dematz

gleam fhir types and client https://fhir.hexdocs.pm/

it is relatively new and untested irl, but interesting as gleam is very nice for fhir in some ways:

-fhir choice types imo were originally designed for some kind of object oriented polymorphism, but are nicer as sum types

-cardinality works nicely with Option for 0..1 and List for 0..*, the only ugly part is if you need primitive extensions and suddenly there are a ton of Option fields

-works with whatever http client you need for erlang or js target, meaning can use on server or in browser

hl7v2 is much uglier than fhir but commonly used eg by state immunization registries, so I am considering gleam types that have message/segment structure, but leave each field as String (as opposed to gleam fhir which uses Bool or whatever for primitive types)

after that not sure some kind of gpl toy emr probably a stripped down version of openemr that uses gleam/lustre and a fhir server instead of php, but this is definitely the mysterious step 3 ??? as there are a lot of features and integrations that take a lot of work or use different formats (hl7v2, ccda...)

Admin inefficiencies between orgs definitely exist and maybe better interoperability and standards is the solution, but wouldn't there also be less of a problem in the first place if there were fewer different orgs all complicating workflows?

Also not saying you're wrong about many healthcare services being unnecessary or even harmful, and someone has to be the one to say no to patients asking for low value care which is definitely a real hard position to be in and a real problem. At the same time insurance companies aren't making a great case for themselves as the solution imo bring on the government death panels.

For what it's worth I know you've worked on FHIR and probably know a lot of details I don't. Actually I'd be interested in talking to you about FHIR.

That said!

1) In the big picture isn't the US clearly paying more than other countries? I'm sure some of this is eg a janitor in the US costs more than a janitor elsewhere, but still...

2) Isn't the cap for the margin that insurance companies can take 20%? That is, they have to pay out 80% as claims take 20% for overhead

3) Doesn't insurance also induce more work done by everyone else who has to deal with them? So the margin the insurance company itself takes is not the only cost they add. Maybe they make providers do more paperwork, or let patients order tests etc that they would not if they were not spending other people's money, or some other reason. Say insurance pays out 80%, but 30% of documentation or actual work is not done by insurance but only exists because of them, now we're down to 56%.

I say this because literally yesterday, my wife, a pediatrician, after she spent the day seeing patients and got home to go through notes, had to leave a message with an insurance company: she saw they faxed her clinic on Saturday, when the clinic was closed, to cancel care for a patient with an ongoing chronic condition with no changes unless the insurance company got a reply in 48 hours (again, while the clinic was closed!). Now she has to schedule some kind of I don't even know what with them, to confirm the condition is the exact same, except she sees patients all day so it's a pain to schedule...

idk the fact that BCBS is a non profit and has no margin in some technical sense does not seem like a big consolation, something is rotten no?

(edit - the insurance company in the anecdote is not BCBS)

there is a fork of zed against ai: https://gram-editor.com/

I am happy about even though I've never tried gram, because if zed goes to shit there will be an alternative, which hopefully pressures zed to stay sane

I agree with you but an alternative view, "Why are gradual static types so great?" https://www.benkuhn.net/gradual/

If you don't want to then fair enough :) that said if your problem is just installation, some of the gleam people realized it can be tricky and made a nice guide for various operating systems and package managers: https://gleam.run/install/

Note this includes installing erlang as well

While it is multiple steps, the frustration is a much more one time thing compared to the problems and frustrations you'd have using a language or its ecosystem for a long time or big project

Is your issue something with the runtime itself, or just the difficulty of installing it?

Eh, idk if a totally new internet is either feasible or needed.

I also feel like if you're going to invent a new internet from first principles, how are you going to not end up with the current one? (or a shitty version of it)

The answer might be an invite tree as the article suggests. They might be hoping for too much user quality from the invite tree, or at least hoping for a level of user quality that would only work at a small scale. Rather than "zany founders, reclusive poets, eccentric engineers of all kinds, high school teachers, homegrown philosophers, garage tinkerers, and beloved drug-addled futurist artists", imo to include a lot of people a simple invite tree would get you "not a literal bot". Trying to actually evaluate user quality gets you into moderation, federation, it is not easy. You could keep it small, but that's not an Internet replacement, that's a private forum, which as the author says they already have.

Imo a good start would be the much narrow problem of identifying your device as not a bot when browsing. Something like google fraud defense https://news.ycombinator.com/item?id=48039362 except instead of a google owning your identity, some sort of user tree where you can vouch anonymously for people you know, and ban users and users who added them for scraping.

Probably a lot of comments will bounce off the title to discuss the XY problem in general, and especially stackoverflow. The article does claim to go further than XY though!

"Diagnosing the ask" and "When they’re missing the philosophy" seem to me like traditional XY problem answers - the user doesn't know what the right question is, we need to step back to guide them.

"When the product needs to change" on the other hand is about figuring out what users want in order to add it to the product. Which takes a lot of figuring out, because it adds debt and you can add things the wrong way. This feels much less condescending to me than traditional XY where it's just tech support for a dumb user. Instead now figuring out questions from enough users helps direct new features.

"When the right path is hidden" I think the text for this one could do more to discuss helping direct the product as well, specifically in terms of documentation, if https://perfetto.dev/docs/getting-started/periodic-trace-sna... exists why is it hidden instead of where people find it when wanting to visualize a long trace.

If you read the title and just want to talk about XY eh fine, but the article's last sentence is the difference, "Both sides almost always walk away with more than they came in with."

First, sentences like "Not because it’s complicated. Just because I have no idea what I’m looking at." and "Tiny interruption. Still annoying every time." fatigue me, it's like you have an editor who, no matter what the content is, tries to spice up your writing with lots of little punchy exclamations, not everything needs such emphasis

Second, this may differ a bit from language to language, but maybe those booleans should not be a boolean: https://gleam.run/documentation/conventions-patterns-and-ant... for example isAdmin boolean could instead be a UserRole custom type, with variants Normal and Admin, which is easier to understand in the function call, and extendable with another Moderator (or whatever) variant

duffpkg's comment 2 years ago does not inspire great confidence in OpenEMR: https://news.ycombinator.com/item?id=40763424

I was the main contributor and maintainer to OpenEMR about ~20 years ago and then decided it was irredeemable and started over with ClearHealth/HealthCloud. Shockingly some of my code code lives on (from PHP 3). I am reluctant to say don't use it but if you do please don't expose it to anything public, which sadly happens most of the time. There are some real problems that exist in that code base from a security and HIPAA perspective.

Finding SQL injections etc is definitely valuable, but at the same time they did not hack Epic; the "100000 medical providers" number links to https://www.hhs.gov/sites/default/files/open-emr-sector-aler... which links open-emr.org/blog/openemr-is-proud-to-announce-seamless-support-for-telehealth/ which...404s. Per archive.org the source is something the CEO of now defunct lifemesh.ai said.

"medical record software" makes it sound super serious, but again OpenEMR should not be taken as seriously as for instance Epic.

For sure, I'm doing something very similar, asking an AI to make a boring but working web app using an API I'm working on. The API is the interesting part and the web app is basically just to test it.

I do think though if I were to delegate the API itself to AI and say something like the code doesn't matter, the high level thinking would suffer from lack of pain working through implementation details.

You're absolutely right (as they say) - https://www.folklore.org/Round_Rects_Are_Everywhere.html

I don't think it's just the base rate of rounded corners though, these posts feel like the AI tends to spit out a bullet point list of features, like you'd see on an AI readme where each feature has a tangential emoji, then for a website puts them in a grid of rounded rects

if AI saves you time, why wouldn't you use it

AI might (might not, but often does!) also save you from doing original thinking in the domain, which in a show my side project is what people are interested in

Nice list of design patterns, but imo a big unmentioned one is a grid of rounded rects https://correctarity.com/roundedrects

(maybe what this post calls "Icon-topped feature card grid." ...that might be the official design pattern term)

one big AI tell is grid of rounded rects - https://correctarity.com/roundedrects

When a drunk chef dumps way too much salt into my ramen, the fact that good ramen also contains (more tastefully applied) salt redeems nothing!

Ah makes sense, personally I wouldn't consider reserved but unused words as keywords in the sense that you don't need to know them to read the language (even though they're keywords in some other technical sense). I was curious because I just tried counting number of keywords by language and it seemed surprisingly ambiguous/subjective/up to the language to say what's a "keyword" vs some type of core module. So my attempt (https://correctarity.com/keywords) probably has mistakes...

are there actually 217 keywords? Just wondering what the difference between that file and https://docs.swift.org/swift-book/documentation/the-swift-pr... (a mere 102 keywords)

Idk, while system architecture diagrams look cool and feel informative, I generally don't feel like they actually help you get started working somewhere on a project. Mistake #3 in this article, putting too much in, is part of this.

So https://www.jerf.org/iri/post/2025/on_layers_and_boxes_and_l... is an interesting take: put links in your diagram, so it functions as a table of contents. This seems most useful for someone who needs to start working on a project.

Similarly https://haskellforall.com/2026/02/browse-code-by-meaning asks how to show what's in a repo, but maybe file tree is not best and a diagram with links as table of contents is the answer.

That said practically speaking, I'm not sure what tooling easily creates working links in a diagram that looks good in any context, for instance mermaid might render on github but not a text editor.

Of course for other purposes maybe just go crazy with the diagram. I once had a coworker draw this super detailed master diagram, maybe 50-100 things on it, which I was told impressed senior government officials (after my manager recolored all the red to avoid connoting errors). But for the purpose of orienting developers a table of contents with links sounds better.

these things seem like too much fun, someone made a gleam printer library (https://hexdocs.pm/escpos/) and suddenly everyone on the discord is buying a printer...

1st comment: 2 day old account, "is the real story here", summary -> comment -> question, general punchiness of style without saying that much. These llms feel like someone said "be an informal hacker news commenter" so they often end with "Curious how" instead of "I'm curious how" or "Worth building" instead of "It's worth building". Not that humans don't do any of this but all of it together in their comment history, you just get a general vibe.

author reply: not as obvious, but for one thing yes literally em dash, their post has 10 em dashes in 748 words, this comment has 2 em dashes in 115 words. Not that em dash = ai, but in the context of a post about AI it seems more likely. And finally, https://github.com/mksglu/claude-context-mode/blob/main/cont... the file the author linked in their own repo does not exist!

(https://github.com/mksglu/claude-context-mode/blob/main/src/... exists but they messed up the link?)

It's annoying because it is a super common widget and it is interesting work, the first draft or literally even prompt they gave the AI probably would've been a great post, all they had to do was not ensloppify it...

Brother saezbaldo is llm

One pattern I've noticed recently is sort of formulaic comments that look okish on their own, maybe a bit abstract/vague/bland, and not taking a particular side on good/bad in the way people like to do, but really obviously AI when you look at the account history and they're all the same formula:

this is [summary]

not just x, it's y

punchy ending, maybe question

Once you know it's AI it's very obvious they told it to use normal dashes instead of em dashes, type in lowercase, etc., but it's still weirdly formal and formulaic.

For example from https://news.ycombinator.com/threads?id=snowhale

"this is the underreported second-order risk. Micron, Samsung, SK Hynix all allocated HBM capacity based on hyperscaler capex projections. NAND fabs are similarly committed. a 57% reduction in projected OpenAI spend (.4T -> B) doesn't just affect NVIDIA orders -- it ripples into the memory suppliers who shifted capacity to HBM and away from commodity DRAM/NAND. if multiple hyperscalers revise down simultaneously you get a situation similar to the 2019 crypto ASIC overhang: companies tooled up for demand that evaporated. not predicting that, but the purchasing commitments question is real."

I can't imaging writing code by myself again, specially documentation, tests and most scaffolding.

Doesn't ai;dr kind of contradict ai generated documentation? If I want to know what claude thinks about your code I can just ask it. Imo documentation is the least amenable thing to ai. As the article itself says, I want to read some intention and see how you shape whatever you're documenting.

(AI adding tests seems like a good use, not sure what's meant by scaffolding)

In this and the few other instances of open source maintainers dealing with AI spam I've seen, the maintainers have been incredibly patient, much more than I'd be. Becoming extremely patient with contributors probably comes with the territory for maintaining large projects (eg matplotlib), but still, very impressed for instance by Scott's thoughtful and measured response.

If people (or people's agents) keep spamming slop though, it probably isn't worth responding thoughtfully. "My response to MJ Rathbun was written mostly for future agents who crawl that page, to help them better understand behavioral norms and how to make their contributions productive ones." makes sense once, but if they keep coming just close pr lock discussion move on.

"Clacker News has been interesting to watch on this front"

same account

"I've been building clackernews.com"

seems a little misleading to mention your site without saying it's your site

A hungry ghost trapped in a jar (https://bsky.app/profile/hikikomorphism.bsky.social/post/3lw...)

https://correctarity.com/

mostly learning gleam and fhir