I'm a mostly self-taught CSE with no CS degree, with well over 25 years of professional experience. I feel that not having the degree has stunted me in some ways, but has also gave me a bit more freedom in others, because I could explore areas for solutions without realizing "how things are done".
That can be both a good and a bad thing.
Lately, I've been exploring other areas of CS which have always interested me, even as a child with my first computer, namely that of "machine learning". Where I have found I fall down in that area is in understanding the deeper level math concepts. I know these aren't really required to come up with solutions to problems using the existing toolsets for machine learning (as long as one can understand the math and concepts at those levels, which is arguably simpler) - but not having that understanding (or only a partial understanding) makes me frustrated that I can't understand exactly what is going on inside "the black boxes".
For instance, I (mostly) understand how MSE and backprop works in a neural network. I also understand why and when to use RELU vs sigmoid (or some other activation function). But could I derive any of that from first principles? Not at my current level of understanding (all those damn rules of calculus - which I don't understand). In many cases, though, I don't need to - I can treat them as a black box. But I don't like it. I do intend to fix this someday.
Of course - this subject - machine learning - has over the years led me down interesting and surprising paths (long before I started really studying it as a subject, in 2011, when I took the "ML Class" and "AI Class" MOOCs). Things more philosophical in nature, but all seemingly related in some manner, at least to my mind:
* Chaos theory
* Network theory
* Emergence and Complexity
* Theory of Mind
* Various topics in neuroscience
* Various topics and ideas in psychology
* Etc
...with "Etc" encompassing robotics, engineering, electronics, ethics, history, religion - and all the interactions and branches and spaghetti in between.
So much of that I wish I could have a more formal grounding in; I also wish I could speak with (and have the language to speak with) those who have this knowledge and grounding. I know that to be an impossibility, even if I were 20-30 years younger.
It astonishes me that many don't see how much CS in general touches and interacts (and both informs and is informed by) with all those topics and more. I see this, even if I don't understand it completely, and sometimes wonder or suspect that maybe I am wrong at some level? Maybe the lack of a formal education in all of those subjects has caused me to see things which aren't there...?
Alternatively, it could be that by not having such an education, I am - like before - not "constrained in a bubble" so that I don't see those things?
I am not saying I am special in this regard - I have read and spoke with others who have similar ideas to one extent or another, and in many cases their understandings have informed mine.
I'm just not sure if I should focus narrowly at this point in my life at 46 years of age, or go more broadly; both are fascinating paths for me. From what I have seen and read, it seems like the "broad path" would be more immersion in studying philosophy, perhaps with a greater focus on the philosophy of mind and/or consciousness.
At the same time, I like to think about and focus on the idea (and fantasize of solving it - fat chance) of the "wrongness" of backprop - and whether another solution exists for neural network learning that is more biomemetic?
...and at this point, I'm rambling - so I'll shut up.