Datasets:

typedef-ai
/

hacker-news-dataset

Modalities:

Tabular

Text

Formats:

Size:

Libraries:

Dataset card Data Studio Files Files and versions

xet

Community

Dataset Viewer

Auto-converted to Parquet Duplicate

Split (10)

2025_comments · 2.54M rows

id int64	type string	by string	time int64	title string	text string	url string	score int32	parent int64	descendants int32	dead bool	deleted bool	kids list	parts list	poll int64	fetched_at int64
42,562,835	comment	MegaDeKay	1,735,689,820	null	Give "The Ultimate History of Video Games, Volume 2" a read. Sega America did well marketing-wise but Sega Japan was a shitshow and really dragged them down. Sega could have done much better with the Dreamcast than they did given the cheaper price point, one year head start, and solid lineup of games.	null	null	42,562,594	null	null	null	[ 42563815, 42563862 ]	null	null	1,758,903,346
42,562,839	comment	sheepscreek	1,735,689,857	null	Why is this trending on HN? This is how most execs are compensated. Doesn't mean they are selling out. I wouldn’t read too much into it.	null	null	42,560,897	null	null	null	[ 42562854 ]	null	null	1,758,903,346
42,562,850	comment	lucubratory	1,735,690,127	null	I think basically everyone should support a carbon tax. It's a really obvious solution that is both environmentally friendly and should be acceptable to free market fanatics because it is explicitly and only taxing a negative externality on the public - it's hard to imagine a more justified tax.<p>Combined with the increased cost effectiveness of renewables & batteries, & the new build-out of nuclear, it could plausibly speed up the clean energy transition, rather than just disincentivising building out more polluting power plants.<p>There are two main options for what to do with revenue from a carbon tax. The one that makes the most macroeconomic sense is to use those proceeds to fund subsidies for clean energy roll outs & grid adaptation. You are directly taxing the polluting power grid to fund the construction of a non-polluting power grid. As CO2 emitting industry (and thus carbon tax revenue) declines, we have less required spend on clean energy roll out, so the tax would balance nicely. The downside would be that a carbon tax would increase cost of living and this does nothing about that.<p>The other option is a disbursement. Give everyone in society a payment directly from the proceeds of the carbon tax. This would offset the regressive aspects of a carbon tax (because that tax would increase consumer costs), and would also act as a sort of auto-stimulus to stop the economy from turning down due to consumption costs increasing. The downside of this is that the clean energy transition happens slower than the above, and that there may be political instability & perverse incentives as people maybe come to rely on this payment that has to go away over the next few decades.<p>They're both good options. I don't know which is better and I think that's likely something individual countries will probably choose based on their situation. But we do need some sort of way to make those emitting CO2 pay for its negative externalities.	null	null	42,561,273	null	null	null	[ 42622999 ]	null	null	1,758,903,346
42,562,970	comment	mikewarot	1,735,691,571	null	Metamine reappears from the void, and wins over the hearts and minds of programmers everywhere with it's clever and powerful combination of imperative and declarative techniques.	null	null	42,558,944	null	null	null	null	null	null	1,758,903,346
42,563,038	comment	Dwedit	1,735,692,359	null	Make it 3D and you'll be really really lost.	null	null	42,529,894	null	null	null	null	null	null	1,758,903,346
42,563,066	comment	huijzer	1,735,692,815	null	> it's due to them simply not recognizing that they could actually take their time and enjoy life<p>Ah sure he went on to work extremely hard and build a whole foundry business just because he didn’t recognize that he could actually take his time and enjoy life? I think you’re sort of right but also wrong. Telling an extremely ambitious person to “just enjoy life” is, I think, like telling an autistic person to “just be social”. It’s not as simple as that.	null	null	42,561,244	null	null	null	[ 42565250 ]	null	null	1,758,903,346
42,563,083	comment	darioush	1,735,693,079	null	Sure it does since luxury tax can be 10%-25% of property value. Let's see who wants to hold on to property then. And if they do they are paying their dues to society.<p>When they are "removed" and owned by more people, presumably this is a more efficient allocation of the same amount of houses to more people, which is a solution for the housing problem.	null	null	42,552,448	null	null	null	[ 42563184 ]	null	null	1,758,903,346
42,563,131	comment	ricardobeat	1,735,693,632	null	Models like Llama 3 are trained on <i>sixteen thousand</i> GPUs, OpenAI probably 25k-100k GPUs. This is the kind of scale the sanctions make a lot harder to achieve.	null	null	42,563,036	null	null	null	[ 42563164, 42563918 ]	null	null	1,758,903,346
42,563,161	comment	wolfgangK	1,735,694,034	null	Amazingly thorough ! I love how the author leaves no stone unturned. I had no idea you could do the kind of low level efficiency shaving in Rust. I wonder how a C++ implementation with <a href="https://github.com/jfalcou/eve">https://github.com/jfalcou/eve</a> would compare.	null	null	42,562,847	null	null	null	[ 42563187, 42566194, 42565575, 42563254 ]	null	null	1,758,903,346
42,563,162	comment	throwitaway1123	1,735,694,055	null	> They don't work offline<p>This isn't true. Offline functionality is the raison d'être for Service Workers. You can run an entire HTTP request router on the client to respond to requests while offline: <a href="https://hono.dev/docs/getting-started/service-worker" rel="nofollow">https://hono.dev/docs/getting-started/service-worker</a>	null	null	42,557,673	null	null	null	[ 42563289 ]	null	null	1,758,903,346
42,563,172	comment	wolfgangK	1,735,694,157	null	I don't think that Python would be the right language for such low-level performance maxxing endeavor. I would have picked C++ but t was eye opening for me to see how rust enabled such low level optimization, so I'm grateful for the choice.	null	null	42,563,115	null	null	null	[ 42563219, 42563202 ]	null	null	1,758,903,346
42,563,178	comment	lolinder	1,735,694,262	null	> I just communicated why the market leak theory is both more intuitive, and more probable.<p>No, you didn't, you stated that it was.<p>The rest of your post makes sense as an explanation. Maybe lead with that next time instead of condescendingly telling people that they're politically motivated, stupid, or whatever else you meant to imply by calling it a conspiracy theory.<p>COVID-19, at least in the US, has been an enormous failure in science communication, and being condescending towards those who already feel alienated by the terrible communication isn't going to help.	null	null	42,563,060	null	null	null	[ 42563211 ]	null	null	1,758,903,346
42,563,190	comment	bwag	1,735,694,478	null	Same. I've been a lurker since 2008. Been hitting this site practically every day since. I've learned so much in the process. Thanks for everything hn, happy new year!	null	null	42,562,750	null	null	null	[ 42564794, 42563929, 42572219 ]	null	null	1,758,903,346
42,563,192	comment	HardikVala	1,735,694,501	null	An interesting coincidence is that the first fabless semiconductor company, Chips and Technologies, was also founded in 1985, by Dado Banatao and Gordon Campbell. They didn't partner with TSMC but contracted with companies like Hitachi that had excess fab capacity to manufacture their semiconductors. Wild. They eventually sold the company to Intel, which obviously didn't appreciate the insight of de-verticalization in the semiconductor supply chain.	null	null	42,559,052	null	null	null	null	null	null	1,758,903,346
42,563,193	comment	int_19h	1,735,694,506	null	It's usually the other way around - first countries introduce laws that require ID to buy a cell phone ("because criminals"), and then the phone number starts getting used as a de facto identity.	null	null	42,559,496	null	null	null	[ 42593093 ]	null	null	1,758,903,346
42,563,292	comment	vouaobrasil	1,735,695,689	null	I've used both and darktable is far superior. Most of the edits I do would be difficult in Ansel...	null	null	42,558,948	null	null	null	null	null	null	1,758,903,346
42,563,324	comment	codetrotter	1,735,696,089	null	> Are there emulators for all of these systems?<p>RetroArch is a great place to start. It supports a ton of different “emulator cores” and acts as a frontend for emulators for different systems.<p><a href="https://en.m.wikipedia.org/wiki/RetroArch" rel="nofollow">https://en.m.wikipedia.org/wiki/RetroArch</a><p>Install it and give it a go.	null	null	42,563,255	null	null	null	null	null	null	1,758,903,346
42,563,344	comment	simonw	1,735,696,245	null	Found that comment here, about to reply: <a href="https://news.ycombinator.com/item?id=42562394">https://news.ycombinator.com/item?id=42562394</a>	null	null	42,563,264	null	null	null	null	null	null	1,758,903,346
42,563,397	comment	eigenspace	1,735,696,870	null	I think the little blurb at the beginning is there because not everyone watching it knows English.<p>Because the humour is so physical though, that blurb is mostly all you really need to enjoy it even if you only know German.	null	null	42,562,101	null	null	null	[ 42572623 ]	null	null	1,758,903,346
42,563,403	comment	markus_zhang	1,735,696,917	null	I used to work as a stock broker and it helped tremendously with my soft skills.	null	null	42,557,947	null	null	null	null	null	null	1,758,903,346
42,563,451	comment	cheinic830405	1,735,697,495	null	> Average age of first-time homebuyers is 38, an all-time high.<p>And why is this a problem?<p>What is wrong with renting for life, exactly? I’m seeing more and more people own rental properties, but do not own their primary residence by choice.	null	null	42,562,962	null	null	null	[ 42569272, 42569139, 42594752, 42575512 ]	null	null	1,758,903,346
42,563,458	comment	leptoniscool	1,735,697,639	null	Happy New Year from the West coast!	null	null	42,562,750	null	null	null	[ 42563580 ]	null	null	1,758,903,346
42,563,491	comment	Kiro	1,735,698,254	null	Hilarious that you're trying to gaslight us into "recognizing" your own incorrect assumptions as facts. You've lost all credibility.	null	null	42,561,447	null	null	null	null	null	null	1,758,903,346
42,563,537	comment	int_19h	1,735,699,144	null	There is an old joke that AI researchers came up with several decades ago: "quality of results is inversely proportional to the number of linguists involved".<p>This has been tried repeatedly many times before, and so far there has been no indication of a breakthrough.<p>The fundamental problem is that we don't know the actual rules. We have some theories, but no coherent "unified theory of language" that actually works. Chomsky in particular is notorious for some very strongly held views that have been lacking supporting evidence for a while.<p>With LLMs, we're solving this problem by bruteforcing it, making the LLMs learn those universal structures by throwing a lot of data at a sufficiently large neural net.	null	null	42,562,336	null	null	null	null	null	null	1,758,903,346
42,563,562	comment	bovermyer	1,735,699,444	null	Happy New Year, whether you celebrated it in the past or the future.<p>Remember - the new year is not yet written. Amazing and unexpected things are sure to come.	null	null	42,562,750	null	null	null	null	null	null	1,758,903,346
42,563,604	comment	Spooky23	1,735,700,104	null	Apple and Google care about this because they handle more customer data and require more customer trust than most companies.<p>People were shitting a brick over a pretty minor change in photo and location processing at Apple. That’s because they don’t screw up like this.	null	null	42,563,183	null	null	null	[ 42578260 ]	null	null	1,758,903,346
42,563,626	comment	mvdtnz	1,735,700,589	null	> Every day by automation I feed in the hourly weather forecast my home ollama server and it builds me a nice readable concise weather report. It’s super cool!<p>You feed it a weather report and it responds with a weather report? How is that useful?	null	null	42,561,349	null	null	null	[ 42564202 ]	null	null	1,758,903,346
42,563,702	comment	flaterkk	1,735,701,703	null	Logged in just to say happy new year, also a lurker :-)	null	null	42,562,750	null	null	null	null	null	null	1,758,903,346
42,563,712	comment	varelse	1,735,701,841	null	[dead]	null	null	42,559,744	null	true	null	null	null	null	1,758,903,346
42,563,729	comment	dcre	1,735,702,036	null	Boy does this author know how to segment me out of his readership.	null	null	42,562,513	null	null	null	null	null	null	1,758,903,346
42,563,733	comment	viewtransform	1,735,702,081	null	<a href="https://archive.vn/UxLXt" rel="nofollow">https://archive.vn/UxLXt</a>	null	null	42,563,730	null	null	null	null	null	null	1,758,903,346
42,563,765	comment	ovelv	1,735,702,385	null	[dead]	null	null	42,557,586	null	true	null	null	null	null	1,758,903,346
42,563,767	comment	WWLink	1,735,702,419	null	> Blue Origin has an awfully long way to go if they want to catch up to SpaceX<p>Maybe we should just give the whole space industry to SpaceX because obviously nobody can touch them. /s	null	null	42,563,521	null	null	null	[ 42563868 ]	null	null	1,758,903,346
42,563,790	comment	fatata123	1,735,702,740	null	[dead]	null	null	42,563,618	null	true	null	null	null	null	1,758,903,346
42,563,810	comment	slothtrop	1,735,703,036	null	lol "important". You're free to flock to some other platform that better caters to extremists.	null	null	42,489,945	null	null	null	null	null	null	1,758,903,346
42,563,817	comment	numpad0	1,735,703,192	null	Are Parrot drones insecure or deliberately built with zero security? I thought they're French made, and France tended to be more scared with technological resistance to government interventions than US or China.	null	null	42,532,014	null	null	null	[ 42564148 ]	null	null	1,758,903,346
42,563,818	comment	varenc	1,735,703,195	null	If Xi Jinping died suddenly, it’d cause serious instability. His centralization of power erased clear succession plans, so top CCP factions would likely fight for control. Markets would also freak out. Long-term? Either a return to pre-Xi more collective leadership or another strongman doubling down on Xi’s approach. Centralized systems with no backup plans are fragile.<p>It’s possible that with technology like absolute communication control and ubiquitous surveillance the chance of internal unrest or revolution is greatly reduced. And as long as the country is growing and the average citizen is getting richer they’re much less likely to get unruly. It’s like startups: growth solves all problems.	null	null	42,560,843	null	null	null	null	null	null	1,758,903,346
42,563,842	comment	llm_trw	1,735,703,711	null	That's just making the architecture work better.<p>I'm old enough to remember when everyone outside of a few weirdos thought that a single hidden layer was enough because you could show that type of neural network was a universal approximator.<p>The same thing is happening with the wide MoE models. They are easier to train and sound a lot smarter than the deep models, but fall on their faces when they need to figure out deep chains of reasoning.	null	null	42,562,812	null	null	null	null	null	null	1,758,903,346
42,563,844	comment	LegionMammal978	1,735,703,773	null	It's not really about asynchronous callbacks or their equivalents. (In this case, the thread running it is otherwise meant to be blocked in a safe state, so that there's none of the usual dangers of interrupting arbitrary code.) Instead, it's about any callbacks coming out of C code, even something as trivial as qsort(). If you pass a C library your C++ callback, and your callback runs back through it with an exception, then 9 times out of 10, the C library will leak some resources at best, or reach an unstable state at worst. C just doesn't have any portable 'try/finally' construct that can help deal with it.<p>So I'd say it's more about the basic expectations of a function called from C, which includes a million other trivial things like "don't write beyond the bounds of buffers you're given" and "don't clobber your caller's stack frame" and "don't spawn another thread just to write to output pointers after your function returns" (not that any of these is the issue here).	null	null	42,563,386	null	null	null	[ 42564321 ]	null	null	1,758,903,346
42,563,851	comment	varenc	1,735,703,905	null	> It pains me to see it, but they show more long-term thinking that many of the Western governments…<p>Absolutely agree, and it pains me as well. Besides long-term thinking, they can also just impose sweeping new rules to address certain problems in a way the West never could.<p>For example, with teen gaming addiction, they didn’t hesitate to just ban kids under 18 from gaming more than a few hours a week, crashing the value of certain gaming companies. In the West, we’d spend years debating, lobbying, and litigating over individual freedoms vs. public good, and likely end up with nothing meaningful. It comes with huge drawbacks, but their system allows them to take drastic action quickly, while we’re often paralyzed by process.	null	null	42,558,966	null	null	null	null	null	null	1,758,903,346
42,563,868	comment	foobarian	1,735,704,202	null	They seem perfectly good at doing that all on their own	null	null	42,563,767	null	null	null	null	null	null	1,758,903,346
42,563,892	comment	lxe	1,735,704,694	null	Been on Iosevka for ages. Nothing else comes close for me. Not even berkeley condensed	null	null	42,555,233	null	null	null	null	null	null	1,758,903,346
42,564,004	comment	alphan0n	1,735,706,563	null	Gentleman’s bet. If you can accurately predict the day of four of the next six months of commoncrawls crawl, I’ll donate $500 to the charity of your choice. Fail to, donate $100 to the charity of my choice.	null	null	42,556,491	null	null	null	[ 42564454 ]	null	null	1,758,903,346
42,564,033	comment	tooltalk	1,735,707,020	null	China is quite anti-competitive. BYD/CATL's success is a largely outcome of the CCP's protectionism past ~10 years.	null	null	42,556,690	null	null	null	null	null	null	1,758,903,346
42,564,053	comment	duskwuff	1,735,707,241	null	The conclusion that this "created knowledge gaps" seems particularly inapt. If American universities had not purchased these particular books, they would have most likely been lost just like the rest of their respective print runs. The "knowledge gap" would have been even worse, as <i>no one at all</i> would have had access to those books.	null	null	42,564,018	null	null	null	[ 42564091 ]	null	null	1,758,903,346
42,564,100	comment	thayne	1,735,708,062	null	That is easier said than done. In order to achieve that effectively every employee that has any relation to data needs to be constantly vigilant in keeping PII to a minimum, and properly secured.<p>It is often much easier to use an email address or a SSN when a randomly generated id, or even a hash of the original data would work fine.<p>I'm not saying that we shouldn't put more effort into reducing the amount of data kept, but it isn't as simple as just saying "collect less data".<p>And sometimes you can't avoid keeping PII.	null	null	42,563,699	null	null	null	null	null	null	1,758,903,346
42,564,104	comment	threeseed	1,735,708,147	null	Formula 1 car has over 300 sensors and generates 1.5Gb a lap.<p>So I would imagine this is generating hundreds of megabytes.<p>You are going to be limited by what you can transfer over radio.	null	null	42,563,750	null	null	null	null	null	null	1,758,903,346
42,564,140	comment	andirk	1,735,708,815	null	I use bitcoin as a currency occasionally. I don't often trade my stonks for goods and services either.	null	null	42,539,228	null	null	null	null	null	null	1,758,903,346
42,564,153	comment	wellthisisgreat	1,735,709,059	null	You guys are really great.<p>Happy New Year to all of us	null	null	42,562,750	null	null	null	null	null	null	1,758,903,346
42,564,160	comment	keepamovin	1,735,709,240	null	Probably fairly low because the future is still talking to us so there have to be at least some timelines extending from the present that make it	null	null	42,563,754	null	null	null	[ 42564712 ]	null	null	1,758,903,346
42,564,191	comment	OhMeadhbh	1,735,709,863	null	Starship is heavy lift / heavy expense. If you want to put 1 small-sat in orbit, you don't use Starship, you use a Rocket Lab Electron. It's MUCH cheaper. Or you wait until you find 600 other small sat users who want to launch into the same / similar orbit.<p>The talk I heard was that New Glenn was supposed to be BO's answer to Starship (or more likely Falcon Heavy) that could launch a bunch of Project Kuiper satellites into LEO so Amazon could compete with Starlink and feed Amazon's Ground Station as a Service offering.<p>If we are to believe the published numbers, New Glenn can lift 50 imperial tons to LEO, Starship Block III will hoist 200 tons and Falcon Heavy will lift 50-60 depending on how re-usable you want your launch to be.<p>I'm not a heavy lift sales-person, but I've been in the room when they discussed what they thought they could sell to govt / mil / commercial customers. So take this with a little bit of salt... Seems to me BO was targeting a slightly smaller launch vehicle than SpaceX was going with so they could decouple schedule with Amazon's Kuiper Project. You don't want to have that cool new rocket you developed dependent on a satellite constellation that gets delayed. So you have a rocket that would be easier to fill with a number of small to medium sized satellites to LEO or (fewer) to GEO.<p>And like other people on the thread have commented, it seems BO is a decade behind SpaceX, so... yeah... a big rocket that competes with Starship is pretty risky for BO.<p>And yes, I understand that BO is independent from Amazon, but from what I've seen this is just so they can execute on a schedule that isn't determined by AMZN's board of directors. They seem pretty closely related, just from talking with Kuiper, GSaaS and BO engineers.<p>I don't work for any of the above mentioned companies and have no insider information. YMMV. Just my guesses from watching some of the personalities involved for the last 30 years.	null	null	42,563,777	null	null	null	[ 42565177, 42564258, 42565191, 42568658, 42566665 ]	null	null	1,758,903,346
42,564,256	comment	krupan	1,735,711,431	null	Cruise is close to Wayno, but nobody is willing to invest in Cruise anymore	null	null	42,564,159	null	null	null	[ 42564698 ]	null	null	1,758,903,346
42,564,257	comment	teractiveodular	1,735,711,434	null	New Glenn's main customer will be Kuiper, Amazon's answer to Starlink.	null	null	42,563,812	null	null	null	[ 42564282, 42564441 ]	null	null	1,758,903,346
42,564,263	comment	thisislife2	1,735,711,554	null	Not really. But I understand your outrage as you only have partial information - the western media often has a tendency to whitewash their foreign policy to show themselves as "benevolent" to poor and developing nations, when in fact most of it seeks to take advantage of weaker nations.<p>In all this "tit for tat" convoluted policies, like the PL-480 for India, it is easy to miss the fact that the US, one of the <i>wealthiest nation</i> in the world even in the 50's and 60's, was asking India, then an impoverished nation, to pay for the wheat they supplied to it.<p>> A key event took place in 1943 that extended India’s food shortage until the early 1950s. In 1943, Prime Minister Winston Churchill ordered foodgrains meant for eastern parts of the country to be diverted to the British troops engaged in World War II. It resulted in a man-made famine in Bengal and the deaths of millions of people ... India signed a long-term Public Law (PL) 480 agreement to get food aid under Government agricultural trade development assistance, with the US in 1954. The agreement was signed a few more times before the US ended it in the late 1960s as Lal Bahadur Shastri and then Indira Gandhi were not willing to make policy changes, especially to allow privatisation of the industrial sector, in return for food aid. > > ... Food aid under PL480 not only drew flak but some critics pointed out that the wheat sent to India was “fit enough only for pigs!”. It was termed as a great insult that was to linger for long in people’s memory. Despite repeated pleas, the US gave only “a small quantity of rice” to India.<p>(<i>How India overcame food emergency, attained self-sufficiency (August 2022)</i> - <a href="https://web.archive.org/web/20240616192936/https://www.thehindubusinessline.com/india-at-75/overcoming-food-emergencies-through-imports-from-us-via-pl480/article65753881.ece" rel="nofollow">https://web.archive.org/web/20240616192936/https://www.thehi...</a> ).<p>> Now, the US wants India to export its wheat to other parts of the world amidst rising prices due to the Ukraine war whereas India is avoiding falling into the wheat trap all over again ... In 1954, then US President Dwight D. Eisenhower launched the Public Law 480 (PL-480) ... It was an initiative to offload all the excess wheat that the US had cultivated with price support. The US also started using it as a tool of diplomacy, because when another nation is dependent on you for its food, you can easily compel it into submission ... It was however realised by India in the mid-1960s that American food supplies came with strings attached and were being used as a means of diplomatic interference. At one point, the US even came close to rejecting wheat shipments to India, which threatened to push the country to the brink of famine. > > ... Six decades after the PL-480 controversy broke out between the two countries, India and the US find themselves at a crossroads over the staple commodity all over again. This time around, India imposed restrictions on its wheat exports ... India has restricted exports in the background of a persistent heatwave and lack of rain compelling officials to lower wheat output estimates ... However, the US, Germany and other G-7 countries have been critical of India’s stance. They are arguing that the restrictions on wheat exports could worsen the current food crisis in the world. Asking India to export wheat on the terms and conditions set by the Western world for its own advantage is more of a condescending attempt to tell the country that it somehow owes the sole responsibility to relax the soaring global prices. The richer nations, therefore, want India to do most of the heavy lifting ... This is in sharp contrast to Washington’s own track record. Remember, the US could have left India in a severe food crisis in the 1960s had New Delhi not managed to augment agricultural production within time. At that time, the US was motivated by its self-interest and India’s refusal to take sides between the USSR and the US. Today, India isn’t abandoning its moral responsibility, yet the US is sermonising it over wheat exports.<p>(<i>Ghost of PL-480 Returns as India Avoids the Wheat Trap All Over Again (May 2022)</i> - <a href="https://web.archive.org/web/20220525074511/https://www.news18.com/news/opinion/ghost-of-pl-480-returns-as-india-avoids-the-wheat-trap-all-over-again-5227951.html" rel="nofollow">https://web.archive.org/web/20220525074511/https://www.news1...</a> )<p>The only kind thing I have to say about the PL-480 or the "food for peace" program is that the American politicians taught us Indians a very valuable lesson - the need to become self-reliant to stave off exploitation. (Ironically, when it comes to people to people relationship, ordinary Indians and Americans really hit it off and work well together - for e.g. India's space, nuclear and even agricultural programs owes a lot to American scientists and thinkers. But this still hasn't translated to a close political relationships because Indian and American politicians just don't see eye to eye).	null	null	42,563,920	null	null	null	null	null	null	1,758,903,346
42,564,268	comment	saidinesh5	1,735,711,636	null	Just build one yourself... It's quite simple and a lot cheaper. All the components you need are open source. Open hardware and software. When something inevitably breaks you know what exactly to replace.<p>If you don't mind paying a little more and want a ready to fly / kit versions: pick any of these. I remember the Holybro x500 kit used to be very popular (review: <a href="https://m.youtube.com/watch?v=cTVtFYONHiY" rel="nofollow">https://m.youtube.com/watch?v=cTVtFYONHiY</a> )<p><a href="https://ardupilot.org/copter/docs/common-rtf.html" rel="nofollow">https://ardupilot.org/copter/docs/common-rtf.html</a><p>You can control them using mavlink / mavsdk etc... the python libraries are good enough.<p><a href="https://mavsdk.mavlink.io/main/en/index.html" rel="nofollow">https://mavsdk.mavlink.io/main/en/index.html</a>	null	null	42,563,940	null	null	null	null	null	null	1,758,903,346
42,564,281	comment	ethbr1	1,735,711,999	null	Having been involved in similar financial arrangements in software automation, years ago, it makes sense.<p>The end user usually doesn't have the expertise to even maintain the systems, nor does it make sense for them to do it in-house.<p>Charging per item of work (operating hour or thing processed) allows use of consultants but keeps incentives aligned between all parties (maximize uptime/productivity).	null	null	42,564,217	null	null	null	null	null	null	1,758,903,346
42,564,294	comment	tekknolagi	1,735,712,383	null	I got one 1.5 years ago and still enjoying. I hope it will be the laptop of Theseus of the rest of my life, but some unexpected development will probably happen that will make that unlikely.	null	null	42,564,216	null	null	null	null	null	null	1,758,903,346
42,564,368	comment	zizee	1,735,713,590	null	Ok, so your argument for Starship not being the best choice for all mission profiles is based solely on cost. If Starship was cheaper, then you'd agree that it serves the mission profiles?<p>An estimate of 2 billion per launch is laughable, and suggests you are not arguing in good faith. 100m is more accurate for a fully disposable launch, and SpaceX has demonstrated great progress on reusability of the booster, which will cut costs considerably.	null	null	42,564,230	null	null	null	[ 42564483, 42564409, 42564468 ]	null	null	1,758,903,346
42,564,589	comment	mikewarot	1,735,717,721	null	I plan to make a BitGrid chip, through Tiny Tapeout.<p>Once I have an actual chip, I can measure its performance/watt... and decide if it'll bring Petaflops to the masses, or not.	null	null	42,563,662	null	null	null	null	null	null	1,758,903,346
42,564,597	comment	OhMeadhbh	1,735,717,830	null	Don't show up alone.<p>I got recruited to organize a "Nurses for Single Payer" campaign (I was the web person, I'm not a Nurse.)<p>When it was two of us, our congress-critter wouldn't return our phone calls. When it was 10, they still didn't return our phone calls because we didn't know we were supposed to put our membership numbers on the web-site. When we were about 50 and realized we should put something like "representing over 50 nurses in our geographic area" we got a call back and when we had 100 real members, we got a sit-down meeting with a staffer. At about 125 members we got a meeting with our congress-critter.<p>YMMV, but if you can get at least 100 people to sign a petition or agree to be counted in some form of official membership list, that seems to be about where policy-makers start responding to you. I am willing to bet you have to have a lot more to truly "influence" policy, but it's a good start.	null	null	42,563,639	null	null	null	null	null	null	1,758,903,346
42,564,618	comment	olliej	1,735,718,141	null	doesn't include search?	null	null	42,563,858	null	null	null	null	null	null	1,758,903,346
42,564,703	comment	userbinator	1,735,719,627	null	I have no problem with them working on security in the service of the user. The problem is with them claiming to do that, but instead doing the opposite.	null	null	42,555,274	null	null	null	null	null	null	1,758,903,346
42,564,843	comment	simoncion	1,735,721,927	null	> Have you even skimmed the docs?<p>Yup.<p>Any committee that will leave entirely unpunished being lied to, directly, in person, by the fellow in charge of the biggest agency the committee is supposed to be overseeing isn't worth a damn.<p>Any committee that won't raise a big public stink about that agency's lawyers lying to the US Supreme Court? Same thing.<p>It's entirely possible to be both bipartisan and a Congressional committee but still be largely worthless to the public.<p>> Why do you take Snowden at his word yet ignore a bipartisan intel committee...<p>Snowden risked his ass (and is currently living in exile) to alert the public about long-running, major violations of Federal law. The most we get out of the absolute best member of that committee is "Man. The American public would be fucking incensed if they heard some of the things that we've been told in our chambers. Someone should really do something about this.".<p>And yeah, I'm aware that that report was written by a scratch committee assembled in the House and is organizationally unrelated to the permanent Senate intelligence oversight committee on which Wyden and company sit. In a crisis situation, these folks absolutely carry the same water, regardless of where they are on the org chart. One only need look at the retroactive immunity granted to the telcos for their long-standing, obvious violation of Federal law caused by their participation in NSA's then-very-illegal wiretap program to understand that.	null	null	42,563,233	null	null	null	[ 42565252, 42566072 ]	null	null	1,758,903,346
42,564,868	comment	Dylan16807	1,735,722,327	null	So as noted you can configure the length, but you can't actually adjust the probability used by the auto-length formula.<p>The formula calculates how many bits are needed to store the approximate number of objects, then uses twice that many bits, rounding up.<p>For example, at 16000 objects git is still using the minimum length of 7 characters. But at 17000 objects it now takes 15 bits to store the object count, so it wants 30 bits of hash, which means 8 characters.	null	null	42,557,735	null	null	null	null	null	null	1,758,903,346
42,564,914	comment	ustad	1,735,722,947	null	DJI Tello Drone is a “toy” drone costing around 100€/$ with the cool feature of being able to access/control via wifi with python.	null	null	42,563,940	null	null	null	[ 42565493 ]	null	null	1,758,903,346
42,564,957	comment	tuna74	1,735,723,613	null	I am going to assume that Lenovo, Dell, nVidia, AMD, Intel etc actually test Linux (or farm that out to RedHat etc) actually test Linux on the hardware that they support. But I could be wrong of course.	null	null	42,562,964	null	null	null	null	null	null	1,758,903,346
42,565,009	comment	pixxel	1,735,724,498	null	Look up ‘Susan Rosenberg’.	null	null	42,564,789	null	null	null	[ 42566706, 42567784, 42567438 ]	null	null	1,758,903,346
42,565,013	comment	xerox13ster	1,735,724,562	null	I told my crow friends happy new year and I'm reasonably certain they know today is special even if they don't get why. [0][1] (this is just one and I offered peanuts first but it stared me down and did not budge until I put out beef)<p>[0]: <a href="https://media.discordapp.net/attachments/433008121234063360/1323778149749751859/IMG_3230.png?ex=6775bff9&is=67746e79&hm=97400f90459c8d046aa0dd190d8fa929e3f64d94b47cabee7c6874e85cb88659&=&format=webp&quality=lossless&width=463&height=659" rel="nofollow">https://media.discordapp.net/attachments/433008121234063360/...</a><p>[1]: <a href="https://cdn.discordapp.com/attachments/434337586182619136/1323950712014766201/image.png?ex=677660af&is=67750f2f&hm=981477f34cbdf41522a307a1fda718bc7a650bdaabbb827029c342f3a7164708&" rel="nofollow">https://cdn.discordapp.com/attachments/434337586182619136/13...</a>	null	null	42,564,559	null	null	null	[ 42565773 ]	null	null	1,758,903,346
42,565,085	comment	palotasb	1,735,725,976	null	The spiral rule works only if there is no pointer to pointer or array of array in the type. In other words it is an incorrect rule. But take this for example:<p><pre><code> +----------------------------+ \| +-----------------------+ \| \| \| +------------------+ \| \| \| \| \| +-------------+ \| \| \| \| \| \| \| +--------+ \| \| \| \| \| \| \| \| \| +--+ \| \| \| \| \| \| \| \| \| \| ^ \| \| \| \| \| \| int * * ¦ ¦ ¦ VAR[1][2][3] \| \| \| ^ \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| +-----+ \| \| \| \| \| \| \| \| \| +----------+ \| \| \| \| \| \| \| +---------------+ \| \| \| \| \| ---------------------+ \| \| \| +-------------------------+ \| +-------------------------------+ </code></pre> The type of VAR is a [1-element] array of [2-element] array of [3-element] array of pointer to pointer to ints. I drew a spiral that passes through each specifier in the correct order. To make the spiral correct it has to skip the pointer specifiers in the first three loops. This is marked by ¦.<p>The Right-Left Rule is quoted less frequently on HN but it's a correct algorithm for deciphering C types: <a href="http://cseweb.ucsd.edu/~ricko/rt_lt.rule.html" rel="nofollow">http://cseweb.ucsd.edu/~ricko/rt_lt.rule.html</a><p>The spiral rule can be modified to process all array specifiers before all pointer specifiers, but then you'd have to specify that the order to do so is right and then left. At that point it's just the Right-Left Rule.	null	null	42,564,861	null	null	null	[ 42565117, 42566885, 42565807, 42565221 ]	null	null	1,758,903,346
42,565,110	comment	tlrg	1,735,726,359	null	Thanks for the answer, I'm checking it out.<p>I was thinking of a system where each user has its own object storage and autorize the Social Networks - SN access to it, behind the scenes the SN fetches each user files from different sources and the user can revoke access at any point.<p>This would require a client front-end, a object storage backend, a SN front-end and back-end.<p>SNs becomes a place to agregate and search for eachothers data in different ways, how this data is agregated presented ranked etc its the job of each SN, people could peak and choose where they want to present their data. People could make their own unique SN if they feel like it.	null	null	42,563,744	null	null	null	null	null	null	1,758,903,346
42,565,136	comment	m0llusk	1,735,726,816	null	It starts with sorting by moral judgements. Reactionaries are particularly sensitive to unfairness, betrayal, subversion, and degradation of sanctity. Weave these things with a narrative and you can capture interest.	null	null	42,563,048	null	null	null	null	null	null	1,758,903,346
42,565,144	comment	vintermann	1,735,726,973	null	I have a feeling that you identify not with the class which is in need of more motivation, but the class holding the whips.<p>I also think it's likely you are <i>actually</i> in the other class than what you believe.	null	null	42,564,564	null	null	null	null	null	null	1,758,903,346
42,565,203	comment	seanhunter	1,735,727,865	null	Yes actually I think you're right.	null	null	42,565,156	null	null	null	null	null	null	1,758,903,346
42,565,236	comment	mgav	1,735,728,326	null	Excerpt: a corporation deriving receipts from sources within this State shall be deemed to have substantial nexus and is subject to the taxes imposed under the Corporation Business Tax Act (1945)…	null	null	42,565,220	null	null	null	null	null	null	1,758,903,346
42,565,286	comment	krzat	1,735,728,846	null	The whole guide could be retitled as: how to cultivate your own psychosis. The technology is sound but does not lead to wellbeing.<p>I think Buddhists figured it out (at least some of them). For example metta practice could be thought of as a magick that is specifically aimed toward wholesome happiness.	null	null	42,564,787	null	null	null	[ 42565363 ]	null	null	1,758,903,346
42,565,323	comment	creer	1,735,729,435	null	They did pay for the PhD - perhaps even in addition to his salary. He was also rather successful and useful at TI.	null	null	42,561,498	null	null	null	null	null	null	1,758,903,346
42,565,365	comment	atombender	1,735,730,252	null	DejaVu is one of my favorites, and it is odd to not see it included.<p>Input Mono [1] by David Jonathan Ross is what I use these days. It's very similar to DejaVu, but I like it better because its geometry is a little rounder. DejaVu's "m" is very narrow, and it has fewer serifs (l, i, etc.), but it also has a serif version for those who want that (it's still very "sans"). It comes in many weights and has a proportional version which is top notch even for graphical design. The designer spent a lot of time on the fine details of the font. It's free for personal use.<p>Jetbrains Mono, which is included in the competition, is very good. Very similar to both DejaVu and Input.<p>[1] <a href="https://input.djr.com/" rel="nofollow">https://input.djr.com/</a>	null	null	42,558,451	null	null	null	null	null	null	1,758,903,346
42,565,377	comment	Keysh	1,735,730,411	null	I'll admit I've never seen "track" used to mean "write" before.	null	null	42,531,577	null	null	null	null	null	null	1,758,903,346
42,565,382	comment	adwn	1,735,730,489	null	One argument against that line of thinking is that energy production has negative externalities. If you use a lot of electricity, its price goes up, which incentivizes more electricity production, which generates more negative externalities. It will also raise the costs for other consumers of electricity.<p>Now that alone is not yet an argument against crypto currencies, and one person's frivolous squandering of resources is another person's essential service. But you can't simply point to the free market to absolve yourself of any responsibility for your consumption.	null	null	42,564,255	null	null	null	[ 42576484, 42565981 ]	null	null	1,758,903,346
42,565,475	comment	gboss	1,735,732,018	null	Happy new year everyone!	null	null	42,562,750	null	null	null	null	null	null	1,758,903,346
42,565,511	comment	ChrisMarshallNY	1,735,732,670	null	The Programmer’s Credo:<p><i>We do what we do. Not because it is easy; but because we thought it would be easy.</i>	null	null	42,559,882	null	null	null	null	null	null	1,758,903,346
42,565,600	comment	Dolloinfo	1,735,733,998	null	[dead]	null	null	42,565,599	null	true	null	null	null	null	1,758,903,346
42,565,697	comment	sylware	1,735,735,597	null	I want a RP chip with the ARM block which was permanently disabled in hardware on which raspberry did not pay the ARM ISA royalties.<p>How can I get my hands of one of those boards? If they do exist?	null	null	42,565,680	null	null	null	null	null	null	1,758,903,346
42,565,745	comment	tpoacher	1,735,736,219	null	It's not (unless you mean "by popular convention", which actually is not necessarily that popular).<p>int* c is perfectly valid syntax<p>and so is int * c.<p>The latter is my preferred convention: it helps me think of the asterisk as an operator, which, when acting at the "declaration" level, is binary in nature.<p>Whereas when the asterisk is used at the 'expression' level, it is unary in nature, and can be thought of effectively as a completely different operator that happens to share the same name/symbol as the declaration-based one.	null	null	42,565,465	null	null	null	[ 42565844 ]	null	null	1,758,903,346
42,565,748	comment	lazide	1,735,736,273	null	My comment did not refer to the article at all. It was if they had read the link to the law they pasted. Which near as I can tell they hadn’t, because the content of the law is the opposite of what they are asserting it is.<p>Feel free to check the law yourself if you’d like.	null	null	42,565,663	null	null	null	null	null	null	1,758,903,346
42,565,751	comment	HexDecOctBin	1,735,736,330	null	Regarding pre-processing, I feed the result of `clang -E` to the code-generator. Since I use a unity build (single translation unit), it works out fine. In fact, the parser treats pre-processor directives as comments (to work around #line and `-dD`, etc.)<p>Regarding external and intrusive, I used to do it in the intrusive way but found it too limiting. Here, I not only generate code but can also (potentially) add whole new extensions to the language. This was the reason I wrote a new parser instead of just using libclang's JSON AST dump. Well, that and the fact that libclang is a multi-MB dependency while my parser is ~3000 lines of C code.	null	null	42,565,583	null	null	null	null	null	null	1,758,903,346
42,565,861	comment	magic_smoke_ee	1,735,737,782	null	[flagged]	null	null	42,563,194	null	true	null	[ 42568286 ]	null	null	1,758,903,346
42,565,904	comment	YetAnotherNick	1,735,738,407	null	First of all, Putnam is not in the test data, at least I haven't seen OpenAI claiming that publicly. Secondly, removing it from internet data is not 100% accurate. There are translations of the problems and solutions or references and direct match is not enough. MMLU and test set benchmarks show more resilience though in some previous research.	null	null	42,565,774	null	null	null	[ 42566378, 42566458, 42566552, 42566492, 42565924 ]	null	null	1,758,903,346
42,565,960	comment	HexDecOctBin	1,735,738,990	null	Stuff like:<p><pre><code> char ((**foo[][8])())[]</code></pre>	null	null	42,565,865	null	null	null	[ 42565974 ]	null	null	1,758,903,346
42,566,006	comment	unrealhoang	1,735,739,476	null	you should just see &mut and & as 2 separate modifier, maybe that'd help.<p>- & &mut doesn't make sense: and a variable of type & &mut can't mutate the underlying object, it's practically equivalent to a & &, even worse, since the inner is &mut, you can't have 2 & &mut that point to the same &mut (mut xor shared).<p>- &mut &: mutable reference that point to a shared reference, means you can change the outer to point to another shared reference, but you can't change content of the inner, for example:<p><pre><code> let a: &str = "hello"; let b: &str = "world"; let mut c = &a; let d = &mut c; d = &b; // (d).make_ascii_lowercase(); // not allowed </code></pre> - &mut &mut: similar to the above, but you can also change the content of the inner.	null	null	42,565,787	null	null	null	null	null	null	1,758,903,346
42,566,029	comment	technothrasher	1,735,739,807	null	> Since our brains aren’t naturally wired for reading<p>There's actually some evidence that we are, in fact, naturally wired for reading. The below study, for instance, shows that the area in the brain used for visualizing words seems to be already hooked up to language processing areas in newborns.<p><a href="https://www.nature.com/articles/s41598-020-75015-7" rel="nofollow">https://www.nature.com/articles/s41598-020-75015-7</a>	null	null	42,565,347	null	null	null	null	null	null	1,758,903,346
42,566,108	comment	adolph	1,735,740,664	null	Was there some previous interpretation that was utopian?	null	null	42,565,574	null	null	null	null	null	null	1,758,903,346
42,566,114	comment	tyleo	1,735,740,752	null	The car dependency—which I’ve had to live with my whole life—doesn’t bother me as much as what’s happening with trucks specifically. It’s like the only remaining trucks for sale are all monster trucks.	null	null	42,565,781	null	null	null	[ 42566211 ]	null	null	1,758,903,346
42,566,176	comment	jpc0	1,735,741,483	null	Which version of GTK should I pick since you seem so confident?<p>And should I target X or Wayland? Which Wayland extensions are ubiquitous enough so that I don't need to manually reimplement things in my app? Should I target pipewire for audio? Pulse?<p>If I pick a modern stack like Wayland + GTK 4 + pipewire it might not work on older linux distros. Maybe that's fine for my needs, maybe it's not. If I pick X + GTK 3 + pulse I might end up reimplementing half of the desktop manager in my app and need to constantly rewrite my app.<p>Do I distribute deb/rpm or do I distribute a flatpack/snap... Actually which one.<p>And don't bring up that GTK abstracts Wayland/X because if I target Wayland I have access to a ton of Wayland extensions which aren't there in X so I do need to make that decision the moment I need to get access to that stuff.<p>So even just picking GTK I now have a matrix of 9 different configurations. GTK is not the defacto framework on linux. We now need to add QT to the mix and the whole collection of things there.	null	null	42,564,391	null	null	null	[ 42571925 ]	null	null	1,758,903,346
42,566,269	comment	rcxdude	1,735,742,620	null	Music has, through various court decisions through the years, gained a particularly arcane and arbitrary set of rules on what parts of a piece are copyrightable, and to what extent. It's a mess, as some parts are overly-protected, like the layout of the music on the page (unlike the typesetting of a book, or the font used!), and some under-protected in comparison, like the rhythm of a song compared to it's melody (which has a particularly arbitrary rule of 7 notes long). Though, to answer the question of the last part, if that version was copyrightable, it depends what book the performers are using! Though again, it could be that the essence of the performance is not considered to be a violation, even if photocopying the book would be, depending on what form any differences take.<p>If you're concluding copyright is impossibly complicated, vague, and confusing and the only reason the system works at all is because hashing out the details in court is either a nuclear option between two large companies, or enforcement on a large scale against small actors is basically impossible (but will unpredictably come down like a ton of bricks on some unlucky individuals), then you'd be right!	null	null	42,553,291	null	null	null	null	null	null	1,758,903,346
42,566,306	comment	spookie	1,735,743,003	null	Yup, basically that. Furthermore, you are unable to export the meshes themselves.<p>You are better off using Google Maps API if I'm being honest.	null	null	42,565,739	null	null	null	null	null	null	1,758,903,346
42,566,340	comment	gdhkgdhkvff	1,735,743,355	null	Your points would be more convincing if you didn’t preface them with arrogant cynicism.	null	null	42,566,267	null	null	null	null	null	null	1,758,903,346
42,566,361	comment	rep_lodsb	1,735,743,513	null	16-<i>color</i> VGA, not 16 <i>bits</i> (which would be 65536 colors).	null	null	42,565,721	null	null	null	null	null	null	1,758,903,346
42,566,503	comment	amluto	1,735,744,736	null	> ask it for a formula for mass-energy equivalence<p>Way too easy. If you think that mass and energy might be equivalent, then dimensional analysis doesn’t give you too much choice in the formula. Really, the interesting thing about E=mc^2 isn’t the formula but the assertion that mass is a form of energy and all the surrounding observations about the universe.<p>Also, the actual insight in 1905 was more about asking the right questions and imagining that the equivalence principle could really hold, etc. A bunch of the math predates 1905 and would be there in an AI’s training set:<p><a href="https://en.m.wikipedia.org/wiki/History_of_Lorentz_transformations" rel="nofollow">https://en.m.wikipedia.org/wiki/History_of_Lorentz_transform...</a>	null	null	42,565,998	null	null	null	[ 42566728, 42566529 ]	null	null	1,758,903,346
42,566,545	comment	sonabinu	1,735,745,073	null	Could also be husband and kids ;)	null	null	42,566,499	null	null	null	[ 42566761, 42569958 ]	null	null	1,758,903,346

End of preview. Expand in Data Studio

Hacker News Dataset (2025)

Dataset Description

A comprehensive dataset of Hacker News content from 2025, containing stories, comments, users, and their relationships. This dataset enables deep analysis of technical discussions, trends, and community dynamics on one of the most influential technology forums.

Dataset Summary

Total Records: 38.4M+ across 10 tables
Stories: 287K+ submissions including links, Show HNs, Ask HNs
Comments: 2.5M+ discussion comments with full threading
Users: 144K+ user profiles with karma and submission history
Time Period: 2025 data snapshot
Format: Parquet files optimized for analytical queries

Dataset Structure

Core Tables

`stories` (287,204 records)

Primary content submissions to Hacker News.

Fields: id, type, by (author), time, title, text, url, score, descendants (comment count), dead, deleted, parent, poll, kids, parts, fetched_at
Use Cases: Trend analysis, content ranking, topic modeling

`comments` (2,539,135 records)

User discussions on stories and other comments.

Fields: id, type, by, time, text, parent, dead, deleted, kids, score, fetched_at
Use Cases: Sentiment analysis, discussion patterns, community dynamics

`users` (144,479 records)

User profiles with activity metrics.

Fields: id, created, karma, about (bio), submitted (list), fetched_at
Use Cases: User behavior analysis, influence mapping, community growth

Relationship Tables

`user_submissions` (30,116,739 records)

Maps users to all their submitted items.

Fields: user_id, item_id
Use Cases: User activity patterns, prolific contributors

`item_children` (2,536,309 records)

Parent-child relationships for comment threading.

Fields: parent_id, child_id
Use Cases: Thread reconstruction, discussion depth analysis

Additional Content Types

`jobs` (432 records)

Job postings from "Who is hiring?" threads.

`polls` (32 records) & `pollopts` (172 records)

Poll submissions and their options.

`item_parts` (164 records)

Links poll options to their parent polls.

`items` (2,826,975 records)

Union of all content types (stories + comments + jobs + polls).

Data Quality

Coverage

Complete comment threading with hierarchical relationships
Full text content for stories and comments
User karma and creation timestamps
Vote scores and deletion status

Known Limitations

Some comments referenced in item_children may not exist in the comments table
descendants count may not match actual comment count due to deletions
Text content may be null for deleted/dead items
Snapshot represents a point in time, not real-time data

Usage with Fenic

This dataset is optimized for use with Fenic, a PySpark-inspired DataFrame framework that enables:

Quick Start

from fenic.api import Session

# Create session and load data
session = Session.get_or_create()
stories = session.read.parquet("hf://datasets/typedef-ai/hacker-news-dataset/data/2025_stories.parquet")

# Analyze top discussions
top_discussions = (
    stories
    .filter(stories["descendants"] > 100)
    .order_by(stories["score"].desc())
    .select("title", "score", "descendants", "url")
    .limit(10)
)

Advanced Analysis

# Find AI/ML discussions with semantic search
from fenic.api.functions import semantic

ai_stories = (
    stories
    .filter(stories["title"].rlike("(?i)(artificial intelligence|machine learning|\\bAI\\b|\\bML\\b)"))
    .with_column("summary", 
        semantic.extract(
            stories["title"], 
            "Extract key AI/ML topics discussed",
            output_type=str
        )
    )
)

# Analyze comment sentiment
comments = session.read.parquet("hf://datasets/typedef-ai/hacker-news-dataset/data/2025_comments.parquet")
sentiment_analysis = comments.select(
    semantic.extract(
        comments["text"],
        "Determine if this comment is positive, negative, or neutral",
        output_type=["positive", "negative", "neutral"]
    ).alias("sentiment")
)

Example Research Questions

Technology Trends: What technologies gained the most traction in 2025?
Discussion Patterns: How do comment threads evolve on controversial topics?
Community Dynamics: Who are the most influential users by karma vs. engagement?
Content Analysis: What makes a story reach the front page?
Sentiment Tracking: How does the community sentiment vary across different topics?

Loading the Dataset

Using HuggingFace Datasets

from datasets import load_dataset
dataset = load_dataset("typedef-ai/hacker-news-dataset")

Direct Parquet Access

import pandas as pd
stories_df = pd.read_parquet("hf://datasets/typedef-ai/hacker-news-dataset/data/2025_stories.parquet")
comments_df = pd.read_parquet("hf://datasets/typedef-ai/hacker-news-dataset/data/2025_comments.parquet")

With Fenic (Recommended for Analysis)

from fenic.api import Session

session = Session.get_or_create()
base_path = "hf://datasets/typedef-ai/hacker-news-dataset/data"

# Load all tables
stories = session.read.parquet(f"{base_path}/2025_stories.parquet")
comments = session.read.parquet(f"{base_path}/2025_comments.parquet")
users = session.read.parquet(f"{base_path}/2025_users.parquet")

Dataset Statistics

Average story score: ~11 points
Average comments per story: ~9 comments
Most commented story: 1,467 comments
Highest scored story: 3,402 points
Active users: 144,479 unique contributors
Total submissions: 30M+ user-item relationships

Citation

If you use this dataset in your research, please cite:

@dataset{hackernews2025,
  title={Hacker News Dataset 2025},
  author={Typedef},
  year={2025},
  publisher={HuggingFace},
  url={https://huggingface.co/datasets/typedef-ai/hacker-news-dataset}
}

License

This dataset is provided for research and analytical purposes. Please respect the original content creators and Hacker News community guidelines when using this data.

Acknowledgments

Hacker News: For being an invaluable platform for technology discussions
Y Combinator: For maintaining and hosting Hacker News
Fenic Framework: For enabling efficient large-scale analysis of this dataset
Community: All the users who contribute to making HN a vibrant forum for technical discourse

Downloads last month: 59

Size of downloaded dataset files:

1.65 GB

Size of the auto-converted Parquet files: