> I fear the current state of our industry eliminated the possibility for not-great, not-skilled juniors to embark in these journeys
I think both sentiments are a product of their times.
Was porting an OS to a new architecture an extremely skilled thing? 100% then and 1000% today. With each new stage of advancement and increase in the layer of abstraction away from the core metal, newer developers no longer need to know how to program at the lowest level like targeting a processor architecture directly.
Software development from the 1950s till the rise of Windows as the standard was targeted not towards systems like we do today but towards processors and architectures. Processors at that time were simpler to write for. You could get the datasheet for whatever was the latest processor from a magazine, understand it inside and out and start writing software for it. Today I do not think there are more than a few dozen people who understand the x64 line of Intel processors at the same level. So times have changed. We write for operating systems now and not processors anymore.
I think that this is neither good nor bad. It just is simply how it is. I'm sure that people who worked on computers in the 1950s at the assembly level would have been complaining in the 1970s about people writing programs in C/Pascal. And so the cycle continues.
In fact, I think that the current state of generative models that output code is the perfect scenario to separate the wheat from the chaff. Their power function nature gives a clear divide between people who worked in software for the paycheck and those who love technology for it's own sake.
> I fear the current state of our industry eliminated the possibility for not-great, not-skilled juniors to embark in these journeys
I think both sentiments are a product of their times.
Was porting an OS to a new architecture an extremely skilled thing? 100% then and 1000% today. With each new stage of advancement and increase in the layer of abstraction away from the core metal, newer developers no longer need to know how to program at the lowest level like targeting a processor architecture directly.
Software development from the 1950s till the rise of Windows as the standard was targeted not towards systems like we do today but towards processors and architectures. Processors at that time were simpler to write for. You could get the datasheet for whatever was the latest processor from a magazine, understand it inside and out and start writing software for it. Today I do not think there are more than a few dozen people who understand the x64 line of Intel processors at the same level. So times have changed. We write for operating systems now and not processors anymore.
I think that this is neither good nor bad. It just is simply how it is. I'm sure that people who worked on computers in the 1950s at the assembly level would have been complaining in the 1970s about people writing programs in C/Pascal. And so the cycle continues.
In fact, I think that the current state of generative models that output code is the perfect scenario to separate the wheat from the chaff. Their power function nature gives a clear divide between people who worked in software for the paycheck and those who love technology for it's own sake.