or just cubes or squares).

Start with a simple idea and get it working then add the creative stuff later. This way your game should always be in a playable state.

Don't strive for perfectionism. If you are trying to write a quick game then your code doesn't need to be perfect.

Earlier you said:

a whole day's work as I used to do and maybe you refactored your sprite class or improved your level editor.

But now you say...

IMO engines are great for bigger projects but for something small the overhead of setting it up and learning how it works is non-trivial? That said I've not used the more modern engines, maybe they do let you have something right out of the box?

Either you're happy wasting a day tinkering with a low level detail like a sprite class, or you're not. If you are, carry on as you were before. If you're not, grab an engine where hundreds of other people already did that tinkering for you.

Of course there will be a learning curve when working with a 3rd party engine. They've done a lot of work for you and you won't know how it all operates. Luckily they also spent a lot of time writing comprehensive documentation which makes that curve much shallower than just getting a code dump. But I understand how you feel; as a coder, it feels more productive to spend a day writing low level code than to spend a day merely reading documentation about high level systems. You just have to remember that really, it isn't.

(As for 'setting it up', the situation these days is generally: 1) Download, 2) Install, 3) Click/Code. It's not like the old days of downloading a source package, downloading 20 dependency libraries, setting up your compiler to find them all, etc etc. Unity comes with everything you need. UE4 requires Visual C++ 2015 Community Edition (free). That's all.)

Where are we at in 2017 regarding game-makers... I recall 10 years ago one of the better 4E4 entrants used GameMaker or something like that. Are they still popular, is there a definitive tool people use or just loads of competing products?

Most mainstream game devs who aren't using traditional C++ tools are using Unity these days, as a powerful compromise between graphical editing tools and code-driven development. Alternatives exist, such as GameMaker, Construct, Stencyl, Clickteam Fusion, etc. But, for you, although checking out such tools is a good use of a day or two, I would probably recommend a more traditional engine that still amplifies your development speed without restricting your ability to use code to get stuff done.

Point of note: I never said I wanted to make 3D games in fact deliberately left it open... what style/scope of game to make is part of the discussion IMO.

The moment you bring a 3D game engine on board it just feels (to me) like a much more heavyweight style of game. Plus as mentioned artwork becomes a bigger deal.

Wouldn't people advocate some slick, simple 2D game based on my first post... I was rather expecting they would? I'm not targeting mobile devices but perhaps that would be a good way to think about it?

Or even browser-based to really keep focused on the gameplay rather than rich assets and fancy 3D graphics... slither.io proved how really basic graphics can be super-fun (and no I don't want to make a MMO it's just an example)

Are there good, mature 2D engines out there? I seem to recall one of the big engines (Unity?) had a 2D version back in the day but that was quite a few years back and I may be thinking of something else?

Thanks for the input all - sorry if I inadvertently dragged everyone down the 3D route. Or maybe the community here just happens to lean more in that direction :)

Most of the "game makers" listed above will be great for 2D, and yeah, Unity does 2D games just fine too.

Yup, I never said anything about 3D. Unity and UE4 were designed to handle 3D but they have built in capabilities for 2D.

"Paper 2D in Unreal Engine 4 (UE4) is a sprite-based system for creating 2D and 2D/3D hybrid games entirely within the editor. At the core of Paper 2D are Sprites (which are a Texture Mapped Planar Mesh and associated Material). You can edit Sprites inside UE4 with the Sprite Editor and create sprite-based animations with Flipbooks (which animate a series of Sprites sequentially by using key frames and specifying a duration in frames to display them)."

"While famous for its 3D capabilities, Unity can also be used to create 2D games. The familiar functions of the editor are still available but with helpful additions to simplify 2D development. The most immediately noticeable feature is the 2D view mode button in the toolbar of the Scene view. When 2D mode is enabled, an orthographic (ie, perspective-free) view will be set; the camera looks along the Z axis with the Y axis increasing upwards. This allows you to visualise the scene and place 2D objects easily. [..] Unity provides a built-in Sprite Editor [..] Unity has a separate physics engine for handling 2D physics [..etc..]"

This is a technique I employed with decent success this past summer, in developing 96MILL

I went for a short-form first-person adventure, few mechanics, photography instead of modeling.

Used the existing engine tech i'd already sunk for a more complex project.

Managed to bring it from concept to release in three months.

If I'm going to make a game as fast as humanly possible, here's what I'd do:

1) Use Unreal Engine 4. I am quite proficient at it. Never waste your time creating your own engine. I only choose UE4 because it's what I'm experienced in. Unity would be an excellent choice for other people, but I have extremely little experience with Unity, so I'd waste most of my time climbing the learning curve rather than building a game.

2) Buy all of my art assets from online market places. I don't have time to create my own or wait for an artist to spend a week making an animated character.

3) Design an achievable game with a very simple core concept. You have to be able to win AND lose and restart very quickly. Think flappy bird.

4) Build the design asap, make it rough, who cares. The key is to get something playable as a prototype. Does the design work or does it need to be tweaked? Invest as little time up front as you can so that you don't waste time refactoring polished work for something that sucks.

5) Build, build, build, test, polish, test, polish, test, polish, release, release, release. go! go! go! get player feedback asap. iterate. fix. change. tweak. release. repeat.