Instead of just revealing letters when completely stuck, it would be cool to get a hint showing where or how the word is used.
Also, on completion or failure, instead of immediately loading the next challenge, it would be great to have an option to 'learn more' about the word and its usage. It would be amazing for learning new words, and you can just add a 'next' button for those who want to keep playing!
It's the new scraping tax. The delicate balance of the open web is completely broken right now because platforms are getting hammered by bots. Sometimes i wonder how long before the entire internet ends up behind a mandatory login or a paywall just to protect server bandwidth?
I really appreciate that HN still just uses temporary rate limits. You get the occasional "sorry" text, but as soon as traffic drops or your IP changes, you can read without logging in.
I wish more platforms stuck to temporary rate limits instead of building permanent walled gardens.
I think we only notice this 'sameness' because we're swimming in it all day. I recently helped a non-tech friend with a vibe-coded site for his cafe. He had the pure enthusiasm of a kid showing off a drawing. He didn't care that it looked like a thousand other Tailwind sites.To him, the magic was simply that he had nothing, and now he has a website.
I think it's slop to many of us, but to a general user, they just aren't seeing it as slop.
In 2024, a Chevy dealership deployed an AI chatbot that confidently agreed to sell a customer a 2024 Chevy Tahoe for $1. It executed a catastrophic business failure simply because it didn't know the logic was wrong.
Sure, you can patch that specific case with guardrails, but how many unpredictable edge cases are you going to cover? It only takes a user with a bit of ingenuity to circumvent them. There are already several examples of AI agents getting stuck in infinite loops, burning through massive API bills while achieving absolutely nothing.
You can contain a system failure, but you cannot contain a logic failure if the system doesn't know the logic is wrong.
A transmission error has a strictly contained, predictable blast radius. If a packet drops, the system knows exactly how to handle it: it throws a timeout, drops a connection, or asks for a retry. The worst-case scenario is known.
A reasoning error has an infinite, unpredictable blast radius. When an LLM hallucinates, it doesn't fail safely but it writes perfectly compiling code that does the wrong thing. That "wrong thing" might just render a button incorrectly, or it might silently delete your production database, or open a security backdoor.
You can build reliable abstractions over failures that are predictable and contained. You cannot abstract away unpredictable destruction.
The idea of their automated rollback infrastructure sounds good on paper, but at the end of the day, this still reads like a highly sophisticated machine for generating technical debt at lightspeed, mitigated only by an aggressive rollback system. You can't have an AI review code written by an AI and call it a security gate. A true security gate requires a human being who actually understands the context and who is actually accountable if the system breaks.
This article is spot on. I'm feeling the exact same way watching the industry aggressively promote the idea that it's safe to deploy unverified code just because an AI wrote the tests.
We are playing with fire. If we keep treating "I don't read the code I ship" as a feature rather than a liability, it's going to cause a massive, real-world disaster. The resulting regulation will be so heavy that software engineering will end up needing a Bar Council or Medical Board just to ship a basic feature. We're cheering for a trend that is going to regulate us into a corner.
Friendly reminder: There is no ghost in the machine. It is a system executing code, not a being having thoughts. Let’s admire the tool without projecting a personality onto it.
It does feel like planned obsolescence when companies like Apple limit software support for older hardware, Ubuntu run smoothly on much older devices. They could certainly do better by extending support and focusing on sustainability.
Exactly, token per dollar rates are useful, but without knowing the typical input output token distribution for each model on this specific task, the numbers alone don’t give a full picture of cost.
I bought into that idea a month or two ago, that more control and detailed instructions would deliver a clean result. That just led me down a rabbit hole of endless prompt re-runs and optimization loops. Many time I thought I had the final, perfect prompt, the next iteration slightly worsened the output. And sometimes the output was the same.
The last 20-30% of precision is brutal. The time and tokens we burn trying to perfect a prompt is simply not an optimal use of engineering hours. The problem is simple: Companies prioritize profit over the optimal solution, and the initial sales pitch was about replacement then it changed now its all about speed. I'm not making a case against AI or LLMs; I'm saying the current workflow, a path of least resistance means we are inevitably progressing toward more technical debt and cleanup at our hands.
Vibe Coding is accelerating the death of documentation and architectural clarity.
Companies are measuring success by tokens generated and time-to-prototype, ignoring the massive, hidden cost of cleanup/maintenance.
Instead of just revealing letters when completely stuck, it would be cool to get a hint showing where or how the word is used.
Also, on completion or failure, instead of immediately loading the next challenge, it would be great to have an option to 'learn more' about the word and its usage. It would be amazing for learning new words, and you can just add a 'next' button for those who want to keep playing!