Written by 
Ever get home, tired after work, sit down on a couch, and spend an hour or two sitting down without even managing to change into your home clothes? It’s a seriously unpleasant in-between state – almost comfortable, but you know you’re not really at rest, likely hungry, and even your phone battery is likely about to die. This kind of tiredness can get self-reinforcing real quick – especially if you’re too tired to cook food, or you’re stuck in an uncomfortable position. It’s like the inverse of the marshmallow test – instead of a desire, you’re dealing with lack thereof.
I’ve been dealing with this problem a lot within the last two years’ time. Day to day, I could lose hours to this kind of tiredness. It gets worse when I’m sick, and, it’s gotten worse on average after a few bouts of COVID. It’s not just tiredness, either – distractability and tiredness go hand in hand, and they play into each other, too.
My conclusion, so far, was pretty simple. When I’m tired, delayed but proper rest is way better than “resting” in a half-alert state, even if that takes effort I might not have yet. So, it’s important that I can get up, even if I’m already in a “crashed” position. Sure, I could use tricks like “do not sit down until I’m ready to rest”, but that only works sometimes – other times, the tiredness is too much to handle.
Audio files and sound playback library in hand, negative reinforcement methods fresh in my mind, I went and cooked together a very simple solution.
Anti-Crash Script
When I noticed myself being tired and in a “crash” state, I would think “oh, no worries, I’m going to get up any minute now”. Of course, it was never just a minute, and I decided to hook into that realization, subsurface but close enough that I could justify some intervention to myself.
Would you be surprised if I told you the solution was to ring a siren into my headphones? The algorithm is simple – every time I’m “crashed” and planning to get up “real soon”, I press a button that starts a five-minute timer, programmed to ring a siren into my headphones. When the seconds stop ticking and the siren triggers, I have a choice – get up and then re-trigger the alarm for five more minutes. There is no second choice, really – I don’t give myself one. The part where I get up before turning the siren off is crucial, of course – though, in case of missing willpower, an accelerometer measuring activity could do as well.
Not that much of my willpower would be required – turned out, it typically would be enough of a shock to realize just how quickly five minutes have passed. Consistently, every time I got tired, time would pass much quicker than I could feel it, and the “oh damn it’s been five minutes already” thought made for a surprisingly powerful reality check.
Initially, the script was a tiny local webserver – I had some Flask examples fresh in my mental toolbox, so I took those and wrote two tiny HTML pages, crash and uncrash. The crash page received a seconds argument, indicating how many seconds to wait before ringing the alarm, and the uncrash page stopped the alarm. Keep the two webpages open, and hit Ctrl+R on the page I need – simple enough.
Resistance Is Counterproductive
Later on, I beautified the pages a little – adding background colours, so that it’d be easy for me to find the pages in my laptop’s window switcher and not get confused between them. That was my first attempt to make the crash/uncrash “hooks” more accessible – since, unsurprisingly, having to Alt-Tab a couple times before finding the right page required some mental energy, so I would often forget about them altogether, and developing a habit of using these pages was significantly harder. Thinking back to the very first article and principles I outlined in it – reducing resistance to use was a must.
So, the “crash” webpages got turned into keybinds accessible on my laptop globally. Surprisingly, despite the crash endpoint’s arbitrary integer delay, I didn’t need much granularity. Right now, I only use three buttons , “uncrash”, “crash in 300 seconds” (5 minutes), and “crash in 1 second” (immediate). The “immediate crash” button was a surprisingly helpful one, too. See, the “oh, five minutes truly can pass quicker than expected” lesson has stuck with me – so, when I’d notice myself crashing, I knew better than to waste time trusting in the “just a few minutes” notice.
The keybinds got me to use the script more often – which has helped me find more usecases, and use it even when I’m not sick or super tired. Really, most of the trouble nowadays is noticing when I need to press the button – which, generally, is in the mornings, when I am still groggy and a scheduled appointment might not feel as important as it actually is.
One important aspect turned out to be retriggering the alarm instead of turning it off after five minutes. I get up either way, but usually, the crash doesn’t – I might “crash” immediately afterwards, or a minute-two later. Stopping the alarm ended up being a very intentional “crash is over” decision – so, the “stop” button never got into my muscle memory. I’ve indeed had muscle-memory cycle restarts, giving myself five more minutes without realizing – but I’ve never had muscle-memory stops, which is nice, because stopping the script without even realizing it would be a critical failure condition.
Retrospective: It’s Great, Somehow
Anything missing? Definitely! For one, there are some good keybinds I could add, even if maybe they wouldn’t fundamentally impact how the script is functioning. Say I’ve woken up, and I have to get somewhere early – so I use the “crash” script to get up and get with the gravity of my current situation. As I run around the house doing morning chores, five minutes pass and the alarm rings again, even though I’m currently actively doing something around the house.
Now, running back to the laptop and pressing a keybind isn’t a problem. The problem is that I could be pressing the “reset alarm” button in two different states – either I’m doing well, or I’m not, but it’s the same button. Making two different buttons, one “doing good” and one “still crashed”, would help me collect metadata I could use for a good purpose – and, quite likely, add a trigger for some sort of positive reinforcement.
Other than that? This script has eliminated yet another common failure mode from my life – and, once again, helped improve focus. It’s as simple as simple goes, and, it’s gotten me to a more comfortable point – often, making a difference between an evening lost to tiredness, and an evening of recuperation.
One thing you might notice – to actually work properly, this script requires always-on, wireless headphones. In the next article, I’ll talk about the wireless headphone device I’ve built, why I had to build one instead of buying one, and how that device has helped me solved a bunch of other problems I didn’t realize I had.