I think my biggest problem with the "what engine should i use" threads is that they usually are great examples of Sloper's FAQ #65: "how to ask questions".
They're aksing a question that isn't really what they're looking for.
They've heard games run on "engines". They've heard phrases like "gamemaker has limitations". They've seen videos of games like Crysis and how it runs on Cryengine and think "omg I want a game that good, i need that engine.". Yet they don't know what they're looking for. They don't understand what a engine framework provides and what they have to create on-top of that. There's a disconnect between what they see in demos and trailers, and the fact that the engine is only a small fraction of why the demo looks as good as it does.
I think that's why I really like giving the advice to go modify a game. Minecraft clones could be minecraft mods. RTS games could be Starcraft2 or Warcraft3 mods. FPS games become SourceEngine mods. Then, for icing on the cake, remind them that Dota/LoL/Heros of the Swarm were all based on a WC3 mod. Natural Selection used to be a mod. Day of Defeat was a mod. Etc. etc. etc. Because many of the people asking the question can't quite see the difference between the aspects of the game they want fixed. They think that there's some special tie in between story, art, and gameplay that make it impossible to create using existing game modifications.
Yeh you're completely right.
People new to game programming (myself in cluded at one point) expect there to be some golden bullet style framework that does half the work for you.
I am a professional web developer and when I'm looking for web tools things feel more complete, there are components that will mostly solve any problem i'm having.
The issue with game dav of course is that this level of frameworking / example code doesn't appear to be there so I guess this is partly why people are asking questions like "What is the best engine?" not really knowing what the implications of that really are.
In my case, working with voxels, there's pretty much no resources out there that are both world class and cheap.
You either get a complete decent framework that can pretty solve all your problems for $100k or you get some demo type code that soon cracks under the weight of the code you try to run on day 3.
But here's the real point ...
Voxels as a concept is a done deal, a solved problem, why can't there be an open source engine out there that has the solution to this and includes the marching cubes algorithm (also a solved problem).
So when someone asks "What's the right engine for me, I wan't to build a minecraft clone" they can be pointed at that as a concrete example of how to do it but also how it can be improved upon.
When I started writing voxel engine code I was pointed at a bunch of theory and research articles or conceptual technical articles rather than something concrete which would have saved me time.
That said ...
I guess what we really need is for something like Unity to be open sourced, that would really raise the bar ... but if you had unity under your belt would you open source it?
And that's the real problem ...
Game programming is hard and the effort people put in people don't then want to just give it away!
So yes I agree the community should do it but who is really going to step up and do it knowing that all they get is the warm fuzzy feeling they helped a beginner or 2?