I don't think the ratio of bad:good new developers varies much by the stack, but I think some stacks push developers "up or out" more than others. Browser frontend has some of the least pressure here because the performance problems are generally paid by someone else's computer.
Backend inefficiencies traditionally will eventually saturate your RAM / CPU / I/O ("clouds" removes some of this but you still pay in the end - they also increase pressure in other areas since you're now always running alongside other things). In the browser you just need to be faster than the user's expectation or need, and user expectations are today at record lows and needs at highs (more and more "daily life admin" is web-only or web-first).
I also think "tribe 3"[0] developers are both more likely to gravitate towards the frontend, more likely to not think about performance, and (definitionally) more likely to not want to learn about fundamentals affecting performance.
My impression is opposite to yours, perhaps due to the type of software development we do.
I can't remember how many times a performance problem was "sorted" by increasing CPU and RAM limits on a cloud instance. It just so happens that it's much, much easier to do that on the cloud instance than on a user machine.
Of course performance problems do show up that need to be solved, just way less common than just spending more resources.
What I do agree with you (and I have no data on this just personal impressions that are most likely affected by observation bias, as I suspect is the case with you as well), is that in these waves of new programmers not going through the normal university route (which is absolutely perfectly fine in my book), they usually start off with front-end development these days, skewing the experience level ratio to more entry level people, and thus making entry-level mistakes. Perhaps 10 years ago the entry tech was Rails/PHP.
Backend inefficiencies traditionally will eventually saturate your RAM / CPU / I/O ("clouds" removes some of this but you still pay in the end - they also increase pressure in other areas since you're now always running alongside other things). In the browser you just need to be faster than the user's expectation or need, and user expectations are today at record lows and needs at highs (more and more "daily life admin" is web-only or web-first).
I also think "tribe 3"[0] developers are both more likely to gravitate towards the frontend, more likely to not think about performance, and (definitionally) more likely to not want to learn about fundamentals affecting performance.
0: https://josephg.com/blog/3-tribes/