Writing Up a PhD

I am coming up on the end of my allotted time for my PhD and need to write up. There is little content of this follow-along-journal-kind out there (as far as I know), so I thought it might be interesting to document, if nothing else for myself to look back over someday. I’ll update this from time to time, either when something big happens or changes, or when I feel I should add a bit (probably around once per week). Fair warning: this will likely become rambling and incohesive as the stress increases, and for this I ask you to please be understanding.

Welcome to the slow-motion train wreck! Enjoy!

2024-01-15

• Supervisor meeting with details for what next
• Not to worry about not having published. That will come with the write-up (hopefully).
• Need to get my example working with sequences of operations, but then I’m good!
• And then I can start writing.
• Speaking of: we put together a rough plan for what chapters there will be
  • “Hey, I think we’re actually gonna make it!”

2024-02-21

Wow! A lot of time has passed since the last update. I keep meaning to write, but was either too exhausted or had other evening commitments; such is life sometimes. Anyway!

The weeks immediately after the previous update (2024-01-05) were an emotional roller-coaster: I spent 2 weeks trying to solve a problem with my core idea. I needed to make it recursive, but the type checker seemed to stall on the final bit where the recursion actually happened.

The solution suddenly came to me and I face-palmed: I was using auto-implicit syntax (@{...}) for an implicit argument (which uses {...}). ONE SYMBOL! Having celebrated that it all now reloaded fine, I went off to teach, came back, only to discover that Vim had not highlighted my code correctly due to the code-block being bigger than the screen… I never fixed anything; the block was not commented in. That’s why it was all fine. And surprise surprise, commenting things in didn’t magically fix things. My commit from then reads:

Author: Thomas E. Hansen Date: Tue Jan 30 17:59:53 2024 +0000

[ ATMSt ] I tried so hard, and got so far

But in the end, it doesn't even matter
(Or at least, not if you forget to comment the block you're modifying
back in -_-). I really thought I'd done it...

Back to the drawing board… At least I had my supervisor meeting the next day and could ask then.

At the meeting we sat down to discuss this, my supervisor had a go and realised “huh, I see… this is trickier than I thought” (you don’t say ^^). 40 minutes of joint poking and debugging later though, we’d done it!

Author: Thomas E. Hansen Date: Wed Jan 31 18:39:34 2024 +0000

[ ATMSt ] IT FINALLY WORKS!!!

The day before my birthday as well! The amount of relief I felt now that this was finally working can only be described as immense; I now, after all these years, had a working proof of concept which showed that my idea was sound and that a solution could be implemented.

• Rewrote code from meeting to be nicer
• And also played with it to check that implementation wasn’t doing anything silly and still obeyed the same rules as the original theory.
• Tidied the trace to not include the initial state. Led to some fun oddities due to when PRNG was happening.
• Made a note on my whiteboard saying to write ~100 words per day initially (Edwin’s suggestion — “can’t edit what isn’t written”)
• Eventually managed to actually write ~100 words ^^;;
• Ported one of my blog-post-style documents on generating arbitrary Vect instances into LaTeX as part of my thesis. It’s free real estate word count increase.
• Writing is slow to get started, but then it gets going. So far, still not at ~100 words / day, things keep coming up (and writing emails takes an unreasonable amount of time.)
• Need to write a talk for SPLS in March. February is surprisingly short suddenly.

2024-02-26

• Not much thesis writing going on bc. I foolishly agreed to give a talk at a seminar on the 6th of March
  • Which is coming along much faster than I expected!
• Otoh, I have discovered some really neat ways Idris2 can help with what I’m doing, which is cool
  • failing can take a (sub)string of an error message to match, catching when you fail bc. of a typo or mismatched arg.s rather than the error you were trying to demo.
• Also discovered a misconfiguration in a test I’d written, showing that my PhD’s approach helps me write my PhD showcase itself ^^

2024-03-08

Managed to give my talk at the seminar earlier this week, which was really good. I ended up working on the slides until 8 minutes before the seminar was starting, because I suddenly also had to update the website and schedule, but it all worked out.

It is always interesting to see what people are working on, and there were a great many number of interesting talks. Simon Gay gave a talk on train tracks and dependent types which, as anything involving trains does now, made me go “Oooh! Can we do fun Factorio-like things with this?”. It was also really nice and validating to have people come up to me during the coffee break and mention that they were doing similar things in industry, but worse, due to the lack of dependent types. Or that they had some ideas for how I could generalise it. One person came up to me in the pub after the seminar and said that it was the best introduction to dependent types they’d seen and that they were really interested now, so that was really awesome to hear. All the more reason to give these talks I guess : )

Now it is time to get back to writing. On the bright side, the talk from the seminar should definitely be portable into a part of my thesis. On the less bright side, I need to actually write out the content from it, I can’t just speak it.

Oh and I want to get some info-vis-like progress meter going for my word count. So that’s what I’m procrastinating with right now ^^

2024-03-18

I’ve spent the past however many days (10 according to this changelog, aaaah!) working on a little automatic word-count plot for the top of this page. That way I can point people at that whenever they ask me how things are going : P

Not the best in terms of actually making progress, but very good in terms of fun (and learning new tools? no no: growing my skillset; yes, that sounds much better)

It took a bit of work to get going, and because I’d rather do pretty much anything other than writing my thesis, the details of that are up (in rough form) in this other blog post: auto-plotting-thesis-word-count

I even managed to get some writing done! From the talk I gave about 2 weeks ago, but now in more written form. I am still just jotting things down informally and in bullet-points; “easier to edit and expand something actually written” and all that. As part of that, I vaguely remembered hearing LaTeX had neat glossary support, and looked into that. It is pretty nice, not gonna lie, and not even that troublesome to set up (as LaTeX things go). But my enjoyment from it – being able to use a little macro whenever I wanted to refer to a glossary term and/or its definition – did again remind me that LaTeX was written by programmers for programmers to some extent; good luck convincing a regular user this is better than chords in MS Word or however that program does it.

Oh and according to the script, there are about 400-500 words more in my thesis now (no clue if that includes glossary items), so that’s pretty neat! And I’ll be even happier to see the line go up tonight ^^

2024-04-03

I got an email last week saying that my “continuation period” was now starting and that, as such, I would no longer be receiving any funding. My first reaction was “What??” – I thought I had another month before that happened – but unfortunately the admin office was able to confirm that we were indeed 6 months off September (which is when the University actually, properly come breathing down my neck) and my funding was 3.5 years, so it all added up. My supervisor is currently writing a grant application but doesn’t have an active grant, and it is the end of teaching. So now I’m burning savings, yaaay…

Oh well, as my supervisor so nicely put it: “Well, better get writing then” (to be clear: this was done in good humour, and I did/do find the comment funny).

Even with that said, finding motivation to write can be hard. Or rather, it is very easy to spend the morning writing emails and helping people out on the various discords I’m in, going for lunch, and then spending the afternoon going “I really should be writing” whilst simultaneously not getting any actual writing done. The graph is doing its job though: looming over me and publicly shaming the (lack of) work I’ve been doing.

Anyhow, I did manage to eventually get some writing done today – doubly frustrating is that once I get going, it just all comes flowing – thanks to a notebook from 2 years ago which not only has implementation details and explanations (thank you, past me) but also has a summary/birds-eye-view at the end of the section detailing how it all ties together (THANK YOU, past me!).

This, in combination with finding some early-2000s pop and pop-punk music to put on in the background, resulted in me sort of just happily vibing along and writing. Which surprised me somewhat, as I can normally not write anything while listening to stuff with lyrics: either the lyrics end up sneaking into the current sentence, or I end up just listening to the music. However, for mostly just transferring stuff I’ve already written from one medium to another, it seems to work (or at least it did today).

Also apparently I’m giving a talk tomorrow. But it is meant to be a technical one, going over implementation details, so I might just walk through the code live as the talk; won’t be the first time somebody does that at a PL (Programming Languages) Group session ^^;;

2024-06-06

Where have I been the past 2 months? I’ve been writing a TyDe'24 paper! Which was very exciting but also extremely tiring. But I am quite happy with it, and according to a quick texcount, it should net me ~16k extra words on the graph (when I eventually add it in), which is definitely worth it : )

Today I restart trying to write up parts of my thesis which I did roughly 2 years ago. Once again: thank goodness for notebooks. Even with them though, it’ll still be work to retrace my steps. Oh well, I’m sure it’ll come faster and faster the more I work through it.

I also had a look over the tiny thesis outline and drafting I’d done so far in my main thesis repo. I guess the good news are that I am beginning to see how the chapters need merging, changing, what needs to go where, etc. But the bad news are that my first thought to all the old stuff that was there was “Christ! It all needs rewriting!” To be fair, it is bullet-point stuff I wrote for a PhD-review over a year ago, so I would be somewhat surprised if I didn’t have a better grasp on things now. Well, time to get to it I guess…

Oh and I’m trying out a new pomodoro-style thing called “Plantie” – you grow virtual trees or fruit or something during the time slots. Very cute. I’ve always been highly sceptical of the pomodoro technique – something about maintaining “flow” and all that – but a friend said it really helped for larger, marathon-like tasks. Which writing up a huge chunk of work from two years ago happens to be. So I’m giving it a shot. I can see the wisdom in taking frequent breaks for those kinds of tasks, and hey, worst case I just uninstall the app and still have some amount of work done. Seems worth a try.

… Briefly popping back to say: git commits can be a time saver! I was wondering how much work I had actually ported from my notebooks. And all I had to do, thanks to how I structure commit messages, was ask git:

$ git log --grep='ch06'
commit a0104646e9effd495303584a02e134c8e323a43f
Author: Thomas E. Hansen <...>
Date:   Thu Jun 6 14:09:13 2024 +0100

    [ ch06 ] Rename to what it actually will be about

commit a9be52442b2f953f4c24e0a451f0046f78dd6b2f
Author: Thomas E. Hansen <...>
Date:   Wed Apr 3 19:04:19 2024 +0100

    [ ch06 ] Start writing up dsa-gen:edges

    There is _a lot_ to detail in terms of edges; turns out (as I
    increasingly recall) that stuff was really complicated actually! Nice!

commit a0e7af7072e451cb18f8c8f6da485080a4d1608a
Author: Thomas E. Hansen <...>
Date:   Wed Apr 3 18:04:29 2024 +0100

    [ ch06 ] Port dsa-gen overview from blue notebook

commit a0846ba7a782c1701269f62819a95a1b7d4f0273
Author: Thomas E. Hansen <...>
Date:   Mon Apr 1 17:14:42 2024 +0100

    [ ch06 ] Reconstruct mental map of blue nb

    Not a huge wordcount increase (might even be a decrease), but this was
    massively helpful in terms of remembering how things went and figuring
    out where they were written down.

Now I know where to resume my work, avoiding wasting time copying stuff from my notebooks that was already there!

2024-06-13

“When the going gets tough […]” – Aye right. More like “When it rains it pours.”

Thanks to a “minor” incompetence by Registry at my uni, it turns out the letter telling me I had a right to 12 months continuation was a template and was sent in error. Continuation can only take me up to month 48, the end of the 4th year, of my PhD, and beyond that we’re going into “You should apply for a formal extension”-territory. Our department’s Director of Postgraduate Research (DoPGR) had to reach out to Registry personally to get the correct information, and they never sent me the correction, the DoPGR had to tell me later when he realised Registry hadn’t told me directly.

Institutional communication is hard, but of all news, I really wish they hadn’t screwed up this one.

If you’ve been following this for updates, you might recall that my funding ended and my continuation started in April. What this means, is that I knew I had 6 months from April, then trusted the information I was sent from the University saying I had 12 months – calming me a bit – and have now been told that the Uni was wrong and I still only have 6 months left. From April, mind you, so at this stage it’s closer to 3 months. Oh and my flat is up in the middle of August because the Uni doesn’t believe I’ll be a student long enough for them to extend it beyond that. Brilliant. Just bloody brilliant…

“A log of the train wreck to come” is the subtitle of this post/series. Well here’s the first major derailment.

As you might imagine, all this means I’m incredibly stressed. Which means I find it hard to focus on writing, which in turn makes me more stressed.
🎶 The wheels on the bus go round and round 🎶
On the other hand, writing all of this down has reduced my stress slightly – now it lives on the page and not in my brain – and, if you’re reading this and are similarly incredibly stressed, I hope it at least reassures you that it unfortunately seems to be on par for the PhD course. It’s not just you.

To try to end this update on a positive note: My supervisor absolutely believes I can do the write-up in that time – “It is impossible for you to fail at this point. Worst case, you get Major Corrections and then that’s just another year to finish things. And most people get Minor Corrections.” – which is genuinely immensely appreciated. He has also reassured me that he can reach out to Accommodation and talk some sense into them. (Turns out not having a place to live is bad for Student Welfare and for writing up, who knew?) I continue to be incredibly grateful and lucky to have such an excellent supervisor. Another positive thing is that I got accepted for my university’s “Thesis Boot Camp”: three 10-hour days of writing, surrounded by peers equally desperate to write, with the aim of getting at least 10k words down over the span of the event. It’ll definitely be intense, but also probably good at this stage (if nothing else, just to commiserate).

I suppose “When it rains it pours” can be true at the same time as “When the going gets tough, the tough get going”. I am really not enjoying it though. The pomodoro-style timer does seem to be somewhat helpful – I’m trying to remind myself that it’s a marathon, not a sprint – as does writing the occasional rant (👋 oh hi), and I am incredibly lucky and grateful to have the support I have.

Well, time to get back to it I guess. Until the next update, and thank you for reading along so far. Good luck!