I guess the key phrase there is "fairly successful". I'm struggling to think of a game that has been an outright hit that has been built on top of unity.

Depends on how you define successful. Unity games tend to be built by smaller teams and therefore don't need to sell millions of copies to be a financial success like big studio games do.

I mean Indie games specifically. The most successful indie games I can think of (e.g minecraft / braid) didn't use any particular monolithic engine.

What I'm wondering is whether this is simply a co-incidence or whether it's that the most successful Indie titles are ones that take big risks in terms of introducing wildly different gameplay and perhaps using a monolithic engine like Unity makes building such a game more difficult because the engine itself will be inherently opinionated.

In other words, if you make a decision on which technology to use before you have a complete game design are you in fact reducing your risk/reward without necessarily realising it?

It depends on how you formulate your risk model.

The industry titans can afford to develop "any" amount of custom tech, subject to schedule constraints, since the art and marketing budgets tend to vastly outweigh the engineering now. But in the last decade, they've tended to shy away from really ambitious tech projects, favoring incremental improvements to a general world-simulation model that applies across all the typical AAA genres. In their risk assessments, they've decided what kind of product and technology is being made early on, so that the bigger expenses aren't in danger. As well, sticking with the same toolset allows it to be further refined.

Indies, on the other hand, have a lot to gain by taking on ambitious technology and de-emphasizing the costly assets - and usually, the technology itself isn't risky, so much as it is just an "unknown engineering challenge" - something that could be a few days or a few months. Something like Braid or Minecraft can be achieved by experimenting and chipping away at the concept over a few iterations. Minecraft's earliest versions didn't have a lot the "big features" that are in the game now.

But it's also incredibly scary to take on those projects all alone, so the majority of indies aren't even going to consider it. Indies are looking for "easy" too, even if it's bad for them from a business standpoint.

I do think that engines like Unity have a lot of value and can even be extended to include some new technology. (I started working on something that builds on Unity's physics system recently - and it shows some promise to be a unique product) But the people who are taking on the unknowns just tend, statistically speaking, to be the people confident enough to write everything themselves too.

>perhaps using a monolithic engine like Unity makes building such a game more difficult because the engine itself will be inherently opinionated.

KSP is actually a good example of this as well. They've had to make a few design decisions because of engine limitations. But it's also let them get a game that's quite good that's developed a dedicated following fairly quickly.

Wasteland 2 would also be a project to watch. They're one of the highest profile projects to go with the engine.

The iOS game "ten million" was built on unity as I recall. It has been kind of an indie hit.