Communing With The Codebase

Content Me, trying to write code while sales is hyping up their latest deal

So you call yourself a "prompt engineer" because you type fancy sentences into ChatGPT? Congrats, you've achieved the same skill level as someone who presses microwave buttons. Both require extensive training in... reading instructions and hoping for the best. The brutal honesty here is that "prompt engineering" went from sounding like cutting-edge AI wizardry to basically being a glorified Google search with extra steps. Sure, you can craft the perfect prompt with context, temperature settings, and token limits—but let's be real, you're still just asking a chatbot to do your homework while pretending it's "engineering." The microwave button physicist comparison is *chef's kiss* because both involve zero understanding of what's actually happening under the hood. You don't need to know how transformers work or understand attention mechanisms—just mash those buttons until something edible comes out. Professional seat warmer indeed.

Nothing says "I'm a serious AI startup" like using the universal symbol for "loading" as your logo. The German reply "perfectenschlag" is perfect - it's what Dwight from The Office used to describe "perfect pork anus" but also "perfect day." Just like this logo perfectly represents what most AI startups deliver - an eternal loading screen with nothing behind it. Venture capitalists, please form an orderly queue.

The hot take that's dividing the tech world: AI-generated code has officially normalized "good enough" as the new standard. The argument goes that while pre-AI devs obsessed over clean code, optimal algorithms, and elegant solutions, now everyone's just shipping whatever ChatGPT spits out and calling it a day. The brutal reality check here is that if you're still doing code reviews like it's 2019 while your competitors are deploying features at breakneck speed with AI-assisted "slop," you're basically bringing a fountain pen to a keyboard fight. The market doesn't care if your variable names are perfectly semantic or if you followed SOLID principles—it cares if the feature shipped yesterday. That comment though? "we all died in 2020 and this is hell" has 85.7K likes for a reason. The existential dread of watching software craftsmanship get steamrolled by velocity metrics hits different.

You ask the AI if it'll take your job, and it hits you with "Short answer: no. And in your case, there are two solid reasons. First, you don't have a job yet to replace." Brutal. Efficient. Accurate. Nothing says "job security" quite like being unemployed. The AI can't replace what doesn't exist. Checkmate, artificial intelligence. The civil engineering job search is apparently going so well that even the robots feel bad for you. The second reason is presumably "you're already obsolete" but the AI was merciful enough to stop at one devastating blow.

The contrast is absolutely brutal. Back in 1960, Margaret Hamilton and her team wrote the Apollo Guidance Computer code with literally zero margin for error—one bug and you're explaining to NASA why astronauts are floating aimlessly in space. That stack of code she's holding? Pure assembly language, hand-woven with the precision of a neurosurgeon. Fast forward to 2026, and we've got developers who've apparently forgotten how to code entirely. The task progression is *chef's kiss*: from "Build me this feature" (reasonable) to "I don't write code anymore" (concerning) to "Change the button color to green" (trivial CSS) to the grand finale: "Go to the Moon, make no mistakes" (absolutely unhinged). The crying Wojak really sells the existential crisis of being asked to match 1960s engineering standards when your most recent commit was changing a hex value. The irony? Those Apollo programmers had 4KB of RAM and punch cards. We have Stack Overflow, GitHub Copilot, and infinite compute, yet somehow the bar has never been lower AND higher simultaneously.