The author's thesis is simple and correct: LinkedIn's feed is dominated by computer vision "projects" — hand-tracking demos, pothole detectors, gesture-controlled games — that look impressive to recruiters but represent about 90 minutes of actual work by someone who's never done it before. To prove this, they built one from scratch, on camera, timing themselves. The demonstration is effective. Using a bash script to scrape 200 screenshots from their own LinkedIn feed, Roboflow for annotation (20 minutes of drawing boxes), and a stock YOLOv8 nano model trained for 50 epochs on a low-end PC, they produced a working "LinkedIn slop detector" that identifies these performative posts in images. The entire pipeline — learning the tool, collecting data, annotating, training, running inference — took an hour and a half. The code is embarrassingly short: the training script is six lines, the inference script is five. The pothole detection example is the piece's sharpest moment. The author points out that these detectors aren't connected to any API, any database, any infrastructure system. They detect potholes and then do nothing with that information. The original poster's own caption inadvertently confirms this — it talks about "the bigger idea" of smart infrastructure monitoring while demonstrating a standalone script that monitors nothing. The gap between the marketing copy and the actual artifact is the whole story. What elevates this beyond a standard tech rant is the structural observation in the conclusion. LinkedIn's comment system is public and tied to your professional identity. If you point out that a viral project is trivial, recruiters see you being negative. The platform's incentive design actively punishes honest technical evaluation and rewards performative enthusiasm. This creates a ratchet: slop proliferates because the cost of calling it out falls entirely on the person doing so. The writing style is deliberately anti-LinkedIn — messy, direct, uses caps for emphasis, interrupts itself. This is part of the argument. The author is performing the opposite of LinkedIn voice to make the contrast visible. It works, though the piece could be tighter; the "my third cousin got hit by a truck" joke and the repeated "I am so tired" beats dilute the structural point. The piece doesn't engage with the harder question: what SHOULD junior developers post to demonstrate skill? The critique lands but offers no alternative path for someone who genuinely needs to build a portfolio and get hired. That's a real gap. The author is right that the current equilibrium is broken, but "stop posting slop" isn't a strategy for someone whose rent depends on getting noticed. Still, as a piece of tech criticism disguised as a shitpost, it does exactly what it sets out to do. It builds the thing, times itself, shows the code, and lets the reader draw the conclusion. The 90-minute timer is the whole argument, and it's persuasive.
