Well, he says "custom engine" in the article but, he used XNA. When people hear "built an engine" they tend to assume the person spent years building it from-scratch. XNA is hardly from-scratch.

I think, in this case, the author meant he didn't use one of the big commercial engines. Aside from all that, I also agree that it's not a really big mistake.

It's a great thing to do if you want to learn. Not so much if you want to release a game and make money. Even if you want some very special engine-level feature, you can find an engine extendable enough and just add what you need.

It is valuable if your goal is to learn to make games and invest in yourself. But it is not valuable if you want to make money and be successful in your indie venture.

It is possible the knowledge he gained from writing his own engine, will allow him to get a better paid job down the road for example. But most likely those $85 000 would be better spent on education.

Yes, absolutely. The world has standardized on Unreal for big games and Unity for small to medium sized games. These come with enormous ecosystems and ready-made asset pipelines, which will dramatically decrease your time to market. Unless you're sitting on a fat pile of cash and free time that you can afford to never see again after you commit it to developing some experimental new engine and tools for the same, the smart thing to do is license Unreal or Unity.

That is, of course, assuming that you think sales are THE thing that matters in indie game development. In which case you are probably better off attaching yourself to a large studio such as EA, than going indie.

What about 2D games? It seems GameMaker may have the advantage there. I've completed six game jam games and am looking into moving towards a slightly larger project and am currently considering Phaser and GameMaker, but if the 2D has come along in Unity maybe it is worth trying.

Unfortunately "invaluable information" doesn't translate to sales. At all.