Overwhelmed by endless Hacker News feeds? Get the lowdown fast with HackerDigest. We serve up the latest in tech, condensed into quick, easy-to-read summaries. Stay in the loop without the time sink. 🚀

View on Github


So you think you know box shadows?

11 hour(s) ago

The article explores the use of box shadows in graphic and user interface design, discussing their technical aspects and potential benefits. The author shares their personal experimentation with creating a 3D effect using box shadows and provides examples and tools for others to try. They also discuss the potential for further exploration and optimization of this technique, including using it for ray tracing and CGI. The author concludes by expressing their frustration with the lack of information on box shadow ray tracing and hopes for future advancements in this area.

Top three comments:

### The commenter created a layout engine for their portfolio website five years ago that heavily utilizes animated box shadows. They provide a link to their website, thestrikeagency.com.

Author: noduerme

Posted: 7 minute(s) ago

### The commenter is discussing the impact of using transparent colors on drawing speed. They mention that transparency can slow down the number of draws that can be done, which is surprising given the capabilities of modern hardware. They explain that this is because transparency affects how the GPU can batch draws, and that drawing in the correct order is necessary for blending to work.

Author: sunnyps

Posted: 6 hour(s) ago

### The commenter is expressing their enjoyment of exploring different techniques, specifically mentioning the importance of layering. They also mention a project they worked on 14 years ago involving text shadows, which they describe as both silly and visually appealing. They provide a link to the project for others to check out.

Author: paulirish

Posted: 8 hour(s) ago

Author: yohannesk

Num of comments: 60

Score: 451

User returns after 100k-hours ban to continue conversation that got them banned

11 hour(s) ago

A banned Something Awful forums user, Etherwind, returned after 100,000 hours to continue a tabletop game debate that got him banned. Despite growing and changing during his ban, he chose to resume the argument for comedic effect, impressing other forum members with his dedication. The article also briefly mentions the impact of online communities and the author's love for gaming and writing, as well as a new D&D handbook and a Russian Oblivion mod being translated into English.

Top three comments:

### The commenter shares their experience of trying to return to their Habbo Hotel account after a 10 year ban. They struggled to login and contacted support, only to find out that the account had been deleted after being banned for more than six months. The commenter expresses disappointment and humorously mourns the loss of their account, which they waited 9.5 years to log into.

Author: benry1

Posted: 10 hour(s) ago

### The commenter shares a story from 6 years ago when they accidentally changed their HackerNews settings and set their "minaway" to 6 years without realizing what it meant. They then mention that their first comment after being "self-banned" was not as impressive as they had hoped.

Author: fiatpandas

Posted: 10 hour(s) ago

### The commenter is discussing the practice of "necroposting," which refers to reviving old threads on online forums. They find this aspect of forum culture interesting and note that different sites have varying norms around necroposting. They mention that some modern sites automatically archive old threads, but they wonder why Hacker News disables comments on old posts. They express disappointment when they come across old threads on Hacker News that are locked and cannot be replied to. They ask if anyone else enjoys posting on

Author: aspenmayer

Posted: 3 hour(s) ago

Author: campuscodi

Num of comments: 120

Score: 258

When ChatGPT summarises, it does nothing of the kind

4 hour(s) ago

The author's experience with ChatGPT reveals that the AI is not capable of true summarization, as it simply shortens texts instead of summarizing them. The limitations of using large language models (LLMs) for summarization are also discussed, highlighting the need for user guidance and specific prompts to produce high-quality results. The author suggests that simpler methods such as information retrieval with curated knowledge graphs or BERT QA may be more effective in producing actionable summaries.

Top three comments:

### The commenter is questioning the lack of mention of the GPT model used in the article and believes that the author did not provide enough information for readers to fully understand the problem. They mention the context size, prompt, and number of attempts as important factors that are missing from the article. The commenter also gives an example of how ChatGPT did not accurately summarize their own writing, but has had success using it for summarizing articles and writing a grant application. They praise the model's understanding and

Author: kibbi

Posted: 2 hour(s) ago

### The commenter believes that there is a fundamental flaw with using AI for summarization tasks, as indicated by the disclaimer on these products. They argue that a summary that requires the reader to verify the information is essentially useless. They criticize AI developers for ignoring this issue and using RAG (a type of AI) to make it seem more useful than it actually is. They suggest that AI should stick to its designated purpose instead of trying to do tasks like summarization.

Author: ADeerAppeared

Posted: 2 hour(s) ago

### The commenter read the entire article and found that the three main points they took away were also included in the GPT4o summary they generated. They then provide counter evidence and suggest that GPT4o may have trouble summarizing longer content or may be using a less effective attention mechanism. They also provide a link to the GPT4o summary they generated.

Author: ianbutler

Posted: 2 hour(s) ago

Author: josephcsible

Num of comments: 145

Score: 211

Intel says 13th and 14th Gen mobile CPUs are crashing

7 hour(s) ago

Intel has clarified that the instability affecting their 13th and 14th Gen mobile CPUs is not the same as the instability on their desktop Raptor Lake and Raptor Lake Refresh processors. The cause of the mobile chip instability is still unknown and could be due to various software and hardware issues. This has affected Intel's reputation and may impact consumer trust, especially with the upcoming launch of AMD's Ryzen 9000 processors. Additionally, Tom's Hardware, owned by Future US Inc, now offers multiplayer at a reduced price, as announced by the FTC.

Top three comments:

### The commenter expresses frustration and disbelief that the issue with CPUs impacting desktops is still ongoing six months after it was first reported. They estimate that a significant percentage of CPUs will eventually fail, including their own. They criticize Intel for not being able to recall and replace the affected products, and express concern that the company may be hiding the true cause of the issue, which could result in legal action.

Author: PaulKeeble

Posted: 4 hour(s) ago

### The commenter is referencing a prediction made in 2016 about CPU bugs becoming more severe, specifically with the Skylake processor. They also mention that the article they linked to has been updated with new information.

Author: userbinator

Posted: 2 hour(s) ago

### The commenter mentions that there were previous discussions about desktop CPU issues on Hacker News, and now there is a new issue being discussed. They provide links to the previous discussions for reference.

Author: magicalhippo

Posted: 4 hour(s) ago

Author: markus_zhang

Num of comments: 91

Score: 149

Trench collapses have killed hundreds of workers in the US over the last decade

11 hour(s) ago

An investigation by NPR, Texas Public Radio, and 1A reveals that trench collapses, which have killed hundreds of workers in the US over the last decade, are preventable and often occur due to employers ignoring safety regulations. Despite being fined by OSHA, companies face little consequences and criminal charges are rare. More needs to be done to protect workers, especially with the expected increase in construction workers due to President Biden's infrastructure deal.

Top three comments:

### The commenter discusses a news article about preventable deaths in the construction industry and shares their personal experience with a friend who works in the industry. They argue that blaming the companies is not the only solution, as they have witnessed workers refusing to follow safety protocols and being fired for it. They suggest that while companies should be held accountable, the issue is more complex and requires constant monitoring of workers to ensure they are following safety procedures.

Author: Aurornis

Posted: 10 hour(s) ago

### The commenter is discussing the issue of companies violating OSHA regulations and facing little to no consequences, even in cases where employees have died. They believe that there should be stricter consequences, including criminal charges, for those who violate OSHA standards. They also mention that the Supreme Court may be targeting OSHA in the future, based on recent events.

Author: BadHumans

Posted: 9 hour(s) ago

### The commenter mentions that OSHA requires a protective system for trenches deeper than 5 feet. They then share a personal experience of witnessing a gas leak being fixed by PG&E, where the workers were standing in a trench about 6 feet deep without any protective walls.

Author: AlbertCory

Posted: 8 hour(s) ago

Author: rntn

Num of comments: 136

Score: 122

PgManage: Modern, cross platform graphical database client

11 hour(s) ago

PgManage is a modern, cross-platform graphical database client with a focus on Postgres management. It leverages the work of the now dormant project OmniDB and offers features for developers. The website and documentation provide all necessary information, and the tool can be downloaded and installed easily.

Top three comments:

### The commenter tried out a program and found that it visually looked okay, but did not support basic connectivity options like unix sockets. They were disappointed that the hostname field for new connections could only take a hostname or IP address and not things like `/var/run/postgresql`. They also mentioned that the program creates config files in specific directories on Linux and suggested deleting them if the program is no longer needed.

Author: justinclift

Posted: 10 hour(s) ago

### The commenter is excited about the fact that PgManage, a SQL IDE, is based off of OmniDB, which is one of the few free SQL IDEs that has a debugger. They acknowledge that the debugger's user interface may not be the best, but it still functions. They provide a link to the GitHub page for the debugger.

Author: infamia

Posted: 8 hour(s) ago

### The commenter is asking if there is a comparison between pgManage and pgAdmin, as they seem to serve the same purpose. They mention that pgManage targets postgres primarily, but they are wondering if there are any other differences between the two.

Author: jand

Posted: 10 hour(s) ago

Author: thunderbong

Num of comments: 38

Score: 107

A brief history of Dell Unix (2008)

9 hour(s) ago

The article discusses the development and release of Dell UNIX, a UNIX operating system created by Dell in the late 1980s and early 1990s. Despite its initial success and integration into the Dell Station, the project was eventually terminated due to lack of profitability. The last major release was in 1992 and support for the system ended in 1993, although it is still used by some without DHCP support.

Top three comments:

### The commenter is asking for information about Jeremy Chatfield, who was mentioned in an article and has seemingly disappeared since 2014. They are wondering if he may have passed away, but have been unable to find any information about his whereabouts.

Author: rwmj

Posted: 4 hour(s) ago

### The commenter has had positive experiences using Dell computers with the operating system FreeBSD for 25 years. They mention a dislike for the iDrac system and how it affects the SAS hba.

Author: ggm

Posted: 1 hour(s) ago

### The commenter suggests that Dell's decision to sell UNIX was primarily driven by their desire to sell hardware, but most of the UNIX copies were actually used on other vendors' machines. They speculate that Microsoft may have influenced Dell's decision through threats or incentives related to their software prices.

Author: bluedino

Posted: 3 hour(s) ago

Author: fanf2

Num of comments: 5

Score: 69

Friendship and Social Fitness

6 hour(s) ago

The article emphasizes the significance of maintaining friendships and the consequences of neglecting them. It also highlights the growing concern of loneliness and friendlessness in society, particularly among younger generations. The author shares personal experiences and urges readers to prioritize and invest in their friendships before it's too late.

Top three comments:

### The commenter shares their experience of having close and meaningful friendships over the past 43 years, some of which have been transient. They attribute their ability to maintain these friendships to being open-minded, honest, and accepting. They also mention the importance of not burning bridges and the negative impact of attacking others in online interactions. They believe that remote communication and limited interaction can lead to false narratives and a lack of understanding of others' life stories. They express concern about this antagonistic behavior translating into real-life

Author: ChrisMarshallNY

Posted: 3 hour(s) ago

### The commenter believes that it is important to maintain investments and relationships, even if it means having occasional conversations. They criticize a quote in an article for being out of touch and argue that maintaining relationships is not as simple as the author suggests. The commenter also mentions that the author themselves acknowledges the difficulty of maintaining relationships.

Author: zokier

Posted: 3 hour(s) ago

### The commenter enjoyed reading the article and agrees with the author's thoughts on reconnecting with old friends during COVID lockdowns. They also ponder the relationship between having close friends and a close family, and whether one holds more importance than the other. They suggest making time to meet up with a friend this week instead of going to the gym.

Author: two-sandwich

Posted: 4 hour(s) ago

Author: paulpauper

Num of comments: 16

Score: 65

Why I left Google

2 hour(s) ago

The author left Google due to a misalignment of values and a realization about the use of AI in their job. They reflect on a transformative experience at a theater performance and ultimately choose to embrace it and let it change them.

Top three comments:

### The commenter shared a link to a blog post about why the author left Google, which was archived on the Wayback Machine.

Author: neonate

Posted: 53 minute(s) ago

### The commenter reflects on a management training course they took 20 years ago, which stated that if one's personal values do not align with the company's, it is best to leave for the sake of one's mental health. They have been contemplating this advice ever since.

Author: ggm

Posted: 1 hour(s) ago

### The commenter is frustrated that the author went off on a tangent about theater and never fully explained their issue with AI. They are interested in hearing more about the author's perspective and how Google's approach to AI has changed in the past 3 years. They also want to know what the author's outlook was on AI 3 years ago and how it has evolved since then.

Author: jamra

Posted: 33 minute(s) ago

Author: doodpants

Num of comments: 48

Score: 64

The data that powers AI is disappearing fast

4 hour(s) ago

The data used to train artificial intelligence systems is rapidly disappearing due to privacy concerns and stricter regulations. This could hinder the development and progress of AI technology in the future.

Top three comments:

### The commenter shared a link to an archived webpage and did not provide any additional context or explanation.

Author: neonate

Posted: 53 minute(s) ago

### The commenter discusses the potential negative impact of AI outside of the tech industry. They believe that AI poses a threat to those in positions of power, such as monopolies and nation-state autocrats, and compares it to the computer and internet revolutions. They also mention how these powerful individuals are trying to resist the disruption caused by AI, particularly in the content creation industry. However, the commenter acknowledges that the future impact of AI is uncertain and that content creators are not necessarily facing a worse outlook than others

Author: worstspotgain

Posted: 23 minute(s) ago

### The commenter believes that the worst possible outcome for the use of AI-based tools would be a licensing regime that gives large companies a monopoly on training large models within their specific niche. They predict that any calls for regulation will result in this outcome, making it illegal for individuals to use AI-based tools except for creating works for hire. They believe that the current law, which allows for the use of facts without copyright, is better for society and creators than a licensing regime. They are currently in favor of

Author: hedora

Posted: 3 hour(s) ago

Author: sgammon

Num of comments: 72

Score: 59

Show HN: Moocable – find people studying the same online course/book

10 hour(s) ago

Moocable is a new platform that helps users find others who are studying the same online course or book. It has been mentioned multiple times on Hackernews and offers a list of the most popular courses on the site.

Top three comments:

### The commenter has been attempting to create a study group for a specific course, but is having trouble because the course is not currently in the library. They are wondering if it is possible to add the course to the library so they can create the study group. They provide a link to the course they are interested in.

Author: SamBorick

Posted: 3 hour(s) ago

### The commenter congratulates the launch of the site and asks if the person has personally found success in finding a study partner on similar sites. They suggest promoting the site on a subreddit as a way to attract potential study buddies.

Author: raybb

Posted: 1 hour(s) ago

### The person is experiencing an issue with their Safari browser on their iPhone and is unable to proceed due to a "checking your browser" message.

Author: BostonFern

Posted: 1 hour(s) ago

Author: junaid_97

Num of comments: 14

Score: 52

DeepL's LLM Outperforms Google Translate, ChatGPT-4, and Microsoft

8 hour(s) ago

DeepL's new language model, powered by LLM technology, surpasses Google Translate, ChatGPT-4, and Microsoft in translation quality and productivity. It is preferred by language experts in blind tests and has a reduced risk of errors, making it a trusted AI solution for businesses. The model is currently available for select languages and has shown a 345% ROI, with improved security and accessibility for Pro users.

Top three comments:

### The commenter is skeptical of DeepL's claims of having the best translation models. They point out that DeepL's models are trained with high-quality linguistic data and the expertise of language specialists, but they question the validity of recent blind tests that show DeepL's translations are preferred over Google Translate, ChatGPT-4, and Microsoft. They argue that if Google Translate is preferred over ChatGPT-4, which is considered superior, then the test must be biased or not representative of real

Author: A_D_E_P_T

Posted: 8 hour(s) ago

### The commenter has used DeepL as an alternative to Google Translate in an attempt to reduce their use of Google products. They find that DeepL is generally good, but the Google Translate UI is better for Chinese due to its inclusion of Pinyin. However, the quality of translation using DeepL varies and can sometimes be better or worse than Google Translate. The commenter still recommends DeepL for its privacy policies, API integration, and the fact that using both DeepL and Google Translate together can result in

Author: sshine

Posted: 8 hour(s) ago

### The commenter uses deepl regularly and finds it to be the best translation tool due to its context awareness. They also appreciate the company's use of LLMs in a targeted manner for their business model and product.

Author: bilekas

Posted: 8 hour(s) ago

Author: geox

Num of comments: 18

Score: 40