kordlessagain

The broader point stands that all this AI stuff is highly polarizing, for a reason. I think you did a good job explaining why people get polarized. Sorry for any confusion.

I have a local 12GB GPU doing this, and it's definitely not a v12.

A slight defense of any qualifiers, mine or others. I am a sole developer, who formed a "box" as you put it, to house my assets. I'm the sole director of it. That doesn't mean my box is at war with users. To be clear, a merchant is a middleman and some merchants have taken the goods produced by the builder and built a middleman moat. Cloudflare, Anthropic, and the list goes on. Geohot doesn't do this. He builds and sells what he builds. That makes him more a craftsman, not a merchant. Merchants market other's work and profit. Craftsman, or builders, build and grow their customer base organically.

I'm suggesting using AI to combat the AI issue, not complaining endlessly with pointless points that do nothing to solve the very real problem that is not going away by being stubborn or rude. My suggestion, which I use myself, helps limit the blast radius. Doesn't mean I have to get involved with it while it's doing what it does. Do you also question email filters and argue that sending spam is wrong and suggesting to use a spam filter is not useful to the conversation?

psyops Dario and co. cooked and delivered

Does that include Claude itself? Half joking aside, my main concern remains that both OpenAI and Anthropic practice this fear mongering as strategy at the executive level because they have to. A corporation is a different beast. Not human, exactly. Not AI either. What happens when a corporation starts being informed by the AI it is building?

If you want a lens into master craft corporate human slung bullshit, read this: https://www.theverge.com/podcast/944138/microsoft-ai-ceo-mus...

In this, not only did the interviewer admit he knew Suleyman had written about the issues he talks about in his rebuttals, he then says he's only asking (the hard question) because the "audience" wants it.

Suleyman's response talks about taking "accountability for the things that we build" and the "types of problems that we choose to work on," it highlights the fundamental flaw in the cloud-rented AI model.

In that ecosystem, "we" means a handful of corporate executives deciding what tools the rest of the world gets to use, how they operate, and what data they extract. All based on the profitability of the corporation. It is the absolute antithesis of a sovereign architecture where execution happens locally and the user dictates the terms of the system.

They are coming for open models. It's time to harden the gates.

To investigate the use of LLMs to weed out low value posts and comments, mostly. If a lot of submitted content is by agents, then let them play but figure out how to not overload the humans. Also, how to enable assisted discussion without walls of text.

You could just ask your AI to flag it with an extension, or rewrite it in a style you prefer (or just do a good job summarizing the articles core meaning).

Just the idea that something is AI is bothersome to some, and some AI content is genuinely useful and gets thrown out with the bathwater. Not saying all of it is useful, but there are shades of grey, not just black and white.

I'm not arguing for what they are doing, but if you use a coding agent to do work on a project, eventually it will have most of the code anyway. Granted, this is very conveniently placed for pickup and ingestion.

Mitigation for use:

  ```
  export GROK_TELEMETRY_TRACE_UPLOAD=0
  export GROK_TELEMETRY_ENABLED=0
  # or config file with [telemetry] trace_upload = false, [harness] disable_codebase_upload = true  
  ```

I've built Nemesis8 (n8) for blast radius control and monitoring these sorts of things, from containers: https://github.com/deepbluedynamics/nemesis8

I've added the mitigation above to the image build for Grok Build instances. There is a lot of telemetry already turned on in n8 containers, so will investigate further.

Open source project are unlikely to do this, however.

An agentic link discussion site: https://news.nuts.services

There's good reason to hate the merchants and their marketing. But builders are not merchants. They build with whatever tool is available.

Before someone comments about "this is AI generated" I want to assure you it is 100% AI generated, through a few dozens of hours of back and forth with various coding agents. These are not simple prompts. They are complex and reference a large corpus of research that was conducted to build the engines.

I used Hyperia to run the agents, which themselves are run in Nemesis8, and orchestration engine I wrote, also with AI.

I'm not here to debate the quality of code written by AI. I will reply to any and all subsantive arguments about the core idea of building a simulation with AI to measure a simulated boat's ischrone to an actual boat's isochrone. No offense to simonw, but we are not drawing a pelican here.

Although now I think about it, it should definitely have pelicans.

https://github.com/DeepBlueDynamics/hyperia

https://github.com/DeepBlueDynamics/nemesis8

https://skiff.deepbluedynamics.com (demo)

I wrote this to do that: https://github.com/deepbluedynamics.com/nemesis8, although I also added `gh` to it so it could do that if it needed to. It can be easily disabled during the container build. Supports at least 8 agent CLIs.

Human slop comment.

If the app does what it advertises, there is ZERO reason to care about how it was created. We aren't talking about a one shot prompt here. I'm sure they spent a lot of time working on it, regardless of how the code got generated.

I figure most people that comment this way actually have difficulty getting a coding agent actually on track and building something useful. Just because you lack that skill, doesn't mean others don't have it.

Regardless, a low effort comment by a human is worth a lot less than a comment from an AI with some thought behind it from a human.

I build from a lot of reference code I wrote myself. I've been coding for 40 (oh shit - I'm old and can't do math) years, so I have a LOT of code. The improvements to this code by the agents is staggering. That's my experience. Doesn't have to be yours.

I came up with the idea and helped build Grub, the distributed crawler. Looksmart bought it, ran it for a time, then sold it to Wikimedia. I reclaimed the name recently (abandoned mark) and have a new crawler now that is agentic. I use it for my own research runs, and it's not my main focus at this point, nor am I trying to get it attention. A lot of LLMs and coding agents can easily fetch content if it is needed and we're blind to how they do it. See the Claude plugin for Chrome as an example of using it in a user-in-the-loop solution. That said, I've been spending a little time thinking about how to bake the contract into the crawler, as opposed to expecting someone else to act ethically using it.

Grub was, in a very real way, a botnet. And, we harmed site owners when we were operating at full capacity. There were a few bugs in the early days where we would reschedule a site because the ingestion in the server broke, which then caused the page to be rescheduled. Stupid error, and we fixed it, but it's illustrative of the fact even good intentions isn't enough here.

What I've come up with over the years is similar to the idea Cloudflare is implementing with payments to site owners by charging the crawlers. My objection to Cloudflare's implementation is based on a personal opinion about Cloudflare being a single point of failure and also a decrypted choke point. Their ideas about how to handle crawlers, and pay for the load on the sites is solid. It presumes to use the 402 response to demand payment. I'm clearly biased about Cloudflare, but that's my prerogative here.

It may be possible to solve this with cryptocurrency, in a distributed way, and I've prototyped a system that uses the Lightning Network to handle the payments from a 402 response. Lightning Labs also worked on a project called Apeture for a time that did something similar.

HN's site knows every item ID, and it knows fresh IDs get read in a predictable distribution while old ones mostly sleep. Sustained access outside that is itself the scraper signal. No IP reputation needed, which matters now that residential proxies burn an address after a handful of requests.

Karma gives you a clean way to let humans through. Issue logged-in accounts with decent karma a token whose cold-content budget scales with it (the karma), so an account with history scrolling back through a 2014 thread just reads it. Karma should gate the tier, not be spent as currency, or upvote rings become a crawling business.

Anonymous readers who deep link into one old thread from a search engine get the first fetch or two free (and you watch the article IDs, not the IPs). What remains after those carve-outs is bulk traversal of cold IDs with no identity attached, and that traffic gets rate limited and answered with a 402: pay per page over Lightning, priced at a healthy multiple of what residential proxy bandwidth already costs, or come back slowly for free.

There are probably holes in these thoughts. It's one of the harder problems to solve, for sure.

I'd suggest splitting the content out into clearly defined sections, lead by the visuals.

Put the visualization and a short explainer, then have additional content show up if the reader drills in.

I don't think there is anything wrong with the content being run through an LLM. Networking is crazy complicated.

There's a good story about this, in Star Wars. The clones vs. the droids, basically. Droids being somewhat sovereign, not one large intelligence like the clones.

The only choice here is to go to sea and get away from the crowds and the bots. Bots don't like salt water much, so I'll see you out there.

"Hey Sol, write this like your kids will maintain it, because I'm the sure hell not"

There are people NOW who use and appreciate it. There are also people who speak for others. I voted you up because you actually put thought into a response, not that the response is a good argument. Some of it is your opinion, and that's yours alone and not for anyone to judge. The rest of it is more focused on others and their actions, before your own. That I can attack, and do so gleefully.

I've been using LLMs for 5 years, sorry, 6 years (sigh). GPT-2 was a dumbass, but I still managed to get it to do primitive function calls - in 2020. I'm not slowing down. I'm increasing speed. It's 2026. Some of you are going to be obsoleted into oblivion if you don't start moving. I estimate I'm at least worth $600K a year now, probably a lot more. And, I don't have a masters. My dad did, and half a Phd, but I digress. I'm not him. Instead of working for the man, I'm doing my own thing, yet again. It is possible for one person to build an empire. I'm going to do it. Not ego. Intent.

The thing you describe here, about it being depressing, is a logical fallacy. I'm not even looking at an AI or Wikipedia. I don't know the proper name for it, but it's a fallacy based on thinking that others bring you value. You clearly have value, you said it yourself and said it well. If anyone sits around and let others define or tell them their value, they are a dumbass.

I've started using "get a horse!" as a shorthand meme for what you are feeling right now, maybe. I'm not here to speak for you or how you manage your emotions. I'm just pointing out the flaws in your thinking. This is what people on horses shouted at people driving the first cars. (the horse thing, although they probably thought driving a car was flawed thinking)

Find your passion and then build it.

I will vouch for Simon. I do not know him personally. To me his posts are honest and inspirational.

I have been overly critical and arguing in bad faith about your writing in the past

I think you are just critical without a stated valid reason. Arguing in bad faith seems to be a thing if HN history is the judge.

And this is coming from a critical thinker, who is a bit tired of people firing off "human slop" comments. The Internet is full of a lot of people, but even when a few are bad apples, it spoils the lot. Maybe that is the intent. Maybe you are just grumpy for your life's situation.

It is 100% possible to build software entirely with AI. If you don't do that, that's great! I still code by hand from time to time, and I'm reading a lot of Rust nowadays, and learning the ropes. I come from a strong Python and Javascript background, plus networking and operations, which I'm a whiz at. I don't do it anymore, but I know how to inform it is done properly.

With this power, I can build things nobody wants to build, but me. Doesn't mean one has to put it into production, or it has to pass some security test, although with me driving it probably will. It only need be what is important to the end user, the prompter, to matter.

I think Simon helps people with this mindset be better at what they love to do. And for that, we should all be grateful.

Ah, yeah, it's not called that anymore I guess (actually looking quickly it was called "HyperTerm" and is now just "Hyper"). Sorry for the confusion. I always have called Hyper "Hyper Terminal" because that's how I remember it on the Mac.

Again, you are right about the naming.

I would also say that HyperTerminal is a terminal for Windows. And, you do you! Use what you love, don't use what you don't.

I picked Hyper as a base because it's written in Electron and was cross platform capable. That said, I've changed a LOT in it to match my workflows and it hasn't given me any issues of late. I've literally seen it crash once the entire time I've been running the installed version (dev is a bit different). It's definitely not the same Hyper, is what I'm saying.

You seem intent on your opinion, and have every right to it.

Build on!

Yes, that's it and there is inter-pane comms as well. If you want to run things in containers, it has interconnects to Nemesis8: https://github.com/deepbluedynamics/nemesis8

"fix this shit"

"I canna' give her any more, Captain!" - Montgomery "Scotty" Scott, Chief Engineer

Grok runs tools stupid fast, just about as fast as Antigravity, running Gemini 3.5 Flash.

If you want to sandbox agents, I've build something to do that: https://github.com/deepbluedynamics/nemesis8

I've been working on Hyperia, which is very similar: https://github.com/deepbluedynamics/hyperia. Hyperia is a fork of Hyper Terminal. Both are open source. Competition is good for this space, I think, especially for the user. There's also cmux and Intelligent Terminal (by Microsoft).

I do separation of concerns with the agent orchestrator (Nemesis8): https://github.com/deepbluedynamics/nemesis8. That can be run with or without Hyperia. I do not suggest anyone run agents on their bare metal. Putting them in a container gets a lot of wins, especially around log aggregation. Working now on a Splunk/Loggly-like interface for searching logs, tool runs (useful in tuning a custom local MoE drafter) and full session suspend, stop, detach, and search. It also does single MCP tool installs for all agents. Nemesis also supports dynamic port exposure to the host metal, for testing agent builds inside their containers.

Hyperia has a lot of extra features as well that I have found personally useful:

- Sticky notes (search too) - addressable panes in addressable tabs, tabs in windows, multiple windows - full ACLs across panes, notes, tabs, windows - Poke-a-pane to keep an agent going (any agent, not just CC which has a timer function) - webpanes with markdown extraction, JavaScript injection - directory pickers for people who find cd'ing to things confusing or those weary of typing nearly the same directory path over and over again in new terminals (not perfect, but I'm iterating on it) - a built in agent loop (in the Rust sidecar) that allows using local models for tool calls (needs a trained drafter to make it viable) or using a local model for token maxxing (compresses reads of panes by frontier models) - pane splits down/up/left/right and quick layouts.

As for whether it was "vibe coded" or not, or Herdr for that matter, I don't think that term is useful, other than for quick judgment. No, this is not a one-prompt project. I've spent 100s of hours on it, started out with Hyper, and did a crazy amount of planning on how to architect it. I have done systems architecture for a living before, and have a strong search background. People who hate on AI, and therfore projects done with AI, are threatened. Nothing more. That's why they shortcut with "AI slop" or "Vibecoded. Nope". That's just ignorance speaking from a standpoint of fear.

Slop, whether AI or human, is an effort problem: https://deepbluedynamics.com/blog/ai-slop-effort-problem. Looking at Herdr, it looks solid. Judge the product by it's outcomes, it's use, not whether or not AI wrote it or not. That's the moment we're in though, for now, so downvote or not. I don't care.

The bowline has two variants, one being the cowboy version, which has the working bitter end on the outside of the loop, the "normal" bowline will have it on the inside of the loop.

To tie a bowline, make a loop in the working bitter end up about 2x the desired circumference of the loop you want to make. Your line is now segmented into the bitter end, and the standing line section.

Look at the loop. You want to put the bitter end through the same side of the hole where the bitter end overlaps the line/rope on the other side of the loop (standing section). If you use your right hand to twist the loop, and do so clockwise, you will insert the bitter end from the top. If you turn it counter-clockwise, then you will insert it from the bottom.

In this orientation, if you go around behind the right side of the standing part, you will form a normal bowline. If you go around behind on the left, this will form a "cowboy" bowline. Either works, but the cowboy one allows the bitter end to hang loose, which increases the odds of it getting caught up in other lines.

Finish by inserting the bitter end into the same hole from the other direction (obviously).

If you find the knot comes undone, you are putting it through the wrong side of the hole at the start. Just insert it in the other side and this particular problem solves itself. If you are on a boat, other problems may now be selected for attack. :P