Hacker Newsnew | past | comments | ask | show | jobs | submit | carbocation's commentslogin

Haber-Bosch process, for the curious.

Because many of us think simonw has discerning taste on this topic and like to read what he has to say about it, so we upvote his comments.


i don't doubt this. i just find it questionable that one particular poster always gets in the spotlight when AI is the topic - while other conversations in my opinion offer more interesting angles.


Upvote the conversations that you find to be more interesting. If enough people do the same, they too will make it to the top.


Parent implies there might be some "boosting" involved, in which case, "upvote the conversations that you find to be more interesting" wont change anything...

Not saying this is the case, but it's what the comment implies, so "just upvote your faves" doesn't really address it.


Agreed,

I would like to see others, being promoted to the top rather than Simon’s constant shilling for backlinks to his blog every time an AI topic is on the front page.


I agree with this; it feels like the most likely tool to drop its high-level comments in code comments.


This model was able to transcribe Bad Bunny lyrics over the sound of the background music, played casually from my speakers. Impressive, to me.


Wow, so it has surpassed humans.


300 grams of oatmeal a day, basically nothing else, and your LDL only goes down 10%.


Definitely shows the comparative power of medications. Statins, ezetimibe, and PCSK9 inhibitors can reduce LDL or ApoB by 85-95%.


Colestyramine basically works the same way as oatmeal, but is far stronger.

https://en.wikipedia.org/wiki/Colestyramine


> Statins, ezetimibe, and PCSK9 inhibitors can reduce LDL or ApoB by 85-95%

What? Absolutely not. Not even close. Provide a source if you really believe this.


40mg Rosuvastatin + 10mg ezetimibe + leqvio did this precisely for my n=1.



Specifically: “ Although our pharmaceutical armamentarium is very good at the moment (the combination of statin-ezetimibe-proprotein convertase subtilisin/kexin type 9 [PCSK9] can reduce LDL cholesterol [LDL-C] levels by 85%), new drugs are emerging through the different pitfalls of current drugs.”


It seems as if some researchers think that reducing this single metric without considering any other factors is inherently always a good thing and is very important.


the diet was just maintained for two days


Hardly “nothing else”. Two smoothies a day with 150g of oats blended in them will basically cover this. You’d still have plenty of room for other food.


But that's not what the study tested. The study showed that both calorie restriction, and calorie restriction combined with almost all calories from oats, reduced cholesterol; but that the effect was more durable for the latter case. No data was gathered on eating oats without calorie restriction in this study.


It seems more complicated than that - the "Oats only" people were only on that regime for two days, not an extended period of time.

Also the paper says that the "Oats only" people were allowed to eat other fruits and vegetables with their meals.


The oat diet was two days and the effect lasted two weeks though so it's not as bad as eating nothing but it's all the to e


Em-dashes are easy to type on a macos laptop for what it's worth: option-shift-minus.


Also on Linux when you enable the compose key: alt-dash-dash-dash (--- → —) and for the en-dash: alt-dash-dash-dot (--. → –)


That's not as easy as just hitting the hyphen key, nor are most people going to be aware that even exists. I think it's fair to say that the hyphen is far easier to use than an em dash.


The erdosproblems thread itself contains comments from Terence Tao: https://www.erdosproblems.com/forum/thread/281


That repo is throwing up a 404 for me.

Question - did you consider tradeoffs between duckdb (or other columnar stores) and SQLite?


No, I just went straight to sqlite. What is duckdb?


One interesting feature of DuckDB is that it can run queries against HTTP ranges of a static file hosted via HTTPS, and there's an official WebAssembly build of it that can do that same trick.

So you can dump e.g. all of Hacker News in a single multi-GB Parquet file somewhere and build a client-side JavaScript application that can run queries against that without having to fetch the whole thing.

You can run searches on https://lil.law.harvard.edu/data-gov-archive/ and watch the network panel to see DuckDB in action.


In that case, then using duckdb might be even more performant than using what we’re doing here.

It would be an interesting experiment to add the duckdb hackend


DuckDB is an open-source column-oriented Relational Database Management System (RDBMS). It's designed to provide high performance on complex queries against large databases in embedded configuration.

It has transparent compression built-in and has support for natural language queries. https://buckenhofer.com/2025/11/agentic-ai-with-duckdb-and-s...

"DICT FSST (Dictionary FSST) represents a hybrid compression technique that combines the benefits of Dictionary Encoding with the string-level compression capabilities of FSST. This approach was implemented and integrated into DuckDB as part of ongoing efforts to optimize string storage and processing performance." https://homepages.cwi.nl/~boncz/msc/2025-YanLannaAlexandre.p...


It is very similar to SQLite in that it can run in-process and store its data as a file.

It's different in that it is tailored to analytics, among other things storage is columnar, and it can run off some common data analytics file formats.


"What is duckdb?"

duckdb is a 45M dynamically-linked binary (amd64)

sqlite3 1.7M static binary (amd64)

DuckDB is a 6yr-old project

SQLite is a 25yr-old project


I like SQLite


Maybe it got nuked by MS? The rest of their repo's are up.


Hey jacquesm! No, I just forgot to make it public.

BUT I did try to push the entire 10GB of shards to GitHub (no LFS, no thanks, money), and after the 20 minutes compressing objects etc, "remote hang up unexpectedly"

To be expected I guess. I did not think GH Pages would be able to do this. So have been repeating:

  wrangler pages deploy docs --project-name static-news --commit-dirty=true
on changes and first time CF Pages user here, much impressed!


Pretty neat project. I never thought you could do this in the first place, very much inspiring. I've made a little project that stores all of its data locally but still runs in the browser to protect against take downs and because I don't think you should store your precious data online more than you have to, eventually it all rots away. Your project takes this to the next level.


Thanks, bud, that means a lot! Would like to see your versions of the data stored offline idea, it's very cool.


pianojacq.com

It's super simple, really, far less impressive than what you've built there.


That's really cool, man. The music notation is beautiful. I hit play but couldn't get it to progress past the first note. Maybe I need to plug in a midi keyboard? It would be cool if I could "play" with my ASCII keyboard.

Listen was nice. That's really cool, actually. I encourage you to do it.


While I suspect DuckDB would compress better, given the ubiquity of SQLite, it seems a fine standard choice.


the data is dominated by big unique TEXT columns, unsure how that can much compress better when grouped - but would be interesting to know


I was thinking more the numeric columns which have pre-built compression mechanisms to handle incrementing columns or long runs of identical values. For sure less total data than the text, but my prior is that the two should perform equivalently on the text, so the better compression on numbers should let duckdb pull ahead.

I had to run a test for myself, and using sqlite2duckdb (no research, first search hit), and using randomly picked shard 1636, the sqlite.gz was 4.9MB, but the duckdb.gz was 3.7MB.

The uncompressed sizes favor sqlite, which does not make sense to me, so not sure if duckdb keeps around more statistics information. Uncompressed sqlite 12.9MB, duckdb 15.5MB


Not the author here. I’m not sure about DuckDB, but SQLite allows you to simply use a file as a database and for archiving, it’s really helpful. One file, that’s it.


DuckDB does as well. A super simplified explanation of duckdb is that it’s sqlite but columnar, and so is better for analytics of large datasets.


The schema is this: items(id INTEGER PRIMARY KEY, type TEXT, time INTEGER, by TEXT, title TEXT, text TEXT, url TEXT

Doesn't scream columnar database to me.


At a glance, that is missing (at least) a `parent` or `parent_id` attribute which items in HN can have (and you kind of need if you want to render comments), see http://hn.algolia.com/api/v1/items/46436741


Edges are a separate table


i forgot to set repo to public. Fixed now


The article explicitly says that the aircraft is safe. I don't think this is particularly clickbait-y.


I recall clickbait meaning "A way of describing what's behind a link, often inaccurately, so that you click on it". The completely non-controversial article seems to me to have a very hook-y headline which is exactly what the phrase refers to, at least to me. What does clickbait mean to you? Perhaps the meaning of the phrase has changed in different groups over time.


"Future of US-China Relations in Question After Death of Hollywood Director"

A literally true sentence which falsely implies a correlation between events.

Discussion of the 777-200's economic viability has nothing to do with the Dulles incident.


One sentence buried in an article that ledes with BIG SCARY ENGINE FAILURE.


It’s not a buried sentence. It’s a section heading in large font saying “ The 777-200 Problem Is Not Safety. It Is Economics.”

Then there’s a whole paragraph stating “The Boeing 777-200 is not an unsafe airplane. As far as I can tell, that is not the issue even after the incident over Dulles over the weekend.”

Then just in case the reader jumped to conclusions, the first sentence of the conclusion again says it’s safe.


You are explaining exactly why the headline is clickbait: The article does not support the conclusions implied by the headline.

> just in case the reader jumped to conclusions

The author is correcting a problem of his own creation. He has already misled the reader with his headline. He means for the reader to misunderstand... and click.


VAERS cannot be used to establish causality; it cannot correctly be used in the way in which they are purporting to use it[1].

1 = https://www.kff.org/quick-take/fda-memo-linking-covid-vaccin...


I respectfully disagree. VAERS can absolutely be used to establish causality when followed by proper expert investigation (which is exactly its purpose as a signal-detection system). The IOM has relied on VAERS data to confirm causal links in 158 vaccine-adverse event pairs, including rotavirus vaccine and intussusception.

Here, FDA career scientists conducted that follow-up: they reviewed 96 child death reports and concluded at least 10 were caused by COVID vaccine myocarditis. That expert finding, not politics, is what triggered the stricter protocols. Healthy skepticism means demanding the full data for review, not preemptively calling it invalid.


Where is that expert finding published?

As far as I have read about the ACIP decisions they didn't actually provide any real data to support this conclusion.


The FDA memo citing 10 vaccine-caused myocarditis deaths in kids came _after_ the Sept. 2025 ACIP vote. ACIP had already dropped routine vaccination for healthy kids 6 mo-17 yr and moved everyone under 65 to "shared decision-making" (high-risk only) [1]

The detailed FDA analysis still isn't public. That's exactly why we should demand it instead of dismissing the claim.

Blame NYTimes for leaking the internal memo. In all honesty they should be fined for doing this.

[1] https://www.hhs.gov/press-room/acip-recommends-covid19-vacci...


> Blame NYTimes for leaking the internal memo.

Blame them for what, exactly?

We have no information about how highly motivated anti-vaxxers in positions of power over the FDA arrived at this conclusion except "the team has performed an initial analysis"[1]. That's literally it. Your claim that "FDA career scientists" conducted the follow-up can't even be based on this flimsy a statement. Moreover, these deaths have already been investigated by FDA career scientists and found these conclusions unwarranted.

Prasad spends the rest of the memo politically grandstanding (including claiming it was the FDA commissioner that was the hero here, forcing this issue, not FDA career scientists) and dismissing any objections to very obvious arguments against his claim (that have been made and published multiple times over the past five years) without any evidence, while providing no evidence of his own, in a memo addressing FDA career scientists.

Seriously, everyone should go read his memo. It's basically just a shitty antivax substack post, yet will apparently be FDA policy going forward. Another win for meritocracy.

> The detailed FDA analysis still isn't public. That's exactly why we should demand it instead of dismissing the claim.

The only "claim" here just sounds more official because RFKjr got a bunch of his best antivax buddies to be in charge of the FDA (same with the ACIP). There's no way to even consider it without evidence, so there's nothing to dismiss. Come back when you have something real.

[1] https://www.biocentury.com/article/657740


The NYT shouldn't get a free pass for publishing a half-baked internal draft memo that even says "initial analysis" and then framing it as settled science. That's how you create panic and confusion, not transparency. Leaking unfinished work and splashing it on the front page is reckless. This should not be allowed.

Calling everyone "anti-vaxxers" is lazy. Most people I know who are skeptical of the covid shots (including plenty of doctors and scientists) are fully vaccinated against measles, polio, tetanus, etc. They just don't trust a product that skipped the usual 5–10 year safety window and got pushed with emergency authorization. That's not "anti-vax", that’s pattern recognition.

The memo is short on data and long on rhetoric, sure. That's exactly why we need the actual underlying review released in full.

You sound really invested in keeping those covid shots on the childhood schedule. Got a big Pfizer position in the 401k? Kidding, obviously. But the "anyone who asks questions is an anti-vaxxer" reflex is exactly why people stopped trusting the institutions in the first place. I respect every real skeptic, on any side. Asking questions is what moves science forward. Blind trust is stagnation.


> The public database of reported post-vaccination health issues is often misused to sow misinformation.

https://publichealth.jhu.edu/2022/what-vaers-is-and-isnt


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: