It seems strange not to mention ICE or even treatment of transgender individuals if you mention travelers. We've effectively taken out round-the-clock advertisements saying "If you're not a fire-breathing, straight WASP, tread lightly". Normally, out of a sense of egalitarianism for even conservatives, I would tag on "even though it's not true", but I fear the evidence simply doesn't bear this out anymore.
We're searching the phones of visiting punk kids. We're cruelly punishing Canadian residents for clerical errors by unnecessarily detaining them. We've told our own citizens that it's reasonable to interrogate them at any moment for looking brown or black. We told an entire class of people they simply do not exist as they or even their family, friends, and coworkers understand them. Finally, we're deporting people who, although they've been here unlawfully for years, have also contributed a great deal. I've left out the most egregious examples to remain focused on the system.
All of this is happening as a backdrop while droves of US-born and immigrant citizens alike lose their jobs and are unable to find new ones. To say the irony is palpable is an understatement.
Respectfully, 25 years ago someone might've said the same thing about you spending any time online at all. Today, people spend far more time on all number of "artificial" experiences. I'm not going to try to convince you that it's good or lasting or even personally entertaining to me, but it seems that it's entertaining to someone.
I hate to say this. I can't even believe I am saying it, but this article feels like it was written in a different universe where LLMs don't exist. I understand they don't magically solve all of these problems, and I'm not suggesting that it's as simple as "make the robot do it for you" either.
However, there are very real things LLMs can do that greatly reduce the pain here. Understanding 800 lines of bash is simply not the boogie man it used to be a few years ago. It completely fits in context. LLMs are excellent at bash. With a bit of critical thinking when it hits a wall, LLM agents are even great at GitHub actions.
The scariest thing about this article is the number of things it's right about. Yet my uncharacteristic response to that is one big shrug, because frankly I'm not afraid of it anymore. This stuff has never been hard, or maybe it has. Maybe it still is for people/companies who have super complex needs. I guess we're not them. LLMs are not solving my most complex problems, but they're killing the pain of glue left and right.
Isn't it remarkable how much less bitter even the tone of the text in the transcripts are too? This is from far before Citizens United. They talk about it with almost idle fascination.
This seems like it would work if you build a system on solid bedrock, but how often does that really happen? CarPlay, for example, started as a disaster. Unsurprisingly, it has changed a lot but remains one.
Your iOS device mute instructions are too verbose. Not all iOS devices have a mute switch. The latest devices have a mute button. Copy can be simplified to "Make sure your mobile device is not silenced"
I almost think what a lot of people are coming to grips is with is how much code is unoriginal. The ones who've adjusted the fastest were humble to begin with. I don't want to claim the title, but I can certainly claim the imposter syndrome! If anything, LLMs validated something I always suspected. The amount of truly unique, relevant to success, code in a given project is often very small. More often than not, it's not grouped together either. Most of the time it's tailored to a given functionality. For example, a perfectly accurate Haversine distance is slower than an optimized one with tradeoffs. LLMs have not yet become adept at housing the ability to identify the need for those tradeoffs in context well or consistently, so you end up with generic code that works but not great. Can the LLM adjust if you explicitly instruct it to? Sure, sometimes! Sometimes it catches it in a thought loop too. Other times you have to roll up your sleeves and do the work like you said, which often still involves traditional research or thinking.
The answer is people who don't truly understand the way it works being in charge of others who also don't in different ways. In the best case, there's an under resourced and over leveraged security team issuing overzealous edicts with the desperate hope of avoiding some disaster. When the sample size is one, it's easy to look at it and come to your conclusion.
In every case where a third party is involved, someone is either providing a service, plugging a knowledge gap, or both.
I used to be less cynical, but I could see them not honoring that, legal or not. The real answer, regardless of how you feel about that conversation, is that Claude Code, not any model, is the product.
My friend moved into one two months ago that had no fridge. I've seen a handful of similar units. Perhaps it's the price bracket you're looking in? These are often cheap apartments.
This post has aged like milk given the rollback. In the amount of time it's taken them to fix it, including lobbying xterm.js upstream and telling users "use a modern terminal emulator", you'd be hard-pressed to convince me they'd have burned more goodwill with paying customers than they already have if they'd quietly switched to alt-mode. It's a downright embarrassing bug for such a high-profile company.
Imagine if society were a program and laws, policies and procedures were its codebase. The scary thing is that the way it gets patched in reality isn't very different from the way agentic LLMs perform on actual codebases today. Bug in this department? Delete it. People abuse this system? Add more friction for legitimate users. Society has a need? Greenfield a feature that works great but doesn't fit in with existing systems at all. This didn't used to be so on the nose. I can't decide if we should welcome our new robot overlords or even what is a reflection of who anymore.
> For packages typed with JSDoc, CTRL/CMD clicking on a function will take you to actual code rather than a type declarations file. I much prefer this experience as a dev.
More broadly, this is the default behavior of JS even without JSDoc blocks, and it ought to be the default behavior everywhere, including TS. I'm not alone in this sentiment, but it's incredibly contentious. There's been an open GH issue about it with hundreds of replies for years. I have no idea why they can't just pick a different shortcut for viewing types and call it a day. They'd be doing the entire ecosystem a favor.
If only we could take output and reverse-engineer activation layers through some parameters and get the original prompt. Imagine how much time we could save if we could read the chat transcript or the two actually human-written paragraphs this article was based on. They'd be some banal rant about a DevRel dude but at least it'd be more efficient.
GitHub wants to sell a service. Keys are convenient. Better alternatives in authorization and authentication exist, and GitHub is very aware of them. They even offer and facilitate them. For example, see OIDC. But many users either want keys because they're used to them or GitHub is sure they do, so they continue to offer them to avoid friction. The alternatives require more parameters, thought, and coordination between services.
GitHub has deprecated classic tokens, but the new tokens are not backwards compatible. The deprecated tokens have also continued to be available for some time. Real security professionals will tell you flatly "tokens are bad", and they're right. They're leakable attack vectors. The tokens are the problem and discontinuation is the solution. Scanning is simply symptom treating, and given what I know about Microsoft culture, I doubt that's going to change soon or quickly.
We may have some things in common. I'm not a mom, but I am a woman. And I don't want to assume the same is true for you, but breaking into this industry was difficult for me, so even without children, I'm really invested in the ability for juniors to succeed too. I wish I had responded more directly to your article rather than my general ennui. I really admire your willingness to write this. I hope it gets broad engagement, because I think these problems seem obvious to us but based on private conversations I've had with some industry peers in very senior director roles the drying of junior opportunities for growth is not readily obvious to them. I'm going to have to think more about the corporate incentives you mentioned, because reading that in the article, it feels deeper to me, and I think that's what I was trying to get at by sharing my past company details.
I think you succeeded overall at your goal! Thanks for replying. You encouraged me to go back and read your article more closely.
Yeah, issues have been open on GitHub for months. I've tried shortening my scrollback history and using other emulators but it doesn't seem to make a difference. It's pretty frustrating for a paid tool.
I guess I should clarify too: I don't believe in junior titles. They handicap people into the position you describe where they must move on to progress. When I describe "junior" above, I generally mean a candidate with <=1.5 years of experience. When I say mid I mean any amount of experience greater but not senior according to technical review. And yep, I know this is not the best heuristic because there are definitely people with no working experience who have mid-senior coding skills (although they're rare). I think that's sort of part of the problem too. Senior management is disincentived from understanding the roles and growth trajectories, so our heuristics for hiring are totally warped and stomped on.
This article talks a lot about AI, but what I find odd is that in my relatively short (but long enough) ~9 yr career so far, this problem predates AI. I don't deny that it exacerbates it, but you don't kill a disease by addressing the symptoms. From the first time I was ever involved in the hiring process, senior leadership always encouraged me to hire more experienced staff, always most heavily scrutinized juniors, and had negotiations fall through with mid-level candidates the most. This was despite juniors passing technical screens with strong showings. This was not at a Fortune 500. This was a micro-cap subsidiary of a private, billion dollar company.
And although it hasn't discouraged me, I have to admit that I've been burned by juniors when caught in the middle between them and senior leadership on output expectations or strategy because frankly it's much more challenging to mentor how to navigate company politics than it is to mentor professional coding acumen. I want to be humble here. I don't think that's the junior's fault.
It feels like these problems go a lot deeper than AI. Most shops want software teams that are either silently embedded black boxes that you insert rough instructions into and get working software as output or an outsourced team. We've all experienced this. It seems silly to deny that it's directly related to why it's so hard to mentor or hire juniors.
I don't know for sure, but it's definitely the first tool of that value to have a persistent strobing (scroll position) bug so bad that passersby ask me if I'm okay when they see it.
LLMs can be used to quickly mulch data into a digestible format that at least used to take effort. Friction is a natural deterrent for bad behavior. Beyond that, however, is the fact that your user interactions with most applications used to be quite coarse. A "customer story" was just that: a story we crafted from the data we have available to us about our customers. We have to build it from heuristics like bounce rate, scroll distance, and other thorough, idiosyncratic and diligent abandonment metrics.
Now why bother? Your customer will ask their silver ball (LLMs) anything and everything, and you can directly do bulk analysis on (in theory) the entire interaction, including all of your customer's emotions available via text.
Lastly, your customers are now eager about this tool, so they're excited to integrate/connect everything to it. In a rush to satisfy customers, many companies have lazily built LLM integrations that could even undermine their business model. This pushes yet more data into the LLM. This isn't just telemetry like file names, this is full read access to all of your files. How is that not connected to privacy?
Come down to Ojai valley. They'll be crawling all over you in no time. I grew up in the east. Most of the ticks we had at least couldn't even carry Lyme or Rocky Mountain Spotted Fever. California is wild.
For GP's sake, even before you make it to FYPM levels of angry, you will be in over your head. It's too much work. I remember being very early in my career and feeling like GP does. This is very easily more than a full-time job. The demands people will make of you and the attitudes they will use to do it will make you crazy.
What strikes me about this exchange is no one is talking about the money. In the past, you could do either and no one had to care except you. Now a lot of jobs that people could find fulfilling aren't because the economy is so distorted, so how are we supposed to honestly look at this? I guess let's walk these people off the plank and get this over with...