Last week it was my org-mode agenda, which is the system whose entire job is to tell me what to do with days exactly like this one. Some of my tasks are meant to be chipped away at a little each day, and the agenda kept filing them under Overdue with a note like “scheduled 4d ago”, even when they were perfectly on track. So I spent most of an afternoon adding an ongoing tag that gives them their own group and shows how far along they are instead, something like “day 4 · 2h10 clocked · due in 9d”. It also flags any that have gone three days without anything clocked as stalled, since the main risk with those tasks is that they sit in the agenda unchanged until I stop noticing them. Every time I rebuilt the view and watched it re-sort, it felt like a small reward. Meanwhile, what I was actually supposed to be doing that afternoon was putting in my leave preferences.

I’ve been calling this work for years because it looks like work, in the sense that there’s a text editor, a problem and visible progress. Other times I’ve called it rest, mostly because it’s what I do in place of resting. Lately I’ve started to think it’s neither and that it’s closer to escapism, which is something I only really thought about after watching a video essay about, of all things, reading for fun. What I’m trying to work out in this post is whether my desk is a workshop, where I make things worth making, or a hiding place, where I go to avoid things. I suspect the answer is both.

Escapism

The video is auranym’s Video Games and Reading for Fun Again. Nominally it’s about coming back to books after a decade away, but I think it’s really about what any of us are doing when we reach for a screen. It starts with binging The Hunger Games in middle school, goes through years of reading nothing, and then into a long stretch of playing cosy games. The part that got me was how the video describes what those games offered: “a farm to tend, a team to train, and a house to build”. In other words, a small bounded world where there’s always somewhere to be and something to do. Eventually the comfort wears off, and the video compares it to a kiddie pool, which is warm and safe but too shallow to swim in once you’ve grown. The way out, somewhat surprisingly, is through another game, a quiet visual novel about serving coffee to lonely people. It was so obviously made by a particular person that it changed how auranym approached every other game, and eventually books as well.

The video leans on a line from Ursula K. Le Guin, written back in 2001, about “the entrancing flicker of electronics” and our simultaneous longing for the unalterable, meaning imagined worlds that stay the same while everything else changes. The video’s point, which I think is the sharpest part of it, is that the two have since merged. Games are now where a lot of people go looking for a world that holds still.

Watching it, I realised my tooling maps onto those games almost exactly. The farm to tend is my config, the team to train is my org-roam notes, which I level up from seedling to budding to evergreen as I review them, and the house to build is my home server, which is quite literally a box in my house that I keep building on. Like a management sim it’s bounded and endlessly tunable, and it rewards me with small hits often enough to keep me going, whether that’s a keybind that finally works, an agenda block that renders cleanly, or nixos-rebuild switch finishing without errors.

So the obvious question is whether all of this is just escapism with extra steps.

A Person in the File

I don’t think it’s entirely escapism, and I think the video itself gives me half a defence.

Midway through, it quotes a Brandon Sanderson keynote on AI and art. His argument is that a book or a painting is in a sense a receipt, proof that a person did the long, unglamorous work of learning to make it. “You are the art,” he says, meaning the thing matters because a person is in it, and something made with no person in it isn’t meaningful in the end. He’s talking about novels in the face of AI-generated slop, but I think it’s also the best description I’ve heard of what it’s like to read someone else’s code.

The receiving end of tinkering is one of the quieter pleasures of my life. Around the end of 2024, an Anki add-on I used called Insert Symbols started greying out the review screen and occasionally crashing Anki altogether. Fixing it meant reading closely through a stranger’s code, following how they’d wired the thing together until I found the one assumption that held right up until it didn’t. The fix itself was only a few lines, and reading the code was really the part I enjoyed.

I feel the same way about other people’s Emacs configs, which I think are memoirs whether their authors meant them to be or not. Most lines in an old init.el are there because of a specific annoyance on a specific day, so reading one you can feel the person building up in the file over years. There are old versions left commented out, TODOs everyone has clearly made peace with, and workarounds with apologetic notes attached. The literate configs, which are written as prose and published for whoever wanders past, fit Sanderson’s definition of art without much squinting, since they’re years of practice laid out openly on the screen. I’ve read other people’s configs the way I’d hope someone might one day read mine, not so much to copy the elisp but to see how another person decided to arrange their days.¹

The video’s own point is that indie games count here as well, since the warm, talkative ones and the wordless, grim ones alike are clearly made by someone. Stardew Valley, the kiddie pool in question, is a good example. Eric Barone spent over four years making it alone, redrawing the art several times because his skills kept outgrowing his earlier work. So the same game that works as comfort and escape is also exactly what Sanderson is describing, a receipt for years of one person’s stubborn, improving work.

So I don’t think the tinkering is hollow. The things it produces have substance because a person made them and another person can find that person in them, and I think the video’s test for art (a person at both ends) works just as well for repositories, patches and configs. I’m keeping that part of the defence.

The problem is that this only defends the things I make. The worry was never really about the things. It was about what I’m actually doing at the desk on the first day off after nights, and whether making things is the reason I’m there or just the excuse.

The Knowable World

In 1938 the Dutch historian Johan Huizinga published Homo Ludens, which argues that play isn’t a break from culture but something culture grows out of. Huizinga describes play as happening inside marked-off places with their own rules, and lists examples like the arena, the card table, the stage and the “magic circle”. Game designers later picked up that last phrase, most notably Katie Salen and Eric Zimmerman in Rules of Play (2003), and it’s now the usual name for the whole idea. Inside the circle ordinary life is suspended and the game’s rules apply completely, whereas outside it they don’t apply at all.

The part that matters for this post is that the boundary isn’t a limitation on the game, it’s a large part of the appeal. Inside the circle the world is finite and you can know all of it. Chess has sixty-four squares and a Stardew year has four seasons of twenty-eight days, and whatever happens inside happens for a reason you could, in principle, work out.

Cosy games are magic circles with the difficulty turned down, which is probably why they suit a tired mind so well. My unorthodox magic circle is not a game but rather a NixOS machine.

For anyone who hasn’t come across it, NixOS is a Linux distribution built around declarative configuration. Instead of installing things onto the system and hoping you remember what you did, you write down in a set of files the state the system should be in (these packages, these services, this user, these dotfiles) and the system builds itself to match. If the result is wrong you can roll back, and the previous version comes back intact, selectable from the boot menu a bit like a save file. The whole machine can also be rebuilt identically on new hardware from the same files. Around that I’ve added a mesh VPN, which gives my machines an actual perimeter, so my network is a boundary I administer rather than just a figure of speech.

What NixOS actually offers isn’t just convenience; it’s also legibility. I can read the state of the machine in full, in one place, and change it on purpose. Nothing changes behind my back, and when something breaks I can usually say exactly what changed and undo only that. I should be honest that the appeal of this is at least as emotional as it is practical. Inside those files the world is small enough to hold in my head and it does what I tell it to, which very little else in my week does.

That being said, the boundary only covers what’s inside the circle, and there’s nothing about it that limits how long I spend in there. A few years ago I wrote about my attempt at a dopamine detox, which was mostly about closing off open systems, meaning sources of enjoyment that are effectively endless, like a social media feed (Meurisse 2021; Knapp and Zeratsky 2018).² I closed them either by not engaging with them at all or by setting a time limit on them. The ones I went after were Facebook, Instagram, Reddit and YouTube, and my config wasn’t on that list, although looking back I think it probably should have been. I can read all of it, but there’s always one more thing to tune, so in practice it behaves a lot more like an open system than the save-file comparison above would suggest.

No Rollback

Which brings me to the other half of the week.

A hospital is a lot of things, but knowable isn’t one of them. Medicine as a junior doctor is probabilistic, hierarchical and constantly interrupted. It’s probabilistic in that you can assess correctly, escalate appropriately and chart the right dose at the right time, and the patient can still deteriorate, because bodies aren’t deterministic systems and nobody ever promised they were. Even intensive care, which is the part of medicine most focused on control and where we titrate blood pressure minute to minute, is more of a negotiation. You can set a target, but the noradrenaline is a request rather than a declaration, and the body is still allowed to raise objections. It’s hierarchical in that what I do is decided above me and handed down, and the roster (when I work, when I sleep, and which weeks of my year disappear into nights) is declared at me in exactly the sense that I declare what my server should be. And it’s interrupted in that the pager makes sure no task has clear edges. You don’t really finish things in a hospital so much as put them down in a way you hope is safe.

There’s no rollback for a patient. I don’t mean that as a complaint, because I think it’s close to the whole moral weight of the job, and the stakes are what they are precisely because nothing can be reverted. It does mean, though, that the part of my brain that loves a clean, reversible and reproducible system spends its working week somewhere that offers none of those.

In 2009 Matthew Crawford published Shop Class as Soulcraft, which draws partly on his own move from a job at a Washington think tank to running a motorcycle repair shop. His argument is that manual work offers a kind of agency a lot of office work doesn’t, because cause and effect happen right in front of you. The bike either starts or it doesn’t, and you can see for yourself whether you got it right. He was writing about white-collar work, but I think the argument applies uncomfortably well to medicine, which is supposedly about as hands-on as professions get. Medicine has real craft in it, like the feel of a vein under a cannula, the methodical nature of a good airway plan, or a chest starting to move properly, but the junior years ration it heavily, and on most days I am unfortunately, hour for hour, a documentation interface between the patient and the medical record.

So of course there exists a significant craft impulse on coming home from work, and it ends up building a workshop. The part of me that spent a shift being interrupted, outranked and overruled by probability comes home and goes to the one place where I’m the intern, the registrar, the consultant and the head of department all at once, where the system does exactly what I say or fails for a reason I can find, and where what I build sticks around and is mine. I don’t think the homelab is an accident of personality. I think it makes up for the agency the job takes away, and I’d defend those reasons to anybody.

Which is what makes the next part harder to say.

Both at Once

That said, the reasons above only explain why I end up at the desk after a run of nights. They don’t mean every hour I spend there is worthwhile, and if I’m honest the desk is sometimes a workshop and sometimes somewhere to hide.

By workshop I mean the nights where I’m actually making something in the Sanderson sense from earlier, where a person made it and another person eventually uses it. By hiding I mean the nights where I’m at the desk mostly because a problem with a clean answer is more comfortable than whatever I’m meant to be doing instead, which after nights is usually sleeping. The annoying part is that the two look exactly the same while I’m doing them. It’s the same chair, the same Emacs buffer and the same feeling of getting somewhere, so nobody watching could tell which kind of night it is, and more often than not I can’t either until it’s over.

I think part of the reason they look the same, even from the inside, is that the hiding usually starts with something I genuinely needed to do. Back in 2022 I came across the term yak shaving, which is doing some fiddly little task that has no obvious relationship to what you’re supposed to be working on, but is linked to it by a long chain of reasons. I made a note on it at the time because it described what happens to me in Emacs almost exactly. I’ll go to capture a new task, realise the way I capture it could be improved by changing my configuration, and from there it turns into a rabbit hole. Every step along the way is justified by the step before it, so at no point does it actually feel like I’ve stopped working. Around the same time I also listed yak shaving as one of my hobbies, alongside reading, writing and the gym, which I suppose shows how much of a problem I thought it was back then.

I don’t have a system for telling them apart (more on why at the end), but there are four questions I tend to ask myself afterwards:

1. Did I actually use the thing afterwards, or was building it the whole point?
2. Does anyone other than me use it, read it or run it?
3. Does it have a point where it’s finished? Some projects are done once they work, whereas others I keep coming back to precisely because they never are, which is usually a bad sign and really just the open system problem from earlier.
4. What was I supposed to be doing when I started? This is the one I trust most, mostly because it’s the one I least like answering.

Going through what I’ve actually made with those questions, the answers are mixed. The Malleus add-on I built to help students find and create the right cards clearly has people at the other end. It only has seven stars on GitHub, but somewhere there’s presumably a student I’ll never meet using it the night before an exam. The Insert Symbols fix went out as a pull request. Three people had already reported the same error in the issue thread by the time I sent it, and the maintainer merged it two days later, so that one clearly reached the people it was for. Ramble, the voice-to-org tool I published so that something I say out loud ends up as a task in my inbox, is a softer case. It’s out there under a real licence with a readme written for whoever finds it, and it has exactly zero stars, so if anyone is using it they haven’t told me. The ongoing tasks feature from the start of this post is the awkward one. It fixed a real annoyance, but nobody else uses it.

Writing that list out, though, I noticed something a bit uncomfortable about the questions themselves. The first three are all asking whether the tinkering was useful, either to me or to someone else, and by those standards almost any hobby fails. Reading a novel doesn’t produce anything another person can run, and I don’t think anyone expects it to. Huizinga’s definition of play also includes that nothing material is gained from it, so if I’m going to call yak shaving a hobby then I can’t really complain that it doesn’t always have a person at the other end.

The problem I’ve actually written down before is a different one, which is that my hobbies run over. In a note on morning routines last year, I wrote that on the days I did have time I’d end up indulging in my hobby for too long and the rest of the day would go to waste, which I think is really what the fourth question is getting at. The first three questions are about whether the thing was worth making, which is a separate question from whether I was hiding, and only the fourth gets at that because it asks what the tinkering pushed out of the day. The ongoing tasks feature is a good example of the two coming apart, since I think it was worth making, but I still made it on an afternoon I was meant to be putting in my leave preferences.

What I keep coming back to is that the video doesn’t end where I expected. auranym doesn’t give up games to become a person of letters, and the games stay. What changes is how auranym plays them: going back to the same cosy farm and noticing the quirks in the writing this time, treating the decorating as something artistic rather than something economic, and playing with someone rather than hiding in something. So the games were never really the problem, it was the way they were being played, and I think the same is true of my config. What matters is what I’m doing inside the circle and whether I ever come back out.

What I Carry Out

The respectable way to end this post would be to turn those four questions into a system, maybe a weekly craft-versus-avoidance review with a capture template and a nice org-ql view. I can feel it wanting to be built, and I could probably have a prototype running tonight, which is exactly why I’m not going to. Building a framework for telling craft from avoidance would be the most enjoyable way available to me of avoiding everything this post is about, and it would of course live in my config.

So there’s no system, just the smaller thing the video leaves you with. auranym found the way back by returning to the same things and paying a different kind of attention to them, not by quitting. Going in isn’t the problem in itself, so I think the thing worth checking is what I come back out with. Sometimes that’s a patch or a tool somebody else can run, sometimes it’s a calmer head or a sentence I didn’t have before, and sometimes it’s nothing and another few hours gone.³

The laptop will probably be open again tonight. I’ve stopped asking whether what I make in there is worth anything, since I think it is: a person made it and another person can find them in it. The question I’m left with, which no config option is going to answer, is what I’m in there instead of.

References

\[ N_\text{beds}=N_\text{occupied}+N_\text{available}=C \]

There also exists a bed conservation principle where for every discharge, there exists an equal and opposite admission

\[ \frac{dN_\text{occupied}}{dt}=\text{Rate}_\text{Admissions}-\text{Rate}_\text{Discharges}=0 \]

However, in reality \(\text{Rate}_\text{Admissions}>\text{Rate}_\text{Discharges}\), and therefore \(N_\text{Occupied}\to N_\text{Total}+\varepsilon\) where \(\varepsilon\) represents patients in beds in the hallway and new adhoc ‘short-stay’ locations. The hospital often violates this very conservation law.

Second Law: Entropy of Information Always Increases

\[ I(n)=I_0\times e^{-\lambda n} \] Where:

\(I(n)\)

information remaining after \(n\) retellings

\(I_0\)

Original information from consultant

\(\lambda\)

Decay constant (\(\approx 0.4\) per person)

\(n\)

Number of people in the chain

For a typical consult, the chain of information is Consultant → Registrar → Resident → Intern. Here \(I(3)\approx 0.30 \times I_0\) so only 30% of the original message is kept in the consult made by the intern. The other 70% is lost to entropy.

A useful corollary explores the death of the week explores the heat death of the weekend handover. Over the weekend Day team → Night team → Weekend team → Monday team. This is why Monday ward rounds feel like archeological expeditions through fragmentary records of decisions made by people who are no longer available.

Third Law: Approaching Absolute Zero

As the number of tasks approaches zero, the ward entropy (disorder/uncertainty) of the system approaches a minimum constant value, but never reaches zero.

\[ S = k_B \ln(\Omega) \]

Where:

\(S\)

Ward entropy (in units of “unpredictable events per shift”)

\(k_B\)

1.38 (empirical constant relating microstates to macroscopic chaos)

\(\Omega\)

Number of independent ways the ward could surprise you

Components of \(\Omega\):

\(N_\text{patients}\)

Number of patients on ward

\(P(\text{deterioration})\)

Probability each patient deteriorates (≈ 0.1)

\(N_\text{admin}\)

Number of administrative systems that could malfunction

\(N_\text{discharge}\)

Number of discharge plans that could be complicated

\(N_\text{handover}\)

Number of handover points where information could be lost

Where for a typical thirty bed ward

\begin{align*} \Omega &\approx N_\text{patients} \times P(\text{deterioration}) + N_\text{admin} + N_\text{discharge} + N_\text{handover}\\ &\approx (30\times 0.1) +10 +8 + 12 = 33 \\ \therefore S &=\ln (33) \approx 3.5 \end{align*}

Which means that even with zero tasks, you can expect ~3-4 unexpected events per shift. Essentially, the third law means you cannot achieve a perfectly ordered ward (\(S=0\)) in finite time:

\[ \lim_{N_\text{tasks}\to 0} S = S_\text{residual} >0 \]

Anything can happen including patients deteriorating, eMR crashing, social issues arising on discharge, information lost during handover or random mandatory training modules. Peace in a hospital ward is thermodynamically forbidden.

Newton Laws of Ward Rounds

Fist Law: Inertia

A ward round in motion stays in motion until acted upon by an external force (usually coffee or bladders).

\[ v_\text{round} = \text{constant (unless interrupted)} \]

However, many stopping forces exist on a ward round including family at the bedside asking questions, consults coming in, consultants calls and other miscellaneous interruptions. Indeed these act as friction and the observable phenomena is that:

\[ v_\text{actual}=v_\text{planned}\times \prod_{i=0}^{N}(1-\text{interruption}_i) \]

Empirically, the value of the product of interruptions changes. It may be \(v_\text{actual}=0.4\times v_\text{planned}\) and so a ‘one-hour’ ward round becomes a two and a half hour ward round. This is not poor planning but just the physics of a ward round.

Second Law: Acceleration and Mass

The acceleration of the ward round is inversely proportional to the number of participants.

\[ a = \frac{F}{m} \]

Where:

\(F\)

Driving force (consultant urgency)

\(m\)

Mass of entourage \(\sum (\text{medical students} + \text{interns} + \text{residents} + \ldots)\)

In essence, the more people, the slower the round. This is why consultant-only rounds are fastest.

Third Law: Conservation of Responsibility

The responsibility of patient care cannot be destroyed, only redistributed. When a referral is made to another team, responsibility is now shared between the teams. When another team takes over care, the responsibility distribution shifts in favour of the new team but it is never destroyed. Even when a patient is discharged, the responsibility shifts to the care of the GP, the family and the community services.

Zeno’s Paradox of Discharge

Zeno of Elea famously proposed the Achilles and the tortoise. It is only fitting to name this discharge paradox after him.

The time required to discharge a patient increases exponentially with admission duration, approaching infinity as length of stay increases.

\[ T_\text{discharge} = T_\text{base} \times e^{\alpha \times \text{Length of Stay}} \times (1 + S + E + F) \]

Where:

\(T_\text{discharge}\)

Time until actual discharge

\(T_\text{base}\)

Baseline discharge time (≈ 2 days for simple medical patient)

\(\alpha\)

Complexity accumulation rate (≈ 0.15 per day)

\(S\)

Social issues factor (0 to ∞, typically 2-5)

\(E\)

Equipment delays (0-3)

\(F\)

Family meeting complexity (0-4)

For example, simple pneumonia, day 3: \[T = 2 \times e^{0.15 \times 3} \times (1 + 0 + 0 + 0) = 3.1 \text{ days}\]

Same patient, day 14, developed delirium, needs OT, family interstate: \[T = 2 \times e^{0.15 \times 14} \times (1 + 3 + 1.5 + 2.5) = 124 \text{ days}\]

The patient has crossed the event horizon. Discharge is now asymptotic.

The paradox remains in how to discharge, you must complete some amount the remaining tasks. Then some amount of what remains. All the while, the number of tasks to complete for discharge continues to increase. You approach discharge asymptotically but never quite arrive.

Conclusion

The hospital is a complex physical system governed by thermodynamic laws, force dynamics, and conservation principles. These laws are as fundamental as gravity—you can’t negotiate with them, you can only acknowledge them and adapt.

The physics of ward rounds is the physics of controlled chaos—where the laws of thermodynamics meet the realities of human systems, and somehow, despite everything, patients get better and go home.

Usually.

Eventually.

Asymptotically.

Generally, there was some friction to capturing tasks on the go. Typing on a phone is tedious, and by the time I’m back at my computer, I’ve often forgotten half of what I needed to capture. Capturing all tasks that came up is probably the most important aspect of task management.

Enter voice notes. They’re fast, natural, and perfect for brain dumps. But then you’re stuck with a pile of audio files that need to be transcribed and processed. This was what the Todoist ramble feature offered to solve, and I wanted a rudimentary replica of this. Ultimately, what I made works really well and allows me to even record audio from my smart watch to be processed into my inbox file. It is available here.

The Problem

I wanted to:

• Speak naturally: “Call the dentist next Wednesday, finish the quarterly report by Friday, and study pharmacology tomorrow”
• Have it automatically become properly formatted org-mode tasks with deadlines, scheduled dates, tags, and priorities
• Do it all locally (no sending my voice to cloud APIs)

The Solution

I built a Python script that:

1. Watches a folder for new audio recordings (synced via Syncthing from my phone)
2. Transcribes them locally using OpenAI’s Whisper
3. Parses natural language to extract:
   • Individual tasks from rambling speech
   • Dates (“next Wednesday”, “by February 26th”, “tomorrow”)
   • Context tags (@home, @computer, @phone)
   • Priorities (“urgent”, “important”)
   • Whether something is a deadline vs. scheduled date
4. Generates org-mode entries with proper formatting and properties
5. Archives the audio for reference

Example

I record on my phone:

Call the dentist next Wednesday, finish quarterly report by Friday tag that as important, and study cardiovascular pharmacology on Tuesday

The python script automatically creates in my inbox.org file:

* TODO Call the dentist
SCHEDULED: <2026-02-12 Wed>
:PROPERTIES:
:RECORDED: [2026-02-08 15:30]
:RECORDING_FILE: Voice_001.m4a
:TRANSCRIPT: Call the dentist next Wednesday...
:END:

* TODO Finish quarterly report
DEADLINE: <2026-02-14 Fri>
[#A]
:PROPERTIES:
:RECORDED: [2026-02-08 15:30]
:RECORDING_FILE: Voice_001.m4a
:TRANSCRIPT: Call the dentist next Wednesday...
:END:

* TODO Study cardiovascular pharmacology :@notetaking:
SCHEDULED: <2026-02-11 Tue>
:PROPERTIES:
:RECORDED: [2026-02-08 15:30]
:RECORDING_FILE: Voice_001.m4a
:TRANSCRIPT: Call the dentist next Wednesday...
:END:

How It Works

The script uses:

• faster-whisper for local transcription (no API calls, fully private)
• watchdog for monitoring file system changes
• Regular expressions for parsing natural language dates and keywords
• Pattern matching to split rambles into discrete tasks

It handles edge cases like:

• Files syncing via Syncthing (which don’t always trigger normal file creation events)
• Multiple date formats (“tomorrow”, “next Friday”, “February 26th”)
• Distinguishing deadlines from scheduled dates based on context (“by Friday” vs “on Friday”)
• Automatic tagging based on keywords

Running It

The script runs as a background service on my Mac (via launchd), constantly watching the sync folder. When a new voice memo appears, it’s processed within seconds.

The Workflow

The friction of capture is real. Every extra step between “having a thought” and “having it in your system” is a chance for that thought to be lost. Voice notes remove that friction, but only if they’re automatically processed into your actual workflow.

Now my capture flow is:

1. Pull out phone or tap record on my watch
2. Record voice note
3. Done

The tasks appear in my org-mode inbox, properly formatted, with the right dates and contexts. No typing, no manual processing, no friction. I do need to later process my inbox to ensure it gets refiled to the correct spot, but that’s part of my daily review workflow.

Here are all the addons I find useful although noting that I have more addons as well.

• A New Card Counter Speed vs Retention Forecast Graphs
• Add Hyperlink
• Add Table
• Advanced Browser
• Advanced Review Bottom Bar
• AnkiConnect
• AnkiHub
• Ankimote - remote to control Anki from your phone
• Batch Editing
• BetterSearch
• Clickable Tags v20
• Colorful Tags Hierarchical Tags
• Contanki - Controller Support for Anki beta
• Fastbar- with nightmode support
• FSRS Helper Postpone Advance Load Balance Easy Days Disperse Siblings
• Image Occlusion Enhanced
• Malleus Anki Helper Addon
• Mini Format Pack
• More Decks Stats and Time Left
• NoteSync
• Onigiri Beta - A more modern Anki for your studies by Peace
• Progress Graphs and Stats for Learned and Matured Cards
• rebuildempty some or all filtered decks
• ReColor
• Speed Focus Mode auto-alert auto-reveal auto-answer
• Study Time Stats
• Symbols As You Type
• TabOut of Cloze
• Top toolbar icons by Shige
• UpToDate Addon

Obsidian

Obsidian is an okay electron based markdown editor for note taking. My primary reason for using it over other text editors such as Emacs is its mobile and cross platform support. Mobile support remains crucial for me because, the main time I reference my notes is while at work so I wanted a more fully featured application to view my notes on mobile as opposed to using Emacs within Termux on Android. I do occasionally use Emacs to edit my markdown files in my obsidian vault using obsidian.el.

My notes are hosted onto a website using netlify (https://juniordoctornotes.com) which makes it easy to access my notes for reference on any computer while at work. In addition, if I ever want to edit my notes while at work, I could do so by logging into github and editing the repository to which they are hosted with github’s native markdown editor. Although it lacks all the features, it gets the job done.

Here are all the obsidian extensions I use:

• Advanced Tables
• Auto Link Title
• Code Editor Shortcuts
• Dataview
• Dataview serializer
• Excalidraw
• File Hider
• Footnote Shortcut
• Git
• Notebook Navigator
• Obsidian_to_Anki
• Omnisearch
• Past URL into selection
• Recent Files
• Smart Typography
• Text Extractor
• Update modified date
• Vim Yank Highlight
• Vimium
• Vimrc Support
• Zotero Integration

Zotero

Zotero is a reference manager. I mainly use it to store my textbooks and read from. I use the better bibtex extension to ensure I always have an updated .bib file to use across my applications. I also use the zotmove extension to move selected entries to my google drive or a directory with syncthing to ensure my textbooks and papers are readable on different platforms.

Calibre

I use Calibre on my macos computer to manage my books (e.g. fiction and non-fiction books). I don’t use it for textbooks and research papers (this goes to zotero; see above).

KOreader

KOreader is probably the best way to read epubs on my phone. I also use KOreader (and pretty much only KOreader) on my jailbroken kindle with great success.

Medicine Resources

Applications

A list of android applications, I find useful to study with or are useful references at work:

• MDCalc
• Eye manual
• APLS
• Nerve Block
• Emergency Procedures
• PalliMEDS
• Opioid Calculator
• Thinksulin

Websites

• Therapeutic guidelines
• UpToDate
• AMHonline
• LITFL & Part One
• Derranged Physiology
• IBCC
• ACI guidelines
• Local hospital policies, procedures and guidelines

I had a few “difficult” cannulas I had been putting off—mostly because the patients were asleep, and I am a merciful coward—but I realised there was no better use of the morning than converting these students into free labour.

Introduction

There is something deeply unsettling about the phrase, “Is it okay if the medical student has a go?” It is usually whispered while a 20-year-old is advancing a needle toward your forearm with the intensity of someone defusing a bomb, despite having only learned bomb disposal from a WikiHow article.

What we are really saying is: “Would you be willing to let this terrified youth, who has hitherto only practiced on plastic arms, pierce your tender flesh with a sharp object while I watch and silently pray I don’t have to intervene?”

And yet, this remains the core paradox of our profession. At some point, every doctor—from your friendly local GP to the “giga chads” of cardiology who “wear the hides of vanquished foes” —had to perform a procedure for the very first time on a real, living, and hopefully consenting human. Medicine has a nasty habit of requiring experience for competence, meaning the only way to become safe is to start off distinctly unsafe.

We are left with the paradox of the healing apprentice: How do we teach medicine without betraying it?

Patients as a Means and an End

Philosophy 101 tells us we shouldn’t treat people merely as a means to an end. This becomes awkward when you realise medical education essentially views patients as very complex, breathing textbooks.

There is a tension here. When a patient agrees to be a guinea pig, I feel a genuine enjoyment in my role as a bedside teacher. But is that enjoyment selfish? Doctors are often criticised for finding pathology “cool”. We might call a rare disease “exciting” or “crazy”, while to the patient it is “life altering”.

But here is the counterargument: Is it ethical to shield patients from learners if the alternative is a future filled with incompetent doctors? If we refuse to treat today’s patients as a teaching opportunity, we ensure that tomorrow’s patients suffer from our short-sighted strive for virtue.

The reality is simple. You need practice. You need the “black humour and insensitive language” that helps you cope. And unfortunately, you need to sometimes be rubbish at something, ideally while someone more experienced—like a taciturn and serene senior—is watching, ready to step in before you accidentally puncture an artery.

This doesn’t mean the patient disappears as a moral subject. It means we have to care for them differently: not as passive recipients of perfect care, but as co-participants in medicine’s grand, messy apprenticeship. As I’ve seen, some patients actually “revel in being ‘rare cases’” and join in on grilling the candidates.

And sometimes, the greatest act of moral courage in healthcare isn’t what the doctor does — it’s what the patient lets them do.

Embracing the Chaos

Medicine loves certainty in the same way that cleaning product advertisements love “before-and-after” photos. Neat, definitive, and sparkling. What we omit is that reality is the “during” photo: a blurry, chaotic mess.

We compare patients post-admission to their state pre-admission and see a definitive change. But the decisions that led there are often plagued by uncertainty. No amount of simulation or exams can eliminate the fact that medicine is often a carefully weighed coin toss taking inputs from experience, improvisation, and gut feeling. Both Daniel Kahneman (2011) and Groopman (2007) speak to terrifying complexities of thinking that makes up this job.

Added to this is the pressure of time. There is a brilliant analogy in How Doctors Think (Groopman 2007) that captures the cognitive load perfectly

Imagine watching a train go by. You are looking for one face in the window. Car after car passes. If you become distracted or inattentive, you risk missing the person. Or, if the train picks up too much speed, the faces begin to blur and you can’t see the one you are seeking. “That’s what primary care medicine is like”

I find this extends to medical practice as a whole, not just primary care. When you are an intern, the train is always moving at Mach 10. Part of what separates the experienced from the inexperienced isn’t just knowing the face you are looking for, but having the ability to decide which carriages require extended deliberation and which ones can just blur on by.

The Ontology of Expertise

But when does the inexperienced become the experienced? When does a medical student actually become a doctor? When does an intern go from incompetent floundering intern to competent resident?

Legally, it’s what the pay cheque says. But spiritually? Is it the first time you prescribe paracetamol without asking an adult? Is it the first time you get an intra-osseous access in an arrest?

The uncomfortable truth is that competency isn’t binary. You don’t go to sleep a danger to society and wake up a healer.

To explain this, I’m going to borrow from Lev Vygotsky’s “Zone of Proximal Development,” but because I’m a child, I will explain it using Pokémon.

Think of a medical student as a Charmander. Cute, full of potential, but if left alone to battle high level trainers on victory road (the Emergency Department), they will faint immediately. The Zone of Proximal Development is the battle they can win if they have a trainer using potions on them (a Consultant listening to their incomplete histories) and substituting them out for more experienced pokemon (a Registrar) when necessary.

True expertise is evolving into Charizard. You no longer need the trainer. You can handle the battle alone, but you know when to turn to your trainer if need be (i.e. true expertise is being able to handle the cases and just as important, knowing when you can’t). However, expertise isn’t just about Flamethrowing the correct diagnosis. A massive chunk of the job is untestable. It is the soft skills. It is knowing how to de-escalate an angry relative, or perhaps the most valuable skill in the hospital: knowing exactly when to shut your mouth during the morning handover so you don’t extend the meeting by twenty minutes.

Medicine is simply the long, slow grind of XP farming.

Learning as a Ritual

Medicine survives by teaching. The only thing more dangerous than a trainee… is a doctor who forgot they once were one. Every competent doctor is standing on a pile of someone else’s patience. Those students who I supervised that morning got the cannula. The student moved on. And medicine, quietly and unceremoniously, began again.

References