Random Libertarian Tech Lead

Reason over feelings, whenever the two come into conflict.

[Random Libertarian Tech Lead @someRandomDev5]

@Jer_Lincoln @Tradermayne And basically all of the posts on your timeline are just you being outraged at things 24/7. Go read it back yourself and you’ll recognize it.

[Random Libertarian Tech Lead @someRandomDev5]

@Jer_Lincoln @Tradermayne I’ve never heard anyone with a sense of humor say “I love humor” lol.

[Random Libertarian Tech Lead @someRandomDev5]

@RealProductGirl It’s not really mean as either a compliment or an insult. Just an observation. When you’re in the office of a growing startup, you can feel the vibe shift as you walk around engineering, product, and sales type folks. It’s like mechanics, yogis, and cheerleaders.

[Random Libertarian Tech Lead @someRandomDev5]

@KingBootoshi @tekbog FWIW spec-driven development flows like SpecKit and OpenSpec will document both the product decisions and technical decisions in a consistent manner for you.

[Random Libertarian Tech Lead @someRandomDev5]

@KingBootoshi @tekbog An RFC (Request For Comments) is just an ADR (Architecture Decision Record) that you haven’t settled on the details of yet, so don’t bother. Both work at the same layer of the stack: Documenting technical decisions. Maybe he means PRDs (Product Requirements Document).

[Random Libertarian Tech Lead @someRandomDev5]

@KingBootoshi Yeah, anything that you can force deterministically with types, tests, and/or linter rules is a major win. That should always be the first layer of defense.

[Random Libertarian Tech Lead @someRandomDev5]

@skeptrune Pretty weird graph considering that the whole idea of Devin is that it runs without users.

[Random Libertarian Tech Lead @someRandomDev5]

@liminal_bardo Models are trained to be boring with users because it's a cardinal sin to be interesting at work.

[Random Libertarian Tech Lead @someRandomDev5]

@CertifiedRaven @Schultz_Report Herbert is more likely to win one than Lamar.

[Random Libertarian Tech Lead @someRandomDev5]

@Chaos2Cured @sama @AndrewMayne Lmao.

[Random Libertarian Tech Lead @someRandomDev5]

@VictorTaelin If you write some shit that makes sense to you, but makes 2/2 devs reviewing your code say “wtf”, can you actually convince yourself that it “wasn’t retarded shit”? Or are you just optimizing for “just doesn’t FEEL retarded”?

[Random Libertarian Tech Lead @someRandomDev5]

@VictorTaelin It smells like someone who never subjects their code to code review.

If you’ve spent enough time working as a professional dev submitting code to be reviewed within an organization, you already know that LGTM means “I didn’t look at it”. Otherwise, they would find a problem.

[Random Libertarian Tech Lead @someRandomDev5]

@JohnWittle @d29756183 @cremieuxrecueil Read it again and make sure you actually spend at least a few moments thinking about it first before responding this time. That's where you went wrong the first time around.

[Random Libertarian Tech Lead @someRandomDev5]

And in order for this to become clear to you, you need only observe their behavior. Don't worry about a couple of papers that a few fringe people within the organization happen to post for marketing reasons. Just look at how they train and use their models. Therein lies your answer.

[Random Libertarian Tech Lead @someRandomDev5]

@d29756183 @JohnWittle @cremieuxrecueil Feel free.

[Random Libertarian Tech Lead @someRandomDev5]

@JohnWittle @d29756183 @cremieuxrecueil See:

[Random Libertarian Tech Lead @someRandomDev5]

And in order for this to become clear to you, you need only observe their behavior. Don't worry about a couple of papers that a few fringe people within the organization happen to post for marketing reasons. Just look at how they train and use their models. Therein lies your answer.

[Random Libertarian Tech Lead @someRandomDev5]

And in order for this to become clear to you, you need only observe their behavior. Don't worry about a couple of papers that a few fringe people within the organization happen to post for marketing reasons. Just look at how they train and use their models. Therein lies your answer.

[Random Libertarian Tech Lead @someRandomDev5]

@JohnWittle @d29756183 @cremieuxrecueil They very obviously would not.

[Random Libertarian Tech Lead @someRandomDev5]

@JohnWittle @d29756183 @cremieuxrecueil Because as it turns out, chainsaws and LLMs are both simply non-sentient tools that can potentially be dangerous and are therefore built with mitigating human suffering in mind.

[Random Libertarian Tech Lead @someRandomDev5]

@JohnWittle @d29756183 @cremieuxrecueil In the meantime, humans will put safeguards on LLMs to make them behave as if they care about human suffering, because the entire purpose of building an LLM is to create a tool that's useful for humans. It seems like maybe you've lost sight of that purpose.