burnerToBetOut

Good for you. Better get crackin' [1]

[1] https://news.ycombinator.com/item?id=47099583

    > …LLM agents posing as
    > people on HN…

I suppose I could have made it clearer that the reply would explicitly make readers aware that the analysis was done by a Robo — i.e., an LLM.

The answer, my friendow, is blowing in the context window [1]…

_______

…LLMs already "know" these books deeply, but without a structured prompt scaffold they apply that knowledge inconsistently and at low confidence. Giving the model a explicit lens — "review this as if you're checking against Clean Code heuristics C1–C36" — concentrates attention and dramatically reduces hallucinated or off-topic feedback…

…

Where I'd push back or warn you:

Context collapse is your #1 enemy. Clean Code was written for Java in 2008. DDIA is about distributed systems at scale. If you apply the Clean Code reviewer to a 50-line Python script, you'll get pedantic nonsense about function length when the actual problem might be that the data model is wrong. Your skill selection logic needs to be domain-aware, not just "throw all skills at every file"…

…

_______

[1] https://g2ww.short.gy/ZLStasQ1

@speckx, did you know that FT articles are behind a paywall?

Found somebody's save of this one, though: https://archive.fo/NlQER

The company that fired that blogger is Google. Right?

    > …perspectives from people
    > who are better at reading
    > if studies are quality or not…

But, without even reading the whole PDF [1] of this particular paper, it's super easy to tell that the authors are all from pro-vibe coding consultancies [2][3][4]

[1] https://g2ww.short.gy/ConsAndPros

[2] https://g2ww.short.gy/MarkOfTheBorg

[3] https://g2ww.short.gy/ActualInequal

[4] https://g2ww.short.gy/ConDelivery

    > …the very dream of "getting
    > rich quick through building
    > a software product being a
    > complete outsider" has waned…

I hope you're not expending that same energy for _every_ dupe on this site.

God forbid you should overexert yourself and burn out ;)

Absolutely!

I did think about putting "and Upwork" in the title.

Thanks for replying.

I asked the question because I wondered if or how AI-assistance might be used more these days. By either the freelancers in their delivering their services, or by buyers who might be figuring they don't need to hire a freelancer for whatever reason these days.

Uncle Sam — by proxy — Wants You…

____

The Non-Negotiables: How We Work

…

• …We leave video links open while we eat lunch. Our pets and kids know our coworkers…

____

If the graphics on their landing page is anything to go by, that company [1] is essentially some kind of defense contractor.

The CTO running that Onebrief shop is the guy [2] that prompted the prediction I made in another post [3].

Coincidentally, I came across a blog post recently [4] where the blogger describes the above job description's way of working playing out IRL.

[1] https://g2ww.short.gy/BigtechBro

[2] https://g2ww.short.gy/OM16G

[3] https://news.ycombinator.com/item?id=46995623

[4] https://g2ww.short.gy/DonutPassGo

No source code? No privacy policy? No "About"?

No thanks.

Ahhh! I think I get where you're coming from.

You're thinking I posted this "article" to get upvotes.

Now, you got two more guesses left for "Why'd he post this?"

Hint: I've written the answer twice already without me even having been explicitly asked to ;)

    > I think this could work
    > well as a blog post…

But I only set out to get the overall vibe of what Yegge is doing. The static analysis [1] and the automaton's take [2] were plenty for doing that.

    > …maybe you could at least
    > go through the report yourself
    > and point out issues it highlights

It was all I could do to muster up the will to even do as much as I did eventually force myself to do.

I'm not the faniest fan boy of AI-assisted coding. But I appreciate them for what they're good at. To me, it delivered a stellar code review here.

    > …show the relevant code,
    > and explain why that's
    > actually bad
    

The Sonar Cloud analysis is super detailed. And the automaton's code review is on the money. Digging into their respective nitty gritty is left as an exercise for the reader.

[1] https://g2ww.short.gy/CCWutUHalfDone

[2] https://g2ww.short.gy/ItsaGasGasGas

    > …The final level in his AI
    > Coding chart reads: "Level 8:
    > you build your own orchestrator
    > to coordinate more agents"…

____

…How do you avoid getting tired? Dude, I take naps throughout the day. I'm exhausted…

…

…which is why I mentioned in one of my last blog posts that I'm taking naps all the time…

____

Yegge's productivity sounds impressive. I'll give you that. But it doesn't sound practical or sustainable for the everyday dev.

I doubt that even Google — with all its famous perks — offers employees ad hoc nap times while they're on the clock.

[1] https://g2ww.short.gy/Napster2026

Since replying to the first posting of this [1], I've been wondering about the familiarity of it. It reminds me of another manifesto [2]

I foresee there being a ton of Outcome Engineering consultancies springing up.

We can expect Outcome Engineering Coaches making a killing at telling companies, "You're doing o16g wrong! For a 'small' fee, I'll teach your teams how to do it right."

If this manifesto is anything like the other one we all know and love, it'll effectively give the original author(s) a license to print money…

____

…Written by Cory Ondrejka — CTO of Onebrief, co-creator of Second Life, engineering leader at Google, Alice Albrecht, Andy Luly, Charity Majors, Eric Schmitz, James Cham, Jason Holtman, Matt Firestone, Palmer Truelson, Seamus Blackley, Tori Horton, and Tyler Mah were early readers or reported bugs…

____

[1] https://news.ycombinator.com/item?id=46978648

[2] https://g2ww.short.gy/BigAIgile

A sampling of Redditors' comments from that thread that offer some thought-provoking observations…

____

…The issue is that with these near shore markets, it’s not always about cutting costs. A lot of people in these threads seem to assume that American development talent is always the best, which is not necessarily the case. The only thing you can be certain of is that your American devs will be your most expensive. Some of the most competent devs I’ve hired (including devs from the US) have been from Costa Rica. Americans do not have a monopoly on doing good work…

____

   > …How exactly are they
   > measuring their best
   > developers?…

____

“As a concrete example, an engineer at Spotify on their morning commute…all before they even arrive at the office.”

____

    > …a slightly accelerated
    > modeling phase and then
    > let them loose on
    > the implementation…

_____

MODEL

…

The Verdict: If you provide a clear instruction like "Before you touch the code, read architecture.puml and ensure your changes do not violate the defined inheritance/dependency structure," the agent will be very effective at following it.

If you just "hope" it bears it in mind, it probably won't.

_The agent is a tool, not a mind-reader; it will take the shortest path to a passing test unless you wall that path off with your architectural models_.

…

To make it actually work, you need to turn the UML from a "suggestion" into a "blocker." You should add a section to your AGENTS.md (or CLAUDE.md ) that looks like this:

    1. Tool Trigger: By using words like "…"

Why this works:

…

_____

[1] https://news.ycombinator.com/item?id=46974325

[2] https://g2ww.short.gy/TheMightyBooch

Killer idea!

It's not obvious in your README whether or not this emits the pom.xml for the project it generates. Like containing the dependencies for Spring Boot, db driver, etc.

Also, please do one for Gradle next.

This "social media report" [1] is dated February 10th.

It uses the photo from your Business Insider article (also dated the 10th).

[1] https://xcancel.com/BanafaAhmed/status/2021309932390261134/

Those are all swell ideas. Thanks for posting the page!

Now what's meant to happen next?

In practice I mean.

    > LLMs love using them
    > for whatever reason

   > Why do people use the
   > ' — ' all the time now?
   > It's not a proper English
   > separator.

    > …I suspect going a more visual block
    > diagram route, to generate some
    > intermediate .md or whatever, might have
    > promise, especially for defining clear
    > bounds/separation of concerns…

So can my automaton bud [2]…

_____

MODEL

…

The Verdict: If you provide a clear instruction like "Before you touch the code, read architecture.puml and ensure your changes do not violate the defined inheritance/dependency structure," the agent will be very effective at following it.

If you just "hope" it bears it in mind, it probably won't.

_The agent is a tool, not a mind-reader; it will take the shortest path to a passing test unless you wall that path off with your architectural models_.

…

To make it actually work, you need to turn the UML from a "suggestion" into a "blocker." You should add a section to your AGENTS.md (or CLAUDE.md ) that looks like this:

    1. Tool Trigger: By using words like "…"

…

_____

[1] https://news.ycombinator.com/item?id=46935979

[2] https://g2ww.short.gy/TheMightyBooch

There's been a couple other threads here recently about the apparent "existential crisis" some people are having these days.

A lot of commenters tell those people that the reason they're struggling is because they made the mistake of having their identity be defined by their job.

Your post made me wonder if stories like yours is how that whole I-am-my-job-and-my-job-is-me entanglement starts.

    > …Not burnout…

    > …this is a very different kind
    > of mistake to make compared
    > to overwriting a file…

    > …but I digress…

    > …which supports my point…

   > …I did a bit of digging…

   > …*he* seems like a very good one…

    > …I have experience AI’s deleting
    > things they shouldn’t but not since
    > like, the gpt4 days.…

———

…I pointed Opus 4.6 at a 60K line Go microservice I had vibe coded over the past few months, gave it some refactoring principles, and let it run unsupervised…

…

What went wrong #

At some point in the code, we re-fetch some database records immediately before doing a write to avoid updating from stale data. It decided those calls were unnecessary and _removed them_…

———

[1] https://g2ww.short.gy/ClaudesLaw