Anthropic loves to release these relatively obvious features with hyperbolic names. But since I've switched to Pi a few months ago and have a much deeper understanding of how the harness works - and more importantly, how simple it is (should be), it's obvious this is a just some agent instructions and hooks.
I've had something like this in my personal pi for a few weeks. The difference is I've tuned it for low context use because my priority is to get work done, not sell tokens.
I'm not saying my one would be able to handle a Rust rewrite of Bun. It's designed for plans where the maximum scope is what I can keep in mind and thoroughly test in ~ 1 day.
One of the major breakthroughs I had was having the executor agent (or manager agent, whatever you call it) split the technical plan into a DAG of small, subagent sized tasks, usually red waves then green waves. Sometimes there'll be 10 or 20 subagents in flight, especially for highly parallelizable red waves which is satisfying to watch. Usually I'm using xhigh for the executor and medium for the workers, GPT5.5.
Besides the DAG organization, other critical pieces were tuning the executor to minimize it's context use. Because it's Pi I'm starting a session at about 10k of instructions and I can let a plan run for a couple of hours before I have to compact. And the other one was having the executor fully responsible for git management, otherwise workers would frequently clash and stall.
It's just subagents, although agent teams on tmux chatting to each other looks cool I think it's just overhead in practice. Give workers clear, narrow scope and review when they're done.
Besides that it's a fairly standard workflow: brainstorm to PRD -> technical plan -> execute -> audit.
That person would have been a skilled laborer like a blacksmith a few generations ago. I'm sure many of those people felt the same way when factories started to produce what they had spent a lifetime learning to make.
Now that is happening to many kinds of knowledge workers. Assembly lines mechanized the work of artisans. LLMs are in the process of mechanizing knowledge and creative work, of certain kinds.
No, we want you to backup your claims and provide sources or stop adding pointless low effort anti-EU noise to the conversation. It's frustrating, any time there's any kind of discussion about anything European on HN it gets flooded with shallow, low effort "EU-bad" posts like your contributions here.
If you're going to make that claim at least put some effort in.
This is a mostly American forum and some people want to piss on the EU to elevate themselves. Europeans do the same to the US but about politics, health care, work life balance, and quality of life. You know, the stuff that matters :D
Someone else mentioned that her problems are more likely stemming from her involvement in the neoreactionary (alt-right) movement circa 2015.
I've never heard of this and generally would say people should be given space to grow from things they said ten years ago... But...
> The Dark Enlightenment, also called the Neo-Reactionary movement (abbreviated to NRx), is an anti-democratic, anti-egalitarian,[1] and reactionary philosophical and political movement
> The Dark Enlightenment has been described as part of the alt-right, as its theoretical branch,[15][16] and as neo-fascist.[15][17] It has been described as the most significant political theory within the alt-right,[2] as "key to understanding" the alt-right political ideology
> The Dark Enlightenment has been described as part of the alt-right, as its theoretical branch,[15][16] and as neo-fascist.[15][17] It has been described as the most significant political theory within the alt-right,[2] as "key to understanding" the alt-right political ideology
And it continues with gems like "freedom is incompatible with democracy" and so on.
It's not like she made a few of the cuff statements here. She was so deeply involved in this movement that people were writing articles about her.
> And then there’s Justine Tunney, “co-founder of Occupy,” proud Google employee and self-declared defender of the tech elite.
Tunney does not just flirt with neoreactionary ideology, the way self-congratulatory “open-minded iconoclasts” like me did in high school and college. She goes full throttle in her embrace of it, doubles down on it, rejects every “politically correct” rejection of sexism or racism or classism that define the modern world.
She makes bold statements that IQ, law-abiding or -breaking tendencies and political alignment are all genetically determined.
As someone else mentioned below, this is why she was uninvited from speaking at the Internet Archive. It seems the llama.cpp drama had little if anything to do with it.
At the very least, if she has grown beyond this, I would expect or rather need her to acknowledge it and distance herself and explain how and why she's grown since then. Otherwise, I can only assume that she hasn't and is still basically that same trollish reactionary from 2015 and that's not someone I would want either working at my company or speaking at my convention.
Justine might have - or perhaps used to have, seeing as this was over a decade ago - political views that are unorthodox and unpopular in mainstream tech circles. I don't agree with these views myself. But shunning him forever for this seems wildly disproportionate.
He is clearly a highly intelligent and highly accomplished individual, and an eloquent communicator. Indeed he is exactly the sort of person who organizations like the Internet Archive should be delighted to have speak on technical matters.
It comes across as very petty and short-sighted of the IA to rescind the invitation based on a handful of complainers complaining about non-technical issues that they are personally offended by.
Whatever happened to being tolerant of the perspectives and opinions of others, even if you strongly disagree?
It's a bad take to suggest that organizations should look past a person's loud supportive positions on eugenics because they might give a good tech talk. If Justine is as controversial as this post suggests (and it doesn't even name the controversial views!) why on earth would any organization want to give them a platform?
"Because they're smart" isn't a good enough reason. I wouldn't want to listen to a podcast that hosts eugenicists. I wouldn't want to appear on a podcast that hosts eugenicists. Even if they didn't talk about their problematic beliefs on the podcast.
There's no shortage of smart folks that don't have a complicated reputation. Organizations like the Internet Archive simply don't have to host them. And frankly, it sounds like they gave her the opportunity to explain her beliefs, and she chose not to. In a way, that's worse: it says "I don't want to explain my beliefs, because they're going to make me look worse" or "I don't want to disavow my problematic beliefs because I don't want to alienate people who share(d) those beliefs".
> true mastery is to let your humanity at the door and pull up a facade
Or say "screw that" and go find work that lets you be a human, not a repressed shell. I'm in my 40s now and have followed that my whole life to great benefit. Barring about two months in a open-plan hell hole in my mid twenties which I still look back on and shudder, mostly out of empathy for people who spend their whole working lives that way.
You can get reasonably close with fewer, however more agents give better signal: e.g. if 3/3 flag something as an issue, the outer one that orchestrates them can view it as something to give more attention to, whereas if it's just 1/3, then it probably begs more consideration. Ofc more doesn't always imply right.
It's also great for totally messing up your brain chemistry and your reward wiring. Wouldn't touch it myself.
I'm sure if you did it once or twice a year it'd be fine but let's be real, anyone who's willing to take it in the first place (outside of having a genuine medical reason like narcolepsy/ADHD) will want to take it a lot more than that.
Besides, in the long run - measured over weeks or months - these absolutely will not give you a productivity boost anyway comparable to sorting out your sleep/exercise/diet/mental health.
ADHD stimulants promised a lot in terms of motivation -- but only made me do the thing I already wanted to do instead of work... more.
So instead of plugging away on house work or chores or my employer's boring work, I was building compilers and databases from scratch at 3am, unable to sleep.
And then I checked my blood pressure. Oops.
Also seemed directly implicated in a loved one of mine acquiring an eating disorder.
Sad! It mush have been something amphetamine-based. Ritalin, from experience, makes me visibly grumpy, because everything is wrong! Everything should be properly fixed! The desk should be cleaned. The code should be reviewed, bugs fixed, tests expanded and enhanced. That ticket is annoyingly obvious, it should be done in 10 minutes, dammit. Well, more like an hour, but now it's fixed for real. The chair squeaks, it's insufferable, where are my hex bits? Etc, etc.
It wears out quickly enough though, maybe in 3 hours.
I've had something like this in my personal pi for a few weeks. The difference is I've tuned it for low context use because my priority is to get work done, not sell tokens.
I'm not saying my one would be able to handle a Rust rewrite of Bun. It's designed for plans where the maximum scope is what I can keep in mind and thoroughly test in ~ 1 day.
One of the major breakthroughs I had was having the executor agent (or manager agent, whatever you call it) split the technical plan into a DAG of small, subagent sized tasks, usually red waves then green waves. Sometimes there'll be 10 or 20 subagents in flight, especially for highly parallelizable red waves which is satisfying to watch. Usually I'm using xhigh for the executor and medium for the workers, GPT5.5.
Besides the DAG organization, other critical pieces were tuning the executor to minimize it's context use. Because it's Pi I'm starting a session at about 10k of instructions and I can let a plan run for a couple of hours before I have to compact. And the other one was having the executor fully responsible for git management, otherwise workers would frequently clash and stall.
It's just subagents, although agent teams on tmux chatting to each other looks cool I think it's just overhead in practice. Give workers clear, narrow scope and review when they're done.
Besides that it's a fairly standard workflow: brainstorm to PRD -> technical plan -> execute -> audit.
reply