russellbeattie

I only recently learned that when using the MacOS trackpad with the terminal, you can simulate a middle button click to paste: Highlight some text anywhere in the console, drag it a little and let go. As you pull the text, a little green arrow will pop up and when you let go of the drag (anywhere) it'll paste the text into the prompt, just like middle click does.

Maybe everyone has known this forever and I'm the last to find out, but I thought I'd share as it was a revelation. I middle click all the time when using a mouse, so it was nice to finally discover a replacement while using the trackpad.

Why aren't they just doing it? The cost for developing, launching and hosting a new social network app is minimal compared to what it was 22 years ago.

Part of it is how cheap and easy cloud providers have made assembling all the services you would need to host and scale the app, the other part is that an AI coding agent can do literally all the development and sys-admin tasks that would be needed to get it done.

It would take an experienced full-stack developer about a week or two of effort to direct and manage the development of the entire web app, plus mobile, plus infrastructure, including everything needed to set it up and launch it online. Given a "social network" is a well known category of app? It might even take as little as a couple days.

There's almost no reason for signups given the name recognition already that I can see. Throw the app together, put a "donate" button on the sign up page to pay for the first few months of hosting until ad revenue kicks in and be done with it.

"AI" is now an unword. The all-powerful delusional sociopath in the White House has decreed the latest bit of Newspeak, and the spineless cowards throughout the government are scrambling to make sure to stay in lockstep with The Party.

This U.S. Attorney is obviously frightened he'll be taken to re-education should he use banned terminology.

In a normal world, I would praise the government for taking a action to address issues with a program that has definitely been abused by major tech companies for a long time.

However, it's really hard to separate the message from the messenger here.

Beyond the fact that this is obviously a political stunt designed to make headlines just a few weeks before the mid-term elections, there is little reason to believe that this is being done in good faith targeting the worst offenders. The people who have been chosen to run the federal government are simply corrupt sociopaths who will do anything to promote their extreme ideological agenda and are not to be trusted in any way.

Until proven otherwise, I have to assume this effort has to do with kickbacks or political retribution mixed with blatant racism, rather than for any sort of public benefit. I doubt this will result in anything besides a larger donation to the Republican party - or direct payments to Trump family members - by the affected companies.

I truly hope this ushers in the end of Markdown as the preferred AI document format.

It's time to promote an .htmd HTML document standard so that we can actually have nice looking documents again. Remember columns? Colors? Typography? Layouts? Graphic design? Interactivity?

Just think - we could actually have a true rich text standard! Imagine if it were adopted everywhere and you could send WYSIWYG bold, italicized or underlined text as easily as you can a custom skin-colored emoji?

It would almost be like were living in the 21st century again!

Muse Spark 1.3: lighthouse

The caveat is that this was done using the phone app, and I've been playing with it since it launched, so who knows what it sent in the initial context that could change the inference math.

Actually, that makes me wonder: Did you do all that testing via a harness or via a straight API call where you control the entire system prompt?

I'd be willing to bet that using the same model from different harnesses produce different results, but I'd have to test.

This comes up every thread. I think we've all noticed how similar they are becoming.

I would guess that they've definitely been trained on previous results, as they obviously share way too many traits at this point to be totally random. That said, I don't think we're seeing any signs of pelicanmaxxing yet from the providers, so it's still a useful (or at least fun) benchmark.

Once all the models produce pristine, elaborate pelicans riding perfectly drawn bicycles, then it'll be time to move on to pigs driving a racecar or something.

This is really sad - Triumph of the Nerds played a big part in my move to California from the East Coast at the beginning of my career in the 1990s and I've rewatched the series regularly ever since... RIP Bob.

Oh, I really meant the latest trend of every AI agent having their own dot folder in root. I think Anthropic set off the trend with ~/.claude/ and I really wish everyone would stop now.

It's taken me a while to get used to "cargo cult coding," but I'm surprisingly OK with it now. (I, for one, welcome our new AI overlords.)

Do these tips work? Probably. Do we know why they work? Sorta. Could we be just copy/pasting whatever text someone else kinda thought worked one time and now enters every time out of some weird feeling that it helps nudge the LLM, like some remote islanders building airport control towers out of bamboo in the hopes of a new airdrop from the gods?? Totally. But here we are.

Maybe it's just my pet peeve, but it's less about my documents and mail and more about programs, tools and AI harnesses deciding they can fill hidden dotfile folders at root with whatever they want (which they do indiscriminately). I don't just hate that there's tons of untracked caches and temp file data that isn't easily discoverable, but more specifically, I don't want all that crap in my root directory.

Who decided that hidden folders are a good thing anyways? .agent, .agents, .aws, .bun, .cache, .cargo, .claude, .codex , .config, .docker, .gemini, etc. I just end up having to show hidden folders all the time, which defeats the point.

It's gotten truly ridiculous. I want the OS to strictly enforce my root directory. There should be a few global preference files in there for like .zsh, and everything else organized in proper, unhidden folders.

I agree with all of that - I'm just saying it must have happened in 1997 as the modems came out too late in 1996 to be particularly useful right away except maybe for your local BBS. I can't remember though - 30 years is a long time - so I'm willing to accept maybe I'm incorrect. I do remember how expensive the modems were - I don't think I ever upgraded, but instead got a DSL line.

I'm sure someone pointed this out in the other thread, but 56kpbs modems essentially didn't exist in 1996. They launched at the end of the year, cost a ton, and the manufacturers hadn't standardized the speeds, so it really didn't matter much as you weren't browsing AOL, MSN or Compuserve with it yet.

33.6k.rip is a harder domain name to get though, so I get it.

Wow, that "bake off" page is better than any coding benchmark I've seen! You can really sense the strengths and weaknesses of each model/harness combo.

I just tried it out - it's pretty cool. The postscript was actually noted in the generated video about this topic, which I find amusing.

I have to say, if HN were to add an AI generated paragraph summary about each link at the top of each comments page, it would probably go a long way to improving the discussions about a lot of topics. The number of people who comment based on the title of a link alone is pretty high, and I'd bet the vast majority of commenters only skim the linked pages anyways. Might as well get everyone on the same page with a quick overview.

I've been working on a custom orchestration system which starts by creating an overall work plan, then splitting up development tasks among agent/model combos depending on their capabilities/expense, then coordinating their output, testing, etc. with various adversarial agents acting as gates throughout the process. It's really satisfying to watch the task board as agents go through their little routines.

It's been an absolutely fantastic way to set tokens on fire and watch them burn, and produce pretty much nothing of value.

I amusingly was doing a "cleanup" of my various harnesses when this happened. Claude Code has a /doctor command which was actually quite helpful, so I decided to swap over to Codex and ask it to do a self-analysis and give me some recommendations if there's something I need to clean up. The response was a connection issue. And so was the next. And the next.

Took me the past 20 minutes to finally check for a status page and realize I hadn't broken something. Not before I created a new project, checked the settings to see my usage numbers, closed and opened the desktop app a couple times, logged out and logged in, used my phone to see if it was a general ChatGPT outage, and generally messing around trying to figure out what I had done.

I mean, you never know... it could have been my prompt that took the whole service down. (Sorry if it was).

Nicely done. I asked Muse for the Hatch binary the day it launched (I actually left a comment about it - literally the last one I made on HN), but I just scanned the strings of it to get a general idea of what it did and left it at that. It didn't dawn in me to ask Antigravity to analyze it.

This teardown answers pretty much all the questions I had after I looked at it. Great job.

Agreed. I'm not a fan of Meta in any way, but I've had fun messing with the app this afternoon - both on my phone and the web version at muse.ai - to see what it could do.

Like the announcement said, it's running in its own VM, but it's an impressively transparent system: You have access to its System Files right in the GUI, which surprised me. You can look at the AGENTS.md and other harness files it uses as well as some shell scripts, bash files, cronjob folder etc. It seems a lot more obvious to me how Muse works than say Claude or ChatGPT, which don't make it nearly as clear where stuff lives and how long its there for. I mean, I get they use VMs as well, but there's definitely no "System Files" menu option in those phone apps at least. The Muse bot seems much more "permanent" in my mind as a result.

The AI is quite helpful as well. It let me download its 350Mb "hatch" binary which runs the bot/harness/agent (written in Rust), plus it zipped up the library of skills it uses and all the system files in my VM for me to look at on my machine. Nothing extra special, but it was nice to browse through all the files and the binary strings to see how it ticked. It also explained the limits of the VM to me - for example, that I can't SSH into the VM as it runs through a proxy and can't be reached directly, but was willing to help me add itself to my Tailscale network.

It's basically a nice looking UI on top of a typical agent harness using standard Markdown files, all running in a little Linux cloud VM spun up just for you. And Muse Spark 1.3 is finally good enough that it was actually useful (and fast), so I can see why they launched the app/service now.

From a non-geek perspective, I asked it about selling my old Nintendo Switch on Facebook Marketplace - something I've been meaning to get around to for about a year or so now - and it said I can upload a picture of all the stuff I have and it'll check prices and create multiple listings for me. That alone might make me hate Meta a little less after this if it works as advertised.

> "As part of the task, they were supposed to have the ability to read the internet but not to write on it."

There's no such thing as only having "read" access with HTTP. If the website saves part of the GET URL as input in any way - like through standard "?input=my+text" params - then the AI has the ability to "write". The way servers use the request URL is completely arbitrary - the params could be "/input/my%20text" or entirely encoded in some way - there's no way to completely prevent this.

Agreed. Though apparently you're being downvoted for a legitimate issue. I'm laying down using my phone and it's not particularly enjoyable to read.

It's 2026 - the site can look like an old-skool zine being displayed with monospace fonts on a computer monitor and easily still have alternate formats for mobile devices. Not having it is just pure laziness.

I wouldn't mind if Reader Mode could deal with the text, but it can't. Oh well.

Nice! I regularly do this with Claude to create little web apps on my phone, so that's great idea to create little Android apps instead!

Amusingly, I started walking recently at a local track. One day while I was walking, I thought, "I should create a little web app for a counter to keep track of my laps." So I asked Claude Code and it whipped up a nice little mobile app with a big "+1" button, a reset, the whole deal, all within less than a lap and a half. Great! A lap later I think, "I should have it keep track of the time it takes me to walk each lap."

As I'm typing out the description, I realize, "You dumbass, you're describing the built-in chronograph feature of the Clock app!" So I just switched to that.

Laws need to be managed like software: There should be a process for testing, user feedback, quick patches for bugs or conflicts, and regular updates to fix issues. If you think about it, both laws and software are called "code".

The problem is our legal system is still based on the waterfall method. Lawmakers try to plan for everything, laws meant to solve one problem face feature creep and create a thousand others, then no one wants to touch anything after launch for fear of making things worse or because that one guy uses the temperature of his CPU as a quick-key and refuses to change his workflow.

Anyways, no law is perfect and never will be, and neither are the fixes.

> "It's a real time-portal to a pre-security internet."

It wasn't that long ago where the network traffic of every website was transmitted in the clear via http, not https. Literally anyone at a coffee shop could listen to the WiFi network and see everything being sent and received in plain text. GMail didn't move to https until 2010, and I think it was one of the first webmail sites to do it. Browsers only started flagging non-https sites just in the past few years.

It already seems far fetched, honestly.

> "Headlong is a complete agent harness with a core of less than 10K lines of Bash..."

Wow. So, be nice or I'll replace you with a very large shell script?

I was pretty surprised at how I both immediately recognized each desktop color, but also how each of them "felt". I could picture the icons and menus in my mind, and their location (left, right, top, bottom) and knew how the mouse would respond as well.

I suspect there's some sort of strong association going on with color and tool use in our monkey brains.

An MBA's rule of thumb is that a full time employee's hourly cost to a business is at least 1.5x to 2x times their salary depending on employer taxes, benefits, offices, travel, training, hardware, perks, etc.

> Then the dedup table, because deliveries arrive twice and the docs cheerfully call this “at-least-once.”

Documenting unexpected or intermittent behavior: The easiest bug fixes of all.

After 20 years my account on reddit was "permanently suspended" after some right wing nitwit reported a comment of mine as threatening. (It wasn't, though it also wasn't kind towards him and his ilk.) Happily, my login still works fine so I just keep using it as a permanent lurker.

I find I spend a lot less time on the site as a result, which really is a blessing in retrospect. I'm pretty convinced it's at least half AI bots now. Given how much AI sites use reddit as a source of "truth", it's not a good sign.

Still, I was among the first couple thousand people to create an account, so it still annoys me that the automated process just happily nuked me. Oh well.

Whoever decided it would be a good idea for ".well-known" to be a "hidden" directory is a complete fool. All it does is provide the opportunity for confusion, misconfiguration, skipped backups, missed git check-ins, forgotten updates and more. Literally the only people a folder like that is hidden from is the whoever is managing the web server.

Sure, if everyone knows what they're doing, it's not a problem. But we all know how long that assumption lasts.