Trying to be somewhat introspective here, I think there are two reasons why I rarely end up taking people on this site seriously.
First, because of the past and present history of the site. It's still mainly about warez, overtly or not. I doubt you'll ever shake off the "gbatemp are a bunch of warez kiddies" reputation.
Second, because my impression is that most of the people doing work on this thread don't know what they're doing
and don't realize how far off they are, or don't care about it. This kind of approach is so far off from the way I and my colleagues work that it's baffling. I work step by step: in order to achieve a goal, I start at my current level of knowledge, use it to build a solid base to work on, and then try things out on top of it as I learn them. I don't use tools until I understand how to use them, and, ideally, also understand how they work (this is how I ended up with detailed register-level documentation on Wiibrew for a bunch of Starlet hardware: because I did extensive probing in order to understand exactly how it behaved). I'm also curious: my primary goal, often, is not to achieve something, but to learn more about the system (and I might end up being able to achieve something as a result of said learning). What I've seen from this thread is people starting with a goal that is way above their heads, then cranking out some hypothetical code without really knowing what they were doing or how the tools that they're using work (the first version with inline asm, honestly, was an atrocity in many, many ways), followed by a bunch of suggestions about how to fix it without really fixing the underlying problem: that you guys need to learn a lot more in order to really understand what's going on and subsequently be able to write intelligent code to solve the problem. Don't use tools (e.g. inline asm) until you know how to use them properly (e.g. constraints). Don't write PowerPC bringup code until you know how things like caches and memory management work.
Edit: A good tip is to learn properly is to never, ever, EVER copy and paste code. Do not use anyone else's "solution" to a problem. Write your own, and make sure you actually
understand what you're doing, why you're doing it, how the code actually achieves the goal, and whether there are any potential pitfalls that you need to consider.
Mind you, a bunch of people who don't know what they're doing or where they're going can still achieve something successful. The prime example of this is probably
PHP, but firmware and software written by makers or hardware engineers with little to no expertise writing code also comes to mind - typically it ends up working for some value of "working", but it tends to be an utterly horrible unmaintainable piece of spaghetti code that needs to be rewritten to evolve sanely. The converse is also true (hardware designs by software guys).
Make of that what you will. Also, I want some of what Kdonix is smoking.