Obsidian/9999 Personal/Journal/20250331_Should_I_Continue.md
2025-04-01 02:15:38 +00:00

123 lines
6.0 KiB
Markdown

# What should I do about my PhD?
---
The quick brown fox jumps over the lazy dog. The dog stays blissfully asleep. :)
This time, I'm actually going to begin writing a journal. I feel like I have a
lot of thoughts that I let swim around, without doing much about them. To quote
*the Happiness Lab* episode that I listened to today:
"Good intentions mean nothing if they don't translate into actions."
So I'll take Dan's advice, and use writing to do thinking. I also think that
writing this journal can improve my writing skills.
So what's going on today?
Well as I'm writing this, I'm on the 61D on my way into school listening to
*Spanish Pipedream* by John Prine. But what I really want to tell you about is
the decision I'm trying to make about whether or not to finish my PhD here at
Pitt with Dan. Last Friday, I had a meeting with Dan about an idea I was
interested in pursuing that at the core involved working with a real, tangible
system. That idea was politely dismissed, and I was 'nudged' back towards formal
methods for critical infrastructure. My immediate impression is that formal
methods for our lab is a crock of shit.
Dan means well and is genuinely looking out for my best interest, but that
doesn't translate into belief into his mission. Formal methods are an intense
mathematical pursuit in order to prove 'correctness' of something to something
else. Formal methods experts may disagree with that characterization, but
ultimately that second 'something' can be a lot of things, such as a model of a
plant, a written specification, or anything that can be logically defined. Dan
wants to use formal methods to prove things about physical systems. His idea is
aligned with what Manyu just finished up: can we use formal methods to prove that
certain systems adhere to requirements using formal methods? This is connected
in part to the HARDENS report, which tried to use formal methods tools at several
layers of abstraction to prove that a written requirement can be translated into
a proof for a determined plant design. There is certainly work to be done there.
But is that work that I want to do? My brain says I could do it, but my gut
screams out a vehement no. This is the core issue--I know I could do it, but
once I would finish, where does that leave me? The answer: a formal methods
expert. Formal methods experts are highly sought after, but it is an intensely
theoretical oriented field. When I think about what I want to do in my career
and what my values are, they are not only working through a computer (despite
my aptitude for such). I want to build real things, works you can touch, and
that interact with the world. A formal methods proof about a reactor that *might*
get built is not in that alignment.
So what the hell do I do? As I write this, it seems pretty clear the relationship
between myself and the Cole Lab's work is fractured. Going forward, I see three
main options:
1) Find another PhD advisor at Pitt. Bajaj is an obvious choice.
2) Master out and go find a job.
3) Go find another PhD opportunity somewhere else. Yichen did this.
Pros and cons of each situation:
1) Find another PhD advisor at Pitt
+ I would be able to keep my NRC Fellowship
+ I know people here already
+ I've already passed the qualifying exam
- Money is garbage
- I don't think anyone is actually doing work that I really want to do
- Have to stay in Pittsburgh
- NRC debt keeps growing
2) Master out and go find a job
+ I would immediately make much more money
+ Can move somewhere else (Boston?)
+ NRC commitment is only a year
+ Could find an interesting R&D job?
+ Could get a second publication out for Bajaj's project
+ Could pick up PhD again in a year or so. Work on projects to make
myself super competitive
- Dr. Sabo is on hold
- No thesis!
3) Master out and start a PhD at another school
+ Dr. Sabo is slightly delayed
+ Could go work in Boston
+ Could go work on something I feel passionate about
+ Better degree diversity
+ Teaching plan could still work out
- Slightly longer timeline
- Have to take another qualifying exam
- Still have the NRC commitment likely
- Would not make much money
I think the smartest thing is to master out, find a job for a year that will
satisfy the NRC requirement, and then if I want to, restart the PhD somewhere
else. This would put me finishing sometime in 2030 or so. That's certainly a lot.
If I went to find a PhD now, I would finish somewhere in 2029 or so, then have
the 1-year NRC commitment. Hmmm.
I checked my current progress in my degree, and learned that I satisfy the controls
track masters degree after this semester. I would need to negotiate two things:
either ME 2046 or ME 2150 would need to count as a DSC elective, and NUCE 2100
would have to be accepted as a ME elective (it is cross listed as ME 2100, so
shouldn't be impossible to argue for). The next thing I need to do for this would
to be to talk to Dan about it.
Thinking about stopping with the Master's feels good. I think I should seriously
pursue this, and maybe talk to Rachel and my family about it.
## Chat w/ Pat
Patrick and I got lunch today. He was generally pretty supportive and gave me
some sound advice that I can use. He asked me "have you read some literature
reviews?" And the answer to that is no, not really. He also advised to "just do
something", which perhaps will help me feel like I'm actually getting something
built. I think that is helpful. Going forward, I'm going to really try and do
two things:
1) Read at least 3 literature reviews by Friday
2) Get ARCADE and I2C working by next Monday
[Could be interesting](https://www.adafruit.com/product/2264?gQT=0)
I really don't know what I should do. This certainly is not a one day decision.
Honestly, I may like the idea of option 3 the best, but I do need to do some work.
I've got it in the works to:figure out how to get the MS along the way with Antoine,
so we'll see how that happens. I think the earliest things would happen would be at
the end of the summer.