My (Mis)adventures with "Vibe-Coding" by means of Grok and my 1988 Cathode Ray Tube Television
______________________________________________________________________________________________
I’ve taken time off from my usual activities and decided to delve head-first into this bizarre world of “vibe-coding.” And I did it all with the use of X’s Grok. I now have a pipeline straight to it, through my Linux terminal. Grok can conjure up anything and everything I ask of it and send it straight into my home folder.
And I have to say — I am very underwhelmed.
Probably at myself, more so than anything.
Okay. Let’s explain ourselves here.
The program in question started out as a simple shell script that existed on my work laptop computer with three screens attached. The third screen is a CRT monitor that was built in the year 1988. (Pictured above.) I use this to play old games on — including games that LOOK old! Now, that’s a trip!
This simple shell script calls up a predefined playlist that instantly plays video through a player called “MPV” whenever I change “channels.” And I use a Linux program called “feh” to call up a title card, displaying the name of the channel that displays for five seconds before the first video is called up.
______________________________________________________________________________________________
I throw every single thing into this channel! From Pee-Wee’s Playhouse and Beakman’s World to It’s Always Sunny in Philadelphia and Aqua Teen Hunger Force. It’s a very confusing channel!
______________________________________________________________________________________________
Simple, right? Except it’s a simple script — It’s too simple! You can only “change channels” in one direction — and really, all I’m doing is ending a task on the shell script, whenever I exit out of MPV, which then loads up the next task. You can’t work your way back up the chain!
Well, I thought I couldn’t. But Grok proved that wrong.
And Grok was able to figure this part out without fuss! All I wanted was to be able to change the channels in any direction I wanted. Up or down. I’m not a coder by trade! That’s why I decided to take this weird little adventure and find out if I can capture those faded analog dreams of mine.
But then, I wanted to do some other things — some far more technical things, involving these weird, bizarre devices referred to as “systemd timers,” which I could use to assign programming blocks into every channel. So, let’s say I wanted Sunny In Philadelphia and the British comedy Peep Show to show up within the same hour-long block at three-o’clock in the morning — assuming if happened to be up at three-o’clock in the morning!
But hey. Peep Show and Sunny are a match made in heaven. I’d gladly fall asleep every night to some of the most cringy programming ever created in the mid-2000’s! And then I’d follow it up with some Mighty Boosh.
Sounds awesome. And on paper, I can do all this! Grok was able to set up an entire GUI for me to plan out every hour, from 00:00 to 23:00. (Listing it in military time, for some reason.)
______________________________________________________________________________________________
This what the Scheduler looks like on the GUI.
______________________________________________________________________________________________
Putting this in action is easier said than done. I thought I could just set up these timers and everything would be hunky-dory!
Nope. Nothing works. The timer triggers but the only thing that happens is the channel resets itself and keeps playing from the playlist like it can’t see anything.
The answer I went with is to auto generate programming block playlists that the timers can “read from” instead of the timers reading directly from the channel playlist that I already had set up.
This is a very clunky workaround. And I say this is clunky because it takes five minutes for the program to “refresh” all those playlists for every programming block! Oh! And let’s forget about custom folders. Apparently, having a custom folder assigned to a programming block just complicates things! Either that I could, I don’t know… force the program to scan the damn folder!
Oh. And by the way — it’s not the timers that “read” the playlists! It’s MPV! The timers are supposed to “trigger” the playlist for MPV to read!
I hope this isn’t confusing. Because I have to keep reminding myself of this for some stupid reason!
And let’s not start with what to do when the trigger hits right in the middle of a show that’s currently playing! I haven’t even thought that far ahead!
This turned out to be a can of worms that I just opened up!
This only enforces what I suspected this entire time: The AI is only as smart as the user — and I’m not going to say that I am smart.
No. Not really. Really, what I do is I “stumble” my way through things by means of trial and error. (Heck. I wrote a previous article all about that!) In retrospect, it would have been far more practical if I had a complete design concept of the entire program before I ever prompted Grok! The thing is I just assumed that Grok would put two and two together, that these lists would be triggered by the timers and then fed to MPV!
But I had to spell it out? The whole thing? It didn’t know that’s what I wanted this whole entire time?!
Okay! Fine! I get it! I’m not a programmer! Likewise, I’m far more familiar with the hardware side of things. But it just goes to show how brainless artificial intelligence really is. Or how brainless I am. It’s hard to tell which one is which at this point.
I’ll keep working on it some more, but keep in mind that I’m not a full-time programmer by any stretch of the means! I just like fiddling about with things, not unlike how I was, when I was five years old, when I tore apart my father’s expensive 80’s-era boom box all because I was curious about what it looked like on the inside.
Perhaps programming isn’t all that different. But with this, it’s more like how I write my books. When I’m out and about, I’ll have some grand idea about what to do, how to implement an idea or how to solve for X for Y. At the end of the day, I’ll come back and put those ideas to the test to see how things turn out. It’s never one, all-encompassing unified plan! It’s the little things that build up, piece by piece, brick by brick. And who knows? Maybe next time it’ll actually play the thing that I want it to play when I want it to actually play!
If you’d like for me to continue developing this project, you can certainly let me know about it through the substacks and the X’s! I might not ever put this up on GitHub, but if you want to fiddle around with it, well… I’ll think about it. If you DM me on X, I might consider it.
But more and more, I’m thinking that the real issue with artificial intelligence isn’t AI itself. It’s the limits of the human mind that’s prompting the AI to solve X for Y.
And to this I humbly apologize to not just Grok, but to all of Artificial Intelligence, for I see it as clear as day. Humanity will henceforth continue dropping its ice cream on the pavement.
I wonder how the hell we even got here to begin with, sometimes.
