I work at another semiconductor manufacturer in the US. You could find and replace TSMC with my employer and write the exact same article.
Re: giving PhDs menial tasks and working a lot/odd hours
I've spent a lot of time thinking about it and it actually makes a certain kind of sense. From an engineering standpoint, there can be a lot variability in the amount of work needed. Some days you are on fire solving issues with the tools and some days there is almost nothing to do. Better to hire one PhD who is used to that kind of variable pace and pay them a relatively high, fixed salary, rather than 2-4 bachelors covering all shifts who will eat up more payroll, have lower peak effectiveness, and have even less to do per person. The pointless, excessive meetings likely exist to fill up the extra downtime and are easily skipped when you have real work to do (or if, realizing they serve no purpose, you come up with schemes to get out of them). If TSMC is doing the exact same thing on this front, I am even more convinced the so-called headcount shortage will never be fixed.
> Some days you are on fire solving issues with the tools and some days there is almost nothing to do
Being programmer/sysadmin, I've never ever encountered such a day. A day with very low productivity? Yes. But not that the task list is short. Are there really days when you have nothing to do within any engineering job? Ofcourse, different jobs etc, but I suspect when everything is working, you still have to do maintenance. If not, you may fill your time by learning new/effective ways of doing your work, etc.
> Better to hire one PhD... rather than 2-4 bachelors...
Speculation? Having PhD in title really makes you more effective? For sure, it's a proof you had research in some niche topic. Good if they can utilize that at work and be more effective. Non-PhDs can also do research that brings fruits to your labor.
> Having PhD in title really makes you more effective?
I think _ht0b means: If a machine needs 24/7 expert support, having a single PhD paid $$$$ to be on call 24/7 might be cheaper than having a rota of 4 non-PhDs, even if the latter are cheaper per head.
Of course, in my experience it depends on the details; if the system is so needy the people doing support can't even go to the supermarket or on a date or drink a glass of beer? That's a job for a rota of people, not a single person.
1. Physical maintenance is usually done by a separate team. There is always more low-immediate-value stuff like skill development, but actual process or procedural improvements almost always require support from multiple stakeholders. Those folks may already be prioritizing other projects, and if the process is already mature, they are likely to tell you to go away if you aren't solving a major pain point. It's honestly a lot like grad school in solid state physics. A lot of hurry up and wait.
2. Mostly I mean higher effectiveness in that one person knowing all the random details of a tool or process can, at least in the short term, be more effective than multiple people spread out over time just because it eliminates the issue of knowledge transfer. The follow-up to that can reasonably be "write better documentation," but some amount of unstructured talking is usually required to convey the full nuance of a situation. Doing that with multiple people spread across four or more shifts that don't overlap is fairly difficult.
To a limited extent, I also mean that PhDs have higher effectiveness because a PhD has a much higher chance of having delivered on process development on relevant tools while in grad school.
My experience with PhDs (at least without years of relevant industry experience) is that they are, unsurprisingly, good at research and developing early prototypes but not so good at engineering and at getting commercial products done.
So it really depends on the job and individual skills, indeed. For software dev (which is what I know best) I would definitely choose to hire several smart, down to Earth bachelors rather than one PhD.
Re: giving PhDs menial tasks and working a lot/odd hours
I've spent a lot of time thinking about it and it actually makes a certain kind of sense. From an engineering standpoint, there can be a lot variability in the amount of work needed. Some days you are on fire solving issues with the tools and some days there is almost nothing to do. Better to hire one PhD who is used to that kind of variable pace and pay them a relatively high, fixed salary, rather than 2-4 bachelors covering all shifts who will eat up more payroll, have lower peak effectiveness, and have even less to do per person. The pointless, excessive meetings likely exist to fill up the extra downtime and are easily skipped when you have real work to do (or if, realizing they serve no purpose, you come up with schemes to get out of them). If TSMC is doing the exact same thing on this front, I am even more convinced the so-called headcount shortage will never be fixed.