The penchant for solving problems that nobody else has but are potentially massively socially useful is actually the entire reason science jobs are compensated so poorly.

You talk to many scientists (myself included) and they'll tell you they'd do the job for free because of how interesting it can be, provided that they were comfortable otherwise. This is well known by the suits, and so, wages trend in the direction of zero.

The problem appears when people aren't paid enough to fit their definition of comfortable. People in science are highly educated by default, and their definition of comfortable takes that into account. This is why people leave science and don't look back. They can be comfortable doing something else, and it only gets more appealing over time. In your 20s, slumming it is to be expected, but beyond that it starts to get awkward.

A postdocship worth 50-60k only happens after 5 years of school (making 18k-25k) working 12 hour days. Realistically for the social utility of science and the sheer effort and difficulty, science should pay 2-3x as much as it does-- but frankly, our society doesn't value science in the way that it did during the Cold War.

Grad school stipends have gone up, not down (was 34kish at MIT last year, although Cambridge is expensive). Public schools usually have smaller stipends. 12 hour days is a bit of an exaggeration too (although it happens sometimes). I get that other fields with similar experience pay more, but we're not exactly living on ramen...

speak for yourself (or for other top-10 schools) - i worshiped my rice cooker for filling my belly day in day out for 50c a pop. i got about $10k per year and even in the mid-90s that was !#@!@# all.