HNHacker News
TopNewBestAskShowJobs

kator

2,679 karma · joined February 26, 2012

https://www.karlbunch.com/

Some of my hobbies include helping people unlock creativity using LLMs: https://conjori.com/

And trying to revive a dead MMO: https://www.swgemu.com/

I work for Amazon, and all opinions I share on this handle are my own.

submissionscomments
kator··on Issue: Claude Code is unusable for complex engineering tasks with Feb updates
Fascinating, I thought I was losing my mind. Claude CLI has been telling me I should go to bed, or that it's late, let's call it here, etc, and then I look at the stop-phrase-guard.sh [1] and I'm seeing quite a few of these. I thought it was because I accidentally allowed Claude to know my deadline, and it started spitting out all sorts of things like "we only have N days left, let's put this aside for now," etc.

Just this morning I typed:

    STOP WORRYING ABOUT THE DEADLINE THAT IS MY JOB
[1] https://gist.github.com/benvanik/ee00bd1b6c9154d6545c63e06a3...
kator··on I'm Not Consulting an LLM
The "calculator ruined the world" argument was actually studied to death once the panic subsided. Large meta-analyses of 50 years of data show it was mostly a non-problem. Students using calculators generally developed better attitudes toward math and attempted more complex problems because the mechanical drudgery was gone.

The only real "catch" researchers found was timing. If you introduce them before a kid has "automaticity" (around 4th grade), they never develop a baseline number sense, which makes high-level math harder later on.

It's a pretty clean parallel for LLMs. The tool isn't the problem, but using it to bypass the "becoming" phase of a skill usually backfires. If you use an LLM before you know how to structure an argument or a block of code, you're just building on sand.

kator··on Updating Codex Contribution Guidelines
Peak irony: OpenAI's Codex project is being DDoS'd by its own creation.

They made code generation essentially free, and now they're drowning in AI-generated PRs that take more time to review than it would take to just write the fix themselves. So they've moved to invitation-only contributions. They built the tools that removed code as a barrier, and now they're living with what that actually means: quantity goes up, quality goes down, and every PR still needs a human with architectural context to evaluate it.

The bottleneck has completely flipped. It used to be finding someone to write the code. Now anyone can dump a 200-line PR in seconds, but the review time from someone who actually understands the codebase is the most expensive thing in the repo. That context can't be generated, it has to be earned.

They built a firehose and their own team can't drink from it.

kator··on Claws are now a new layer on top of LLM agents
Some users are moving to local models, I think, because they want to avoid the agent's cost, or they think it'll be more secure (not). The mac mini has unified memory and can dynamically allocate memory to the GPU by stealing from the general RAM pool so you can run large local LLMs without buying a massive (and expensive) GPU.
kator··on Stop Doom Scrolling, Start Doom Coding: Build via the terminal from your phone
I use Terminus with Zellij and keep about 8 sessions going with a combination of Claude and Codex, and once in a while, Gemini. It's great when you're sitting in a docotor's office lobby bored out of your skull and when you get back to your desk you just join the session and it's all right there.
kator··on I don’t need a Steam Machine
Read all the way to the end: "I’m getting one."

So will I, if anything to support the effort, and check it out maybe I'll buy more for my kids or something.

kator··on Frank Gasking on preserving «lost» games
Bigger challenge is games that die because the back-end servers are turned off and the assets are discarded. I'm on a team reverse engineering an old MMO from 2011. We've spent years rebuilding the server from packet captures and disassembly because everything official got nuked. This is just one of many examples where the customer "buys" a thing in their mind only later to find out they really didn't buy anything.

The legal situation is a mess too. We're not competing with anyone (game's been dead over a decade), we're not selling anything, but we still operate in this gray area wondering what's fair use versus what crosses a line. Copyright law wasn't written with "what if the company abandons it and erases it from existence" in mind.

Meanwhile every day that passes, more of these games just vanish permanently because preservation is treated as piracy.

kator··on Ripgrep 15.0
ripgrep has saved me so much time, I also use it now with LLMs and remind them they have ripgrep available! I added a donation on github, thanks for all your work.
kator··on IQ tests results for AI
LLM vibe coded site and architecture?
kator··on Writing a good design document
I also added an Appendix "Technical Stack Considerations," but I like the PR and the FAQ's to focus on the customer/end-user's needs. The technical details matter, but they serve the customer outcome, not the other way around.

A recent project's tech appendix had headers like "Core Technology Philosophy", "Backend Architecture", "Frontend Architecture", "Service Architecture", "Infrastructure and Deployment", "Security Architecture", "Performance Requirements", "Configuration Management", "Backup & Disaster Recovery", "Development Workflow", "Network Architecture", "Resource Management", "Development Principles", and "Scalability Considerations".

The beauty is that by the time you get to the technical appendix, you've already validated what you're building and why it matters. The technical choices then flow naturally from the customer requirements rather than driving them.

kator··on Writing a good design document
7.5 Years at Amazon, and even for my side projects, I write PRFAQs and share them with my stakeholders to gather feedback. I'm a PMT at Amazon, but in my alternative life, I code on many projects, and develop infrastructure, architecture, etc, and enjoy writing as much of it as I can.

That said, work back from your customer!

kator··on Dancing brainwaves: How sound reshapes your brain networks in real time
A little better link here with a link to the detailed article, too.

https://health.au.dk/en/display/artikel/dansende-hjerneboelg...

Meanwhile, it's interesting that I do find I can focus deeper on code with certain types of music. I also have certain music I listen to when I want to write a document, such as a PRFAQ or some narrative. I've always assumed I was just "programming" myself for these modes, and the music was reminding me of the mode I was in. Perhaps it's a little of both.

kator··on Engineering Sleep
This article makes me want to get a DNA test. In my family, it’s very common to sleep in the six-hour range. I personally sleep from 10 pm to 4:25 am every day, often waking up around 4:15 am before my watch vibrates to wake me.

If I sleep eight hours, I feel groggy, jet-lagged, and generally have a day where I’m slogging through molasses to get from one task to the next.

My wife has raised concerns about my sleep pattern, so I started using sleep-tracking tools like Fitbit and, more recently, an Apple Watch. She tracks her sleep too, and the big difference we’ve noticed is that I fall asleep within about two minutes, and my “sleep efficiency” using these tools is 98%. If I’m traveling and feel a bit jet-lagged, I can take a 20-minute nap (often without an alarm) and wake up feeling refreshed. She also seems to wake up a lot, most nights I "sleep like a log" and I only wakeup in the morning.

My mother has the same pattern but stays up later and sleeps about six hours into the morning. I used to do this too, but around age 23, I switched to an earlier bedtime and a consistent daily routine. When I became a “morning person,” I found I could code like crazy in the morning before “starting” my day, and this rewarding experience reinforced the habit.

I’ve tested this pattern in many ways, including not using an alarm (I still wake up around the same time for weeks at a time) and using a “light clock” I built with a Raspberry Pi to slowly brighten the room. Again, I wake up after roughly 6 hours and 20 minutes. Now, I use my Apple Watch to vibrate as a gentle reminder to start the day. On weekends, I keep the same schedule and use the extra time to read or hack away at side projects, often coding until the late afternoon when my wife protests enough that I need to stop and hang out or do my honey-do's.

About 10 years ago, during my annual checkup, my wife asked my doctor about this sleep pattern. The doctor asked me several questions, seemingly looking for signs of sleep deficit or dysfunction. In the end, he said I could do a sleep study but concluded, “If it works, don’t break it.”

As for productivity, I’ve found I can code effectively from 4:30 am to 8:30 am, then shower and work from 9 am to 6 pm without much trouble. I also practice intermittent fasting, typically eating only at 6 pm, with a protein shake around noon. This habit happened by accident—I realized breakfast slowed me down, and eating lots of carbs impaired my cognitive function and ability to code or handle complex tasks in the morning.

Before you ask, I generally don’t use caffeine or other stimulants. Occasionally, I’ll have one cup of coffee around 9 am as a social habit, but I recently stopped that again and actually feel better. I’ll most likely drop it again for a while until it sneaks back in again.

kator··on Cats are (almost) liquid
Interesting because I have recently been trying to catch a stray cat for a capture-release process and the cat will not walk into a typical trap-door type wire mesh trap. Watching him on video the roof of the trap seems to freak him out. It seems a better trap would have a narrow gap with high door that lets them confidently walk into the trap and trigger would just block the slot perhaps with some sort of sliding door blocking the exit.
kator··on AT&T says criminals stole phone records of 'nearly all' customers in data breach
I read an article in wapo that said you can use this URL to see what data was exposed: https://www.att.com/event/lander
kator··on Ask HN: How bad is the xz hack?
My favorite was sendmail's WIZ followed by SHELL bug.
kator··on Fdupes: Identify or delete duplicate files
Funny to see this here, recently I decided to clean up my personal NAS that had 60 million files and 17 TB used. I tried various tools, but this NAS was the backup target for many of my machines over a decade or more of time. I finally gave up and wrote a multi-threaded c++ program to "inventory" every file and then I imported that into sqlite and started building out scripts to do bulk actions, set a flag on for files to delete and some scripting to show the files that would be deleted etc. It was quite a project and took me "spare" time of a month to trim.

On a related note in 2019 Apple destroyed all my music copies in one of the itunes updates, and part of this process I gathered all forms of media into various zfs datasets and de-duped and managed to find all my mp4's from the '90s lol.

It's amazing how much digital garbage we can accumulate, I often wonder what my kids will do when I pass, will they try and sift through all of it, or just toss it in a closet somewhere to be discovered in 1,000 years by some digital archeologist?

kator··on Have attention spans been declining?
just cut-paste in ChatGPT.. why bother reading it..
kator··on Bottles – Easily run Windows software on Linux
+1 I use CrossOver, I bought the lifetime license to support them since they also upstream crazy stuff like running 32bit Windows EXE in MacOS etc.
kator··on Twitter now requires an account to view tweets
New Coke anyone?
kator··on Rooms.xyz
This reminds me of Metaplace[0][1] back in 2007 created by a MMO veteran Raph Koster. Not sure how they will avoid the same issues that killed Metaplace and other attempts at this over the years.

    [0] https://www.raphkoster.com/2021/09/02/online-world-or-metaverse/
    [1] https://en.wikipedia.org/wiki/Metaplace
kator··on Retro Computer Museum
I would gladly donate on a monthly basis to a patreon or something to keep it going. Wonder what the monthly costs were to operate it.
kator··on Retro Computer Museum
While this looks nice, I really loved https://livingcomputers.org/ when it was alive. Sadly it's in stasis now, I was lucky enough to visit while it was still open and I ran around to all these old machines punching in code and showing my wife all the old machines I've worked on in my 40yr+ career.
kator··on Console Screendumps
Am I over simplifying or does this in general do the same thing with some help from imagemagick on linux with /dev/fb?

    convert -size $(tr ',' 'x' < /sys/class/graphics/fb0/virtual_size) -depth 8 bgra:/dev/fb0 snap.png
kator··on Asking ChatGPT to write my security-sensitive code for me
I engage with ChatGPT daily now on a number of topics. In general, I've been trying to search on google and talk to ChatGPT about the same topic, of course I don't put three words in I try to use natural language with ChatGPT. (I had a stint in NLP systems for 5 years around 1997, so that might bias me a bit on how I engage)

What has impressed me is some stuff like "Can you summarize this for me?" and "How would you parse the datetime out of this log entry in python3: {raw text}", "How could I make the following mysql query more readable?" etc.

At this stage, it's like when Stack Overflow came out. And yes, some SO stuff is crap, but once in a while, you get something that saves you 3 hours of your life. For example, recently, ChatGPT has saved me hours of poking around on topics I wanted to solve quickly without thinking so I could get to the high-value work that would get me closer to my goal.

That said, I am amused at how it can bald-face lie about even mathematically incorrect things, and it gets lost if a thread gets a bit too long.

Something is happening here. I think it'll be a while before these things write code from reading a paragraph from a product manager or a less technical user's "use case".

What's interesting is to watch this pendulum swing back and forth, from expert systems coded by hand to neural nets and now these large language models. If the pendulum keeps swinging, it might land on your head one day if you don't pay attention.

All this said, I enjoy my interactions with ChatGPT more than most SO posts, so I continue to use it, and luckily I have 40 years of coding experience to help me identify where it's a bit off.

I have taken to pasting questions from my mentees into ChatGPT and sharing the result and suggesting they try ChatGPT to learn python3 in addition to SO and other tools. It seems to help them, I worry it will confuse them with a bald-faced lie, but I'm here to help when it does!

kator··on GNU nano is my editor of choice (2021)
By that logic we should just make "ed" the default editor. It's more like every other line oriented CLI interface element in the system than Nano is.
kator··on General guidance when working as a cloud engineer
Don't forget "pets vs cattle", thinking of servers as ephemeral and working towards quickly being able to scale up/down based on demand. So often I see people "lift and shift" from a dedicated server model into the cloud and never convert their pets into cattle. This reduces flexibility later, not to mention makes it harder to respond to patching needs, scaling, and moving to optimize latency or costs.
kator··on I fought the PayPal and I won
Yes very well written and reminds us all that in the end we're humans, and once in a while things can go right if only for us asking why they aren't.
kator··on An Accident at SpaceX
While it is a sad story, it reminded me of a parallel issue with another "Race" for technology:

https://en.wikipedia.org/wiki/Demon_core

kator··on 90s Cursor Effects
When can we get these added to iTerm2?
← PreviousPage 2 of 17Next →