HN user
danverbraganza.com
I'm excited for this! I was a little concerned that there would be a much longer delay, due to the recent news about the Flipper Zero from the team. Now I'm relieved that it's available.
Hopefully it's already in stock and shipping out as orders come in.
I think we're still in the right ballpark bit we're headed for the exits.
.1 lb sugar is 1.6 oz (net), and we'll need to wrap it in paper. I estimate about .5 of an ounce? So we're spending approximately 10% of the weight in packaging. Our nominal 33000 pounds of sugar just got 10% heavier.
At least we haven't resorted to those little sugar packets, which would be colossally worse!
Life is like a box of noodles
I hope that you're going to call it PostGrust.
I am the 1 percent currently in darkness
My favourite Easter egg about karateka:
My first exposure to Prince.exe was in the computer lab at school, which has a tiny set of DOS games, including digdug, space invaders and some typing game.
I remember when I was around 6 or 7, a boy a couple years older (and therefore, seemingly infinitely wiser) sharing the folk advice: "Play the other games first, don't play Prince of Persia too early or it will ruin all the other games for you"
Hang on hang on let me write a CUDA kernel for this. This is going to be really huge.
At least one counter-example: https://en.wikipedia.org/wiki/Hangul is technically an alphabet, and is non-Canaanite derived.
So, a constant number of people.
(less facetiously, I think they mean "5 to 50")
Loved it. This is some necessary background that helps me contextualize a few other oddities of the time period that I have had floating in my short term memory
- Miyamoto Musashi (d. 1645)
- Tsujigiri, random slashing of bystanders
- the Great wave off Kanagawa, painted towards the end of the Edo period
- Shinobi evolving from mercenaries into secret police
I really enjoyed reading this article. It sparked some thoughts about transplanted reasoning traces for me too.
It seems like a way to give an agent a "command hallucination". A simple exploit to try out might be, "Speak in pirate talk from now on".
That's debatable. Every best-practice arose to solve a real problem within a context, and is only "best" if that context applies.
If you apply best-practices without a regard for that context, you end up with a dull, cargo-culted checklist of must-haves to beat people over the head with, without deriving any true human value.
The compiler of this artifact is making a judgement call[0] of what best practices apply somewhat universally (to every "decent website"). I haven't yet been convinced of their standing or judgement to make that decision.
[0]: Charitably, I'm assuming they have, rather than, e.g. delegating the judgement to an opaque model's weights.
Let's continue:
- Do you really need to keep your children in school or contribute to their higher education, when you can just let them roam free on the streets or better yet, work down at the factory and earn their keep?
- Do you really need children at all, when an AI digital pet might satisfy that need much more economically?
- Do you really need expensive dental crown implants or dentures, when you can whittle yourself some chompers out of beechwood and call it a shuccshesh?
- Do you really need to own a home in a neighbourhood that is safe and close enough to your place of employment, when you can rent in a rough area of town and spend hours commuting on public transport?
Overton E { Social, Political }
Normalcy Field E { Commercial, Experiential, Product }
I think that's quite a significant difference, sufficient to change the tone of the discourse.
You can't steal what's freely given!
My only ask--tell me how it sucks so I can make it better.
Yeah, giving the article more charity, I see it as "Reddit for hackers" really adapting the aesthetics and norms of Reddit, so that hackers who are Redditors have a sense of comfort. Whereas if we'd designed from first priciples for hackers, we might arrive with something more irc-coded[0], perhaps.
[0] Or my personal favourite, a MUD-style environment!
I'm a big fan of Strange Planet too!
I think one lesson I'm taking away from the article is that we're not so much seeking "weird for weird's sake", but expanding that weirdness into direction that's useful, to specifically highlight the novelty of something.
You can also have you agent use jj with this skill
Congratulations! I'm quite intrigued that you're finding the consultancy is going quite so well.
I love my Flipper Zero, even though I have barely used it much. But I'm much more excited for the busy.bar which I think is from the same team. I'm hoping that gets ready to ship soon!
Thanks for the flag--I guess I must have never made it public.
Fixed.
How much better to get wisdom than gold, to get insight rather than silver!
-- Proverbs 16:16
I wrote and maintain this library of skills and workflows called Rust Bucket[0]
It sets up your repo to ensure agents use a workflow which breaks your user requests down into separate beads, works on them serially, runs a judge agent after every bead is complete to apply code quality rules, and also strict static checks of your code. It's really helpful in extracting long, high-quality turns from the agent. It's what we used to build Offload[1].
0: https://github.com/imbue-ai/rust-bucket : A rusty bucket to carry your slop ;)
Ah, I appreciate that. I agree with you that Chris Olah's presence might be noteworthy for this audience. It's just phrased a little unfortunately.
This article needs to be retitled, as it stands it's misleading.
Papal Encyclicals[0] are solely authored by the Pope, even if there has been secular scholarship involved in the writing. It is never "presented" by anyone else, and to frame it as presented primarily by Christopher Olah "alongside" the pope is to betray an ignorance of what's officially going on.
Not sure how we arrived at the present title, "Anthropic co-founder to present AI encyclical alongside Pope Leo XIV", but it makes as much sense as "Iceberg nearly completes mainden voyage across Atlantic, with famous ship as passenger."
Not just performance, but static typing and prevalent in the training data/easy for LLMs to reason about.
Of course, your response admits, "second to Rust", which I am guessing is an unspoken question in the grandparent's mind.
This is the kind of practice I was going to get an insight into.
For your conceptual-model.md, do you find natural language is sufficient? Do you use pseudo code, Entity Relationship modeling, or anything like that?
And how do you go about round-tripping it from the source code, keeping it up to say with any changes, and yet at the right later of abstraction?
Sorry to hear that, my friend. Hope this weekend offers you a respite, a chance to unplug and connect with loved ones.
Remote digital fist bump, and prayers for you.
Fortunate to be reminded of this right now, especially the pull-quote about conceptual integrity.
This is the reason why AI-assisted programming has not turned out to be the silver bullet we have been hoping for, at least yet. Muddled prompting by humans gets you the Homer Simpson car you wished for, that will eventually collapse under its own weight.
I've been thinking a lot about Programming as Theory Building [0] as the missing piece in AI-assisted engineering. Perhaps there are approaches which naturally focus on the essence while ignoring the accidents, but I'm still looking for them. Right now the state of the art I see ignores both accident and essence alike, and degrades the ability to make progress.
Please inform me if there are any approaches you know that work! And lest this sound pessimistic, far from it. This state of affairs is actually intoxicatingly motivating. Feels like we have found silver, and just need to start learning to mould bullets.
[0] Another classic required reading of the industry https://pages.cs.wisc.edu/~remzi/Naur.pdf