ezrast

Gotta hand it to 'em though - posting this less than a month after the Matrix boondoggle certainly is, uh, audacious.

Another article that, by the third sentence, namedrops seven different AWS services they want to build their app on and then spends the rest of the argument pretending like that ecosystem has zero in-built complexity. My friend, each one of those services has its own security model, limitations, footguns, and interoperability issues that you have to learn about independently. And you don't even mention any of the operational services like CloudWatch, CloudTrail, VPCs (even serverless, you'll need them if you want your lambdas to hit certain other services efficiently), and so on. Those are not remotely free. Your "real developers" can't figure out how to write a YAML document, but you trust them to manage infrastructure-as-code for motherloving API Gateway? Absolutely wild.

Kubernetes and AWS are both complex, but one of them frontloads all the complexity because it's free software written by infrastructure dorks, and one of them backloads all of it because it's a business whose model involves minimizing barriers to entry so that they can spring all the real costs on you once you're locked in. That doesn't mean either one is a better or worse technical solution to whatever specific problem you have, but it does make it really easy to make the wrong choice if you don't know what you're getting into.

As for the last point, I don't discourage serverless solutions because they make less work for me, I do it because they make more. The moment the developers decide they want any kind of consistency across deployments, I'm stuck writing or rewriting a bunch of Terraform and CI/CD pipelines for people who didn't think very hard about what they were doing the first time. They got a PoC working in half an hour clicking around the AWS console, fell in love, and then handed it to someone else to figure out esoterica like "TLS termination" and "logs" and "not making all your S3 buckets public by accident."

EDIT: I originally had an escalatory reply here but have thought better of it. Trying again but without being an asshole: yeah, what you describe is one way miscommunications can happen. They can also happen in other ways that may be the engineer's fault, or nobody's at all. They may not happen every day, just like how you won't be deploying a new service every day, but anyone who works in an office is going to need to be able to deal with them regardless. Highly functional organizations are so because they can recover from mistakes, not because they don't make them.

Without wanting to be unkind, this author misunderstands the purpose of the exercise, which is to demonstrate to another human being that you can write code that meets the company's standards (as in, would pass PR review), not to ship a fully-functional product solo. The initial email makes it explicit that they don't care if the product even works ("Can use a fake backend") and that the specifications are deliberately open-ended. A little intuition and empathy can tell you that they probably are not going to spend many hours reviewing a complex submission, so the project should optimize for demonstrability of code quality, not completeness.

"I would like to know what kind of response I could expect..." - This is also established from the beginning: you can expect either a "Looks good, let's do some interviews" or a "Sorry, not interested," based on the code you submit. They can't narrow down the choices prior to your submission, because they're grading your submission, not your proposal document with an extensive list of details that they already told you they're mostly ambivalent to.

"So it is funny that my project is so weak, yet it made them update the guidelines to something stricter." - Main character syndrome. As someone who has been on the other side of these kinds of reviews, the far more likely explanation is that they kept getting submissions where the build instructions didn't work (which is not disqualifying by itself; the authors may not be on the same OS as the reviewers) and they got tired of spending time dealing with it.

Ultimately, the failure here was not a technical one but a social one. The author tried very hard to do the thing that it seemed like they were asking for, not the thing they actually wanted. The hiring manager's unwillingness to engage with the proposal doc was itself a form of communication that they were not interested in this level of detail, it was just an implicit one.

It's common for engineer types to want all that kind of communication to be explicit, and I have a lot of sympathy for those folks, but the reality is that teamwork is a skill, and the ability to suss out what a stakeholder actually wants, but isn't saying due to incomplete information and/or office politics, is a reasonable thing to select for. The ambiguity of the prompt is a feature, not a bug: it's kept the author away from a company whose communication style they're not compatible with.

(that said, this project does sound excessively complex for a take-home)

You can't programatically detect novel BS any more than you can programatically detect viruses or spam. You can only add the fingerprints of known badness into an ever-growing database. Viruses and spam are antagonistic to well-resourced institutions, and their databases get maintained reasonably well. LLM slop is being generated by those same well-resourced institutions. I don't think it fits into the same category as Nepenthes.

The whole point of systemic incentives is that there is no conspiracy. Nobody wants a DDOS and every large provider will have people genuinely working to avoid them. But every time there is an opportunity to allocate resources, the team that gets to frame their return on investment in terms of real dollars will always have an edge over one whose value is realized only in murky customer satisfaction projections. Over the lifetime of a company, the impact of these decisions will add up with no need for any of the individuals involved to even be aware of the dynamic, much less conspire to perpetuate it.

Whelp, looks like I'm today's Poe's Law poster child. ;)

This seems like a distinction without a difference, since we can never positively categorize any unobserved phenomenon as impossible (vs merely unobservable). To me, it seems ontologically cleaner to treat existence and observability as the same thing. shrug

What does "impossible" mean to you if not that a thing and it's consequences can never be observed?

It's not really a problem in practice (and I love Ruby), but it's still wild to me that they made the parser do this:

    $ irb
    irb(main):001:0> def foo(x=70) = x
    => :foo
    irb(main):002:0> i = 2
    => 2
    irb(main):003:0> foo / 5/i
    => 7
    irb(main):004:0> foo /5/i
    => /5/i

I'd call that reading poorly-supported enough to be incorrect. The report establishes that waypoint names are changed somewhat frequently for reasons that are largely left to mystery, but can include "a basketball player I like died." Additionally, "This is on fire right now" suggests some relevant context beyond "a single reporter sent us a nosy email." So while it may technically be true that the email preceded the name change in the chain of causality, its framing as the principal cause appears to be a narrative built of incomplete information. The linked NYT article also mentions pushback from pilots.

Think of all the algebra problems you got in school where the solution started with "get all the x's on the same side of the equation." You then applied a bunch of rules like "you can do anything to one side of the equals sign if you also do it to the other side" to reiterate the same abstract concept over and over, gradually altering the symbology until you wound up at something that looked like the quadratic formula or whatever. Then you were done, because you had transformed the representation (not the value) of x into something you knew how to work with.