a) I cannot effectively review more than 2000 lines of code a day. The LLMs can produce much more than that.
b) Even if I accepted my reading throughput limitations as the cost of being in the loop, reading is not enough to keep cognitive debt in check: my skills will atrophy if I do not participate in the writing ("What I cannot create I cannot understand").
So, to me, it seems like we, humans, either have to come up with higher (and deterministic) abstractions than code to communicate with LLMs or resign ourselves to letting the LLM guess what we want from English and then banging on the output to see if it sort of works. This later state of affairs seems to be what the current trend is and I find that absolutely revolting.
Yes, it is a common psychological defense. Rather than suffer the pain of inadequacy one either decries the whole idea of merit as meaningless/harmful or adds enough mushiness to the ranking method as to render it useless. Either way your envy subsides and you can back to living a professional life of un-examined mediocrity.
This is an amazing book, the engineering, bottom-up counterpart to the equally marvelous, but top-down, mathematical SICP. Between these two a serious student can get a wonderful foundation in CS and a deep and hopefully enduring connection to the beauty in it.
It changes. When you don't understand the problem, yes, simplicity is your best bet. Keep the code clean, clear, because you are going to have to change it a lot. As the understanding of the problem increases, and this does not always happen since sometimes the problem changes too quickly to be understood, then a good architecture can make an enormous difference.
This is something I sometimes also ponder. I think the answer is yes, as long as you keep your ultimate business goal in mind and optimize for workflow rather than looks. Yes, you do take longer to get going but once your setup is honed in you gain a little more time every minute you use your system.
a) I cannot effectively review more than 2000 lines of code a day. The LLMs can produce much more than that. b) Even if I accepted my reading throughput limitations as the cost of being in the loop, reading is not enough to keep cognitive debt in check: my skills will atrophy if I do not participate in the writing ("What I cannot create I cannot understand").
So, to me, it seems like we, humans, either have to come up with higher (and deterministic) abstractions than code to communicate with LLMs or resign ourselves to letting the LLM guess what we want from English and then banging on the output to see if it sort of works. This later state of affairs seems to be what the current trend is and I find that absolutely revolting.