HNHacker News
TopNewBestAskShowJobs

nialse

643 karma · joined June 1, 2021

The opinions expressed may not reflect the views of my employers, my business, or myself.
submissionscomments
nialse··on Less human AI agents, please
Asking for code to manipulate the AST is another route. In python it can do absolute magic.
nialse··on Less human AI agents, please
Interesting. When I code, I want a boring tool that just does the work. A hammer. I think we agree on that the tool should complete the assignment reliably, without skipping parts or turning an entirely implementable task into a discussion though.
nialse··on Less human AI agents, please
That is probably the next step, and in practice it is much of what sub-agents already provide: a kind of tabula rasa. Context is not always an advantage. Sometimes it becomes the problem.

In long editing sessions with multiple iterations, the context can accumulate stale information, and that actively hurts model performance. Compaction is one way to deal with that. It strips out material that should be re-read from disk instead of being carried forward.

A concrete example is iterative file editing with Codex. I rewrite parts of a file so they actually work and match the project’s style. Then Codex changes the code back to the version still sitting in its context. It does not stop to consider that, if an external edit was made, that edit is probably important.

nialse··on Less human AI agents, please
Agreed. I am asking for something beyond the current state of the art. My guess is that stronger RL on the model side, together with better harness support, will eventually make it possible. However, it's the part about framing the failure to do complete a task as a communication mishap that really makes me go awry.
nialse··on Less human AI agents, please
It's not necessarily "ignoring" instructions, it's the ironic effect of mentioning something not to focus on, which produces focus on said thing. The classic version is: "For the next minute, try not to think about a pink elephant. You can think about anything else you like, just not a pink elephant."

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

nialse··on Less human AI agents, please
Agreed. We should not be anthropomorphising LLMs or having them mimic humans.
nialse··on Less human AI agents, please
Codex in this case. I didn't even think about mentioning it. I'll update the post if it's actually relevant. Which I guess it is.

EDIT: It's specifically GPT-5.4 High in the Codex harness.

nialse··on Less human AI agents, please
On point. I'm more interested in what comes after LLMs/AI/AI-agents, what the next leap is.
nialse··on NSA is using Anthropic's Mythos despite blacklist
My point was narrower than suggested. If Mythos is in fact a security risk, then the NSA is one of the actors most likely to already understand that. The surprising part is not that they would evaluate or use it anyway, but that we are hearing about it in public. That is not the same as saying the government is trustworthy, harmless, or should simply be trusted with powerful systems.

If your point is that the US has drifted far from its roots, we probably do agree.

nialse··on NSA is using Anthropic's Mythos despite blacklist
That is expected. What is not expected is us knowing about it. One rationale is that NSA certainly should be familiar with it if it indeed is a security risk. Nothing to see here.
nialse··on OpenAI's new $100 tier targets developers hitting Codex limits
This pushed me to try pi.dev with a remote ollama instance running gemma4:26b on a 24GB RTX 3090. It is not perfect, but it's quite okay-ish with 128k context length. It's not a model primarily for coding I guess, but IIWIW. Thanks!
nialse··on OpenAI's new $100 tier targets developers hitting Codex limits
Manufacturing demand. Create the problem. Solve the problem.

The 2x allowance last month was not of the goodness of their hearts. It’s pumping up the demand AND the revenue simultaneously. The squeeze before an IPO.

How many of us have looked into moving to the next tier during the last 48 hours? I know I have both professionally and for my private use. Moving 20% of $20 customers to $100 increases the revenue for the group with 1.8x.

nialse··on Cached knowledge is not intelligence
- Yes it is!

- No, it isn’t.

Is it intelligence to attribute this quote to Monty Python, or isn’t it?

Remembering things is traditionally considered a part of intelligence, memory that is. The subject has been discussed for ages. Personally I favor the statistical definition based on the observation that given a range of different intelligence tasks, individuals that succeed in one task generally succeed in other tasks. There is shared variance, a g-factor. Intelligence.

nialse··on LibreOffice – Let's put an end to the speculation
In terms of communication: The only clearly communicated message is that TDF is not fit for fulfilling its purpose and likely never have been. As an outsider I would suggest ceding the project and IP to a third party not involved in the historic squabbles and infighting. It would be a service to the community and enable the project flourish!
nialse··on Claude Code's source code has been leaked via a map file in their NPM registry
Paranoia. And also ironic considering their base LLM is a distillation of the web and books etc etc.
nialse··on Everything Claude Saw: A Transparent Account of the Chardet v7 Rewrite
Going forward this may become somewhat the norm. The main thing that has stopped major rewrites or reimplementations under other licenses is the amount of cognitive resources that is needed. Many companies have made similar efforts just to steer clear of obstacles in terms of licenses or licensors that were viewed as problematic. The difference is that we now have cognitive resources on tap and we need to get used to it. What do you want to reimplement today?
nialse··on Copilot edited an ad into my PR
Microsoft injecting permanent ads in PRs? Has this been independently confirmed?

Brought to you by Carl’s Jr.

nialse··on Morgan Stanley lays off 2,500 employees across all divisions
That is 3% of their work force according to the article.
nialse··on Show HN: NanoClaw – “Clawdbot” in 500 lines of TS with Apple container isolation
”I don't care deeply about this code. It's not a masterpiece. It's functional code that is very useful to me.” - AI software engineering in a nutshell. Leaving the human artisan era of code behind. Function over form. Substance over style. Getting stuff done.
nialse··on Ask HN: Have we confused efficiency with "100% utilization"?
The trade off between utilization and latency is rarely understood in organizations. Little’s law should be mandatory (management) reading. Unused capacity is not waste, but buffers that absorb variability, and thus keeps latency down.
nialse··on ISO PDF spec is getting Brotli – ~20 % smaller documents with no quality loss
Who is responsible for the terrible decision? In the pro vs con analysis, saving 20% size occasionally vs updating ALL pdf libraries/apps/viewers ever built SHOULD be a no-brainer.
nialse··on Apple Creator Studio
The one-time purchase version of Microsoft Office is not available worldwide. Where offered, it is reduced to Word, Excel, PowerPoint, and OneNote, with Outlook as a Business edition extra. Individual apps can sometimes be bought separately, but pricing usually makes this impractical. This is to push buyers to Microsoft 365 subscriptions which is the primary product.
nialse··on Ask HN: How to automate aesthetic photo cropping? (CV/AI)
Use commercial tools and services for automating cropping. Google is your friend.
nialse··on Ask HN: Has macOS Tahoe been fixed enough to update to?
Upgrade regret here. What used to be solid performance is now random hangs and unresponsiveness. Most things work but it’s Apples least polished OS in many years.
nialse··on Show HN: Human or AI-made song detector and 100% Private Audio Mastering
Very nice! AI detection checks out, even with heavy editing and processing.
nialse··on Opus 4.5 is not the normal AI agent experience that I have had thus far
After recently applying Codex to a gigantic old and hairy project that is as far from greenfield it can be, I can assure you this assertion is false. It’s bonkers seeing 5.2 churn though the complexity and understanding dependencies that would take me days or weeks to wrap my head around.
nialse··on Agentic Patterns
Note: At the point of writing this, the comments are largely skeptical.

Reading this as an avid Codex CLI user, some things make sense and reflect lessons learned along the way. However, the patterns also get stale fast as agents improve and may be counterproductive. One such pattern is context anxiety, which probably reflects a particular model more than a general problem, and is likely an issue that will go away over time.

There are certainly patterns that need to be learned, and relearned over time. Learning the patterns is sort of an anti-pattern, since it is the model that should be trained to alleviate its shortcomings rather than the human. Then again, a successful mindset over the last three years has been to treat models as another form of intelligence, not as human intelligence, by getting to know them and being mindful of their strengths and weaknesses. This is quite a demanding task in terms of communication, reflection, and perspective-taking, and it is understandable that this knowledge is being documented.

But models change over time. The strengths and weaknesses of yesterday’s models are not the same as today’s, and reasoning models have actually removed some capabilities. A simple example is giving a reasoning model with tools the task of inspecting logs. It will most likely grep and parse out smaller sections, and may also refuse an instruction to load the file into context to inspect it. The model then relies on its reasoning (system 2) rather than its intuitive (system 1) thinking.

This means that many of these patterns are temporary, and optimizing for them risks locking human behavior to quirks that may disappear or even reverse as models evolve. YMMV.

nialse··on Focus apps claim to improve your productivity. Do they work?
If there is no scientific evidence for the method then it’s up to if you believe. Placebo is a hell of a drug.
nialse··on Global Memory Shortage Crisis: Market Analysis
That is the logical conclusion. The era of personal computers is coming to an end. It had a good run though.
nialse··on Versioned Cache Keys Can Save You During Rolling Deployments
Versioning the cache is a poor band-aid and introduces failure modes not considered.
← PreviousPage 3 of 10Next →