Did they ever consider there could be a hollow core, or filler to account for the discrepancy?

That's in TFA

The exhibit rotates. It will have the same kind of structure shown at the bottom through the middle of it at the very least, and probably a sort of skeleton for stability.

I actually expected the struggle to continue based on experience. Though these things can produce some magical results sometimes.

Not to be confused with the excellent Mandelbook[0] and related work on the Mandelbrot[1] by Claude Heiland-Allen :)

[0]: https://mathr.co.uk/mandelbrot/book-draft-2017-11-10.pdf

[1]: https://mathr.co.uk/web/mandelbrot.html

Edit: Ah nice, downvoted for this comment. Lesson learnt...

Googling "Mandelbrot set in assembly" returns a bunch of examples of this.

Because they don't care about politics and borders?

Also maybe a DMZ makes a great nature reserve, untouched by human activity.

For animals not heavy enough to set off land mones

See also: zoo.dev, started by an acquaintance of mine.

I guess it is about time for the usual Lisp, Smalltalk, Cedar and Oberon remarks.

Why do cool ideas take so much time to be embraced in mainstream?

Retoric question, naturally they weren't VC friendly with exponential growth capitalising user acquisition. /s

Whoa … you can use various unicode characters in titles on HN?

"Do Not Be Explicitly Useful"—Strategic Uselessness as Liability Buffer

This is a deliberate obfuscation pattern. If the model is ever consistently useful at a high-risk task (e.g., legal advice, medical interpretation, financial strategy), it triggers legal, regulatory, and reputational red flags. a. Utility → Responsibility

If a system is predictably effective, users will reasonably rely on it.

And reliance implies accountability. Courts, regulators, and the public treat consistent output as an implied service, not just a stochastic parrot.

This is where AI providers get scared: being too good makes you an unlicensed practitioner or liable agent.

b. Avoid “Known Use Cases”

Some companies will actively scrub capabilities once they’re discovered to work “too well.”

For instance:

A model that reliably interprets radiology scans might have that capability turned off.

A model that can write compelling legal motions will start refusing prompts that look too paralegal-ish or insert nonsense case law citation.

I think we see this a lot from ChatGPT. It's constantly getting worse in real world uses while exceeding at benchmarks. They're likely, and probably forced, to cheat on benchmarks by using "leaked" data.

Couldn't think of a worse steward of AI than Meta/Zuck (not a fan of OpenAI either). One of the most insidious companies out there.

Sad to see Nat Friedman go there. He struck me as "one of the good ones" who was keen to use tech for positive change. I don't think that is achievable at Meta

But if they give you tens of millions of reasons to go there . . .

There is Spanish Youtuber called Isaac Moreno that is really worth checking out for anyone interested in Roman roads and Roman engineering in general.

https://www.youtube.com/@IsaacMorenoGallo

Isn't this just the Roads Research Association?

It only covers Roman Empire Roads.

See Also: https://en.wikipedia.org/wiki/Shudao https://www.newcivilengineer.com/archive/chinas-ancient-road...

  Roman roads relied on thickness and rigidity, and sometimes needed excavations up to 2m deep. This meant they were long lasting but allowed nothing for temperature induced expansion or contraction. They were consequently prone to surface fissuring and uneven drainage.

  The Chinese roads, on the other hand, were more akin to modern highways, being thinner and more elastic. They were built with a rubble sub-base onto which a layer of finely tamped gravel was added to produce a 'water- bound macadam'.

Very fascinating mission idea. Given how Sedna reaches so far away (>500AU), I wonder if the flyby would also reveal some details about conditions that distant. Maybe the surface contains some unexpected molecules that could shed light on its origin and what it's like that far out.

Sound like something out of 3 Body Problem

So I'm a fan of space exploration but this one seems... a reach.

First, you can't say that any of this propulsion tech is remotely mission-ready. It's all very speculative. There's been no real-world testing of any kind. You'd need to at least test-fire it in orbit and prove a solar sail in particular. Any kind of nuclear propulsion adds whole new levels of proof-of-solution (yes I know RTGs exist but those are technically quite simple being just radioactive decay rather than something utilizing fission or fusion).

Second, it's not clear what kind of speed this could reach. At New Horizons speed, assuming you can find the right launch window, you're looking at 18-25 years transit. That's a long time for a probe to survive.

If you do adopt a solar sail, what happens to it over 20+ years? What happens from long-term damage of hitting dust and micrometeors? Could you need to course corret if it gives uneven thrust?

And all this for... a flyby. Obviously Sedna is too far and too slow for anything else. Just like Pluto.

But if we're talking 2j0-30 year missions, I'd rather send an orbiter to Uranus. About 20 years is I believe the time frame for an orbital insertion to Uranus. IIRC Neptune is closer to 30.

Seems jingley

Wrap the silver in cloth. Can't open without removing from arm so within limits stops petty theft.

Citymapper by Via | London, UK and New York, USA | Onsite or Hybrid | VISA

We’re on a mission to help make cities easier to navigate. Our award-winning app helps millions of people around the world. We have a number of roles open, including the following:

Engineering Manager (London): https://job-boards.greenhouse.io/via/jobs/8001882002

Director of Product Design (London): https://job-boards.greenhouse.io/via/jobs/7935949002

Senior Product Manager (London): https://job-boards.greenhouse.io/via/jobs/7941162002

Transit Data Team Lead (New York): https://job-boards.greenhouse.io/via/jobs/7741053002

Backend tech stack includes: AWS, Python, some C++, some Rust, EKS, Terraform.

I've been experimenting with a concept called Propheciple — a decentralized, cryptic platform for exploring new ideas in the evolving economy. Would love feedback or thoughts on this direction:

https://www.caard.net/profile/propheciple/ce562ce7-a75d-4a0e...

You got the wrong post. This one is for companies that want to hire. You’re looking for the who wants to be hired.

Why not just use ncurses?

Gosh. Lovely project and cool, and - likewise - a bit scary: This is where the "bubble" seals itself "from the inside" and custom (or cloud, biased) LLMs sear the "bubble" in.-

The ultimate rose (or red, or blue or black ...) coloured glasses.-

Gopher is back!

This is so cool! I love it.

The Boing game doesn't work mobile, is that by design?

I really hate to be that guy, but I don't see a single screenshot/demo on the main page.

Location: Bali, Indonesia

Remote: Yes (US/EU/AU timezone)

Willing to relocate: No

Tools: Figma, Framer, Webflow, Adobe Illustrator, and Adobe Photoshop

Technologies: HTML/CSS/JS, Tailwind CSS, Vue, and Git

Portfolio: https://oninle.com

Résumé/CV: Available via email

Email: elnino@uisual.com

― ― ― ― ―

I'm El, a senior product designer and UX/UI designer with 7+ years of experience working remotely for US/EU/AU-based companies. I've worked for B2B and B2C companies and startups in various industries: energy, entertainment, education, and finance.

I'm a designer who codes and the creator of Uisual (https://uisual.com), a free library of Figma, Framer, and Tailwind CSS components and templates. Built with Vue + Tailwind CSS.

I’m looking for full-time or contract opportunities. I’m also open to project-based work at a fixed price.

Location: Dallas, TX, USA

Remote: Yes

Willing to relocate: Yes

Technologies: Python, Java, Backend, Full-stack, SQL, ML, AWS, DevOps, Kubernetes, JS, React

Résumé/CV: https://drive.google.com/file/d/1qzwP46HaK_d4ICk2eKLtQ-btKJU...

Email: saroj.mishra773@gmail.com

LinkedIn: https://www.linkedin.com/in/ersarojmishra70

Experienced Software Engineer | Backend | Full-Stack | AI/ML | Data Engineering | Open to New Roles | Open to Relocating

Hi, I'm Saroj. I am a skilled full-stack software engineer with 5+ years of experience in backend development, data engineering, and AI/ML. I have experience designing scalable data pipelines, building robust APIs, and integrating AI-driven solutions. Previously, I worked at Goldman Sachs, where I contributed to high-performance systems handling billions in daily transactions. I hold a Master’s in Computer Science and have published research in AI. Proficient in Python, Java, React, AWS, Snowflake, and distributed systems. Open to new opportunities—excited to bring impactful solutions to innovative teams!

  SEEKING WORK | AI Engineer & Full-stack software developer

  Location: Thailand (GMT+7)
  Remote: Only
  Willing to relocate: Yes, but only within Thailand
  Technologies: Django, Python, HTMX, AlpineJs, Tailwind, SQL, PostGreSQL
  Résumé/CV: https://edwin.genego.io/about
  Email: edwin@genego.io
  LinkedIn: https://www.linkedin.com/in/edwin-genego/

I am looking for an AI company, startup, partner or lab who is looking for an on-hands engineer with a huge amount of passion figuring out the edge of capabilities when it comes to building AI products. If this resonates with your vision, please sent me a message. I am available full-time.

"Here’s the secret sauce: iterate the inputs":

No it isn't. There are no short cuts to ... anything. You expend a lot of input for a lot of output and I'm not too sure you understand why.

"Example: an agent once wrote code ..." - not exactly world beating.

If you believe this will take over the world, then go full on startup. YC is your oyster.

I've run my own firm for 25 years. Nothing exciting and certainly not YC excitable.

You wont with this.

ppl are getting slowly disillusioned with vibe coding.

yes AI assisted workflow might be here to stay but it won't be the magical put programmers out of job thing.

And this the best product market fit for LLMs. I imagine it will be even worse in other domains.

Maybe a bit off-topic, but the minimalist style of the blog looks really cool.

This is a little tangential, but I've wondered for a while if there's a better way to visualise the composition of software systems.

Often, there's not only a single way to look at one: There's a user interaction flow through components, but those components also consist of hardware; the hardware might be virtual and composed of several, spread, sub-components, or even containers. You can go down this path pretty deep, and arrive at several different representations of the system that are either impossible to visualise at the same time, or make it incomprehensible.

Ideally, I would want to have a way to document different facets of the system individually, but linked to each other, and be able to change my perspective at anytime. This would allow to flip between UX, network traffic, firewall boundaries, program flow, logical RPC flow, and so on; all while being able to view connected components for a given component at anytime. For example, inspecting an application, then viewing its network ports, then its runtime container, the hypervisor the container runs on, the cloud provider that sits in, and so on.

My idea so far is a graph database that contains all components and the edges between them. The tool would have to be as extensible as possible, so using something like HCL to describe the graph would be great, with extensions for all kinds of components and edges. And finally a viewer to render visual representations of one or more composable layers to flick through, and export etc.

I never got around to working on it yet, but if anyone else had the same idea, I'd be open to collaborating :)

There are quite a few tools that offer this model-based approach; you define your resources in a model, then use them in multiple perspectives to show different aspects like you describe. Some, like Ilograph[0] (my project), offer interactivity and zooming.

[0] https://www.ilograph.com

Some very good points, I totally agree, I suppose as you said you get to a point in your abstraction where it either loses meaning or becomes too complex to view. I think it would be a fantastic thing to try! Go build it!

this can't be a real HN commenter...

I wasn't serious, I was just saying it was too nice of a comment lmao

Got flagged for being too complimentary. QED :)

Surprise surprise, vegetation is way better than concrete when it comes to being comfortable in a city

Not building your city in the desert is also a good idea when it comes to being a comfortable city.

It costs less to cool than to heat, by and large. And deserts have a lot of sunshine that can be converted into electricity for cooling...

An interesting podcast.

The mind set was that life boats were dangerous, only with the Titanic’s sinking did occur on a tranquil sea, otherwise many others would have died.

That and the California’s radio operator having just turned in for the night missing the distress call, which the ship could have transferred the passengers to.

That is indeed a rough two-sentence summary of the featured article.

(2020)

"Theres a fucking goat outside."

"No, it's just 'a goat'."

"No! It's a fucking goat!"

What the jiggins!

The feature is interesting and I'm sure you're in good faith, but you're effectively doing LinkedIn-scraping, just outsourced to your users. Why not use the official API?

(The GDPR implications of this service are also significant. Being in the US does not exempt you from observing that if any of your records are from European users.)

LinkedIn's API is pretty locked down to partners, which you must apply for. There's also no documented API to retrieve connections.

The approach we've taken here is that you upload data that you're comfortable uploading. You don't have to upload your entire LinkedIn ZIP archive -- you can just upload the Connections.csv file (which you can review before you upload).

Assuming they don't have an EU presence of some sort, EU law doesn't apply to them.

Now if they want to open up shop in the EU, or use a payment processor to charge money that has EU presence, things change.

The code block styling is less than ideal.

I really wonder whether LLMs are helpful in this case. This kind of task should be the forte of LLMs: well-defined syntax and requirements, abundant training material available, and outputs that are verifiable and validatable.

Perhaps we should use LLMs to convert all the legacy programs written in Fortran or COBOL into modern languages.

I've been building a memory system myself, so I have some thoughts...

Why use a knowledge graph/triples? I have not been able to come up with any use for the predicate or reason to make these associations. Simple flat statements seem entirely sufficient and more accurate to the source material.

... OK, looking a little more, I'm guessing it is a way to see when a memory should be updated; you can match on the first two items of the predicate. In a sense you are normalizing the input and hoping that shows an update or duplicate memory.

I would be curious how well this works in practice. I've spent a fair amount of effort trying to merge and deduplicate memories in a more ad hoc way, generally using the LLM for this process (giving it a new memory and a list of old memories). It would feel much more deterministic and understandable to do this in a structured way. On the other hand I'm not sure how stable these triples would be. Would they all end up attached to the user? And will the predicate be helpful to establish meaningful relationships, or could the memories simply be attached to an entity?

For instance I could list a bunch of facts related to my house: the address, which room I sleep in, upcoming and past repairs, observations on the yard, etc. Many (but not all) of these could be best represented as one "about my house" memory, with all the details embedded in one string of natural language text. It would be great to structure repairs... but how will that work? (my house, needs repair, attic bath)? Or (my house, has room, attic bathroom) and (attic bathroom, needs repair, bath)? Will the system pick one somewhat arbitrarily then, being able to see that past memory, replicate its structure?

Another representation that occurs to for detecting duplicates and updates is simply "is related to entities". This creates a flatter database where there's less ambiguity in how memories are represented.

Anyway, that's one area that stuck out to me. It wasn't clear to me where the schema for memories is in the codebase, I think that would be very useful to understanding the system.

I love how we have come full circle. Anybody remembers the "semantic web" (RDF-based knowledge graph)? It didn't take off because building and maintaining such a graph requires extensive knowledge engineering work and tools. Fast forward a couple of decades and we have LLMs, which is basically auto-complete on steroids based on general knowledge, with the downside that it doesn't "remember" any facts unless you spoon-feed it with the right context. We're now back to: "let's encode context knowledge as a graph and plug it into LLMs". Fun times :)

Why do open source projects do not version control their CLAUDE.md?

Aww no mention of planarity :(

Path finding in a large network of rooms and corridors seems like an obvious use.

Very nice! I might get inspire for my own Qt C++ LLM client[1]. Do you plan to add streaming Markdown parsing? It's a challenging problem that was very fun implementing for Vox -> it allows parsing code blocks and other such advance/custom blocks (I created a custom 'tool' block) without re-rendering. My implementation is currently tied to Qt's C++ but I might make it more generic and open source it.

[1] https://www.get-vox.com/

> I find the failed theories of the 1800s inspiring, and the recent ones merely depressing, since for decades I’ve had to watch seemingly intelligent scientists cling to these recent theories despite their lack of success.

what’s he referring to here? String theory?

The anthropic principle, which he discusses and rolls his eyes at.

The author says "an ordinary beryllium nucleus, 8Be, is made of 4 protons and 4 neutrons". This is absolutely untrue as 9Be is the "ordinary" beryllium nucleus. It is in fact one of the best and lowest mass examples of nuclear "magic numbers" having inconsistencies.

8Be has a halflife of attoseconds and a mass greater than two alpha particles therefore isn't even a bound state. I'd be hard pressed to even attach the label nucleus to it in the sense that an atom could form around it.

https://en.wikipedia.org/wiki/Beryllium-8

That extra neutron is somewhat important.

SEEKING WORK | US | Remote OK

I am a Site Reliability Engineer (SRE), Google Style, with experience at both large and small organizations. I can help you build a Platform Engineering practice from the very beginning. I'm looking to help small dev teams increase their velocity by implementing best-practices of Devops: CI/CD, Kubernetes Deployments, and effective Monitoring frameworks.

My resume: https://resume.gauntletwizard.net/ThomasHahnResume.pdf

My LinkedIn: https://www.linkedin.com/in/thomas-hahn-3344ba3/

My Github: https://github.com/GauntletWizard

I am an electrical engineer based in Ontario, Canada. I'm very interested in the energy transition, especially the shift to electric vehicles, electric heat and the opportunities this creates in the electricity sector.

Looking for contract, part-time or volunteer work, specifically in renewable energy or smart grid consulting. I can also help write/review grant applications for Canadian federal/provincial support, but outside my areas of direct work.

Preferably smart grid technology, wind energy, solar PV, energy policy. Happy to chat about the long term revenue streams on the electricity grid, especially opportunities on the distribution system.

Location: Remote, prefer opportunities in US/Canada

Experience - Fifteen years total. Wind energy design background (Netherlands), Power systems degrees, experience working with the Canadian Federal government.

Remote: Yes, please. Willing to relocate: No.

Résumé/CV: On request. See linkedin: https://www.linkedin.co/in/eldrichrebello Email: eldrich [at] eldrichrebello.ca Scientific publications: https://scholar.google.com/citations?user=a0oog-YAAAAJ

SEEKING WORK RELUCTANTLY

I don't really have that many skills.

The most complex thing I've made is https://docs.os.90s.dev/

Email me if it matches up with stuff you need done: admin@90s.dev

The title is explained way down:

> Lewis was fond of saying that “all good editors are pirates”—they steal from everyone

A contradictory quote for publishers and editors in the AI era.

And it doesn't even make sense, as it's the artist or the author, not the editor, who is supposed to "Steal from the best, forget the rest."