1. Tinnitus linked to undetected auditory nerve damage
Total comment counts : 80
Summary
A study conducted by researchers at Massachusetts Eye and Ear has found a link between tinnitus and undetectable auditory nerve loss. The study challenges previous beliefs about tinnitus and opens up new possibilities for treatments focused on nerve regeneration. The researchers discovered that individuals with tinnitus, even those with normal hearing tests, experience auditory nerve loss that is not detected by conventional tests. Tinnitus can cause debilitating symptoms, including sleep deprivation, social isolation, anxiety, and depression. The researchers hope that understanding the mechanisms underlying tinnitus will eventually lead to a cure. The study was funded by the National Institutes of Health and the Lauer Tinnitus Research Center.
Top 1 Comment Summary
The author discusses their experience with tinnitus and how certain muscle movements in their head, face, and jaw can worsen it temporarily. They speculate that their tinnitus may be caused by physical or muscle-related factors rather than hearing damage or neurological issues. The author shares this information to see if others have had a similar experience and to gather insight into the possible causes of this “muscle-related tinnitus.”
Top 2 Comment Summary
The author recently purchased audiomoths to monitor birds passing through their area. They discovered that the audiomoth can pick up ultrasonic chirps from bats due to its recording capability of up to 192kHz. While using Audacity to search for bird calls, the author determined the bandwidth where their tinnitus overwhelms other signals, shedding light on the extent of their work-related hearing loss. The author uses low and high pass filters to extract signals across discrete frequencies and measures the necessary gain to hear the calls in each extracted band. This information could potentially be used to design a hearing aid that boosts sounds in the affected bands. The author also mentions the possibility of using a notch filter to address the narrowband tinnitus-related noise.
2. Tiny volumetric display
Total comment counts : 41
Summary
The author of the article describes their idea of creating a flickering candle effect from any angle using electronic candles. They suggest using a persistence-of-vision display but decide that it requires too much supporting machinery. Instead, they come up with the idea of a spinning motor and battery. They quickly order circuit boards from China and use a pick and place machine to assemble them. The LED matrix board has only one component, making the process quick. The author mentions using different sizes of LED matrices and experimenting with circuit board designs. They discuss their choice of microcontroller and battery, opting for a minimal RP2040 board and a lithium-ion rechargeable battery. Overall, the author describes their initial exploration and experimentation with creating a small LED matrix display.
Top 1 Comment Summary
This article suggests a couple of ideas for improvement. One idea involves using extra identical motors to create a rotating power transmission system. By disassembling two motors and using the brushes, it is possible to create a reliable system with a cap and rectifier. This eliminates the need for batteries and allows for image stability based on the actual position of the assembly. Another idea mentioned is transmitting power via induction, but the efficiency of this method was not sufficient. In order to make it work, the source voltage had to be increased to a level where fire safety became a concern. These ideas were derived from a Burning Man art project from 2001, and a prototype can be seen in the provided link.
Top 2 Comment Summary
The article mentioned micro LED panels, but it did not provide any information about their commercial availability.
3. You don’t need JavaScript for that
Total comment counts : 69
Summary
The article discusses the core principle of web development called “the least powerful language suitable for a given purpose”. It explains that JavaScript (JS) is the most versatile language, but it can break, fail to load, and be inaccessible to some users. In contrast, HTML and CSS are declarative and handled by the browser, making them more efficient, native, adaptable, and accessible. The article emphasizes that many functionalities that previously required JS can now be achieved with CSS and HTML. It provides an example of creating a custom switch using HTML, CSS, and the :checked pseudo-class. The article encourages developers to test assumptions and consider using HTML and CSS instead of defaulting to JS for certain tasks.
Top 1 Comment Summary
This article highlights several issues with the use of JavaScript in web development compared to CSS and HTML. The author states that JavaScript is often used for better compatibility, as new JavaScript can be transpiled, whereas missing CSS and HTML features are harder to polyfill, and sometimes impossible. The article also mentions some caveats with the appearance property, particularly in older versions of Safari. The datalist element has limited functionality on Firefox Android, while the color picker is deemed nonstandard and not ideal for businesses due to its inconsistency across browsers. The author acknowledges the inconsistencies with the details and dialog elements as well. Overall, the author hopes that browsers without consistent support for these features will eventually become obsolete, but for now, they’re mostly used in side projects where browser support can be controlled.
Top 2 Comment Summary
The article highlights the ongoing challenges with forms and user experience for entering data. The author expresses frustration that these issues have not been resolved despite advancements in technology. They mention a failed attempt to address the problem with XForms, and note that most CSS/JS component frameworks offer their own limited form elements with varying semantics.
4. Show HN: 80% faster, 50% less memory, 0% loss of accuracy Llama finetuning
Total comment counts : 19
Summary
The article mentions that the feedback provided by users is taken seriously. It also provides information about certain qualifiers and recommends using Git or SVN for downloads. It mentions the availability of an official CLI and Codespaces feature. The article specifies that the software supports Linux distros and Pytorch version 2.1 or higher. It suggests changing the CUDA version for specific requirements and provides a link to learn more about Pytorch. The article also mentions an option to receive a sticker if a model is trained with the software. It provides statistics on time taken for a specific process and peak memory usage. It states that Windows is not currently supported and suggests trying to update pip if there are installation issues. Finally, it repeats the claims of being faster, using less memory, and supporting LLM finetuning.
Top 1 Comment Summary
The author of this text is questioning the possibility of a certain code performing faster than expected. They mention that in their own experience, the runtime is mostly dominated by large matrix multiplications in MLP layers, with a bit of attention. The code in question uses the same functions for MLP and attention as HuggingFace, but it seems to be much faster. The author notes that although there are a few Triton kernels present, there is no indication of Triton being used for MLP or attention, which are the main bottlenecks.
Top 2 Comment Summary
The article suggests ignoring criticism about pricing and recommends finding a sales representative or someone with experience in early stage DB companies. The suggestion is to start cold calling high-end customers with a large number of GPUs. The article advises that selling to B2B customers with deals worth 200-300k+ is the best strategy.
5. No new boss at NSA until it answers questions on buying location, browsing data
Total comment counts : 18
Summary
error
Top 1 Comment Summary
The article discusses the issue of Controlled Unclassified Information (CUI), which is a type of information that should be protected with access controls and not widely distributed. The article mentions that Senator Ron Wyden criticizes the labeling of CUI, stating that it is a “made up designation with no basis in law.” CUI was created by an executive order from President Obama in 2010. One of the main concerns highlighted in the article is that CUI can last indefinitely because it is an extra-legal construction and does not follow the mandatory declassification date like classified documents.
Top 2 Comment Summary
The article discusses the potential delay in the confirmation process for a nominee due to a request made by Senator Wyden. The National Security Agency (NSA) will either need to fulfill the request or Congress will have to hold a vote to push through the confirmation. While it may only cause a delay of a few weeks, it sheds light on the situation. The author suggests that preventing private companies from acquiring and selling private data would be the easiest and best solution, although it acknowledges the challenges of going against these companies.
6. Electrocaloric material makes refrigerant-free solid-state fridge scalable
Total comment counts : 11
Summary
The December 2023 issue of IEEE Spectrum introduces the concept of refrigerant-free refrigerators that are portable and efficient. Traditional coolers, such as refrigerators and air conditioners, use vapor compression which is proven but inefficient and relies on greenhouse gas-emitting hydrofluorocarbons (HFCs). However, there are alternative solid-state technologies that avoid these issues and can create miniature cooling devices. One such alternative is the electrocaloric effect, which involves using solid materials that change temperature under an electric field. Researchers have made significant progress in this area by developing a regenerative heat exchanger using a ceramic multilayer capacitor. By activating and deactivating the electric field, the material heats up and cools down, creating a temperature difference. The researchers achieved a record-breaking 20.9 °C difference between the hot and cold ends of the regenerator, along with a cooling power of 4.2 watts, which is a significant improvement over previous materials. However, further design iterations are needed before these electrocaloric coolers can be mass-produced. The current ceramic materials used are not suitable for mass production, and the researchers are also exploring the use of a more effective thermal conductor than the silicone oil they currently use. Additionally, the regenerator needs to be waterproofed to allow for the use of electrically conductive fluids like water, which would greatly increase cooling power. The article also mentions magnetocaloric materials as another technology that could be used for solid-state refrigerators.
Top 1 Comment Summary
The article discusses the use of a syringe pump to push silicone oil through a stack, which is considered an improvement over the current refrigerant cycle. However, it questions the claim that this is a “solid state” solution. The article also mentions that the current ceramics used in the process are not suitable for mass production due to their toxicity, high cost, and association with conflict materials. Therefore, the desired solution is not currently available with the materials being used.
7. New York will plant trees using new tech to maximize foliage impact
Total comment counts : 10
Summary
New York City’s city council has passed a measure calling for a 30% increase in tree canopy by 2035. A recent tree census found that the city has room for an additional 250,000 trees. The increase in trees is seen as a win for the environmental justice movement as cities worldwide aim to adapt to hotter temperatures caused by the climate crisis. However, determining the appropriate locations and tree types is more complex. A researcher at Cornell University has developed a 3D mapping tool called Tree Folio NYC, which uses lidar scans to provide a detailed visualization of each tree in the city. This allows researchers to understand the individual contributions of trees to specific areas and allocate resources accordingly. Expanding the city’s tree canopy has numerous benefits, including boosting immunity, reducing stress and blood pressure, improving mood and focus, and increasing energy levels. However, communities of color have significantly less tree canopy compared to majority-white areas, which contributes to the urban heat island effect. Unequal distribution of trees can have deadly consequences, with black residents being twice as likely to die from heat exposure as white residents. Historically, trees have been planted for aesthetics and economic development rather than maximizing their climate benefits. Strategic tree placement can enhance their impact and ensure more equitable distribution. This approach could also improve the safety of streets by allocating more pedestrian space on sunnier sides, protecting bike lanes, and investing in overall pedestrian comforts. New York City has passed urban forestry laws to create an urban forest plan and monitor progress using lidar data.
Top 1 Comment Summary
This article discusses how Portland is using tree cover to keep the roads cooler and reduce heat islands. This year, the city collected data that identified areas with reduced tree coverage, leading to the formation of “heat islands.” The article also mentions a similar study on urban heat island abatement, which can be found in the provided link.
Top 2 Comment Summary
The article emphasizes the importance of planting and maintaining trees as a beneficial use of city funds. It argues that such initiatives not only provide employment opportunities but also offer a cost-effective solution with significant positive impacts.
8. Review: A Dive into Mikrotik’s Weird SmartNIC (2022)
Total comment counts : 10
Summary
The article discusses Mikrotik’s CCR2004-1G-2XS-PCIe, a 25gbps NIC that combines a router into a PCIe NIC. The writer questions the need for routing and firewall capabilities in a NIC and suggests that it may be targeted towards small businesses. They mention the lack of reviews and availability of the product, as well as inconsistencies in the listed CPU specs. The writer proposes a potential use case for the NIC in single server co-location scenarios. They also provide information about the physical characteristics of the card and suggest a solution for PCIe initialization delay in Linux systems.
Top 1 Comment Summary
The author upgraded RouterOS from version 7.10 to 7.12 on a hAP ac3, which went smoothly. However, when attempting to upgrade on a CRS device, it wiped the entire configuration and the cloud backup was not restorable. In an attempt to downgrade to restore the 7.10 backup, the device became unusable and had to be fixed with NetInstall. The author suggests exporting the configuration to a text file in the future to avoid similar issues.
Top 2 Comment Summary
The article discusses the deployment of the CCR2004-1G-2XS PCIe, providing information and guidance on how to set it up.
9. Easy to use OpenID Connect client and server library written for Go
Total comment counts : 9
Summary
The article discusses an OpenID Connect client and server library written in Go. The library is certified by the OpenID Foundation and provides an easy-to-use implementation of the OpenID Connect standard. It supports both the client (RP) and server (OP) roles and is built using existing packages like OAuth2 for Go. The article mentions that there are limited options for OIDC libraries in Go, which is why this library was developed. It also mentions that other libraries like go-oidc and fosite were not chosen for specific reasons. The library is open source and free to use for everyone.
Top 1 Comment Summary
The author of the article shares their experience using Keycloak, an enterprise software, for their current platform development. While Keycloak has the necessary features such as SSO login, user registration, password resets, and social login, it required significant configuration and had some bugs. The author also mentions using mod_auth_openidc, an Apache module that simplified development by handling tasks like protecting endpoints and refreshing tokens. However, the author still expresses dissatisfaction due to the complexities and challenges of working with OpenID Connect, Keycloak, and configuring mod_auth_openidc. They mention moving permissions/roles back into the app’s database to avoid relying on the Keycloak REST API for every user permission change. Despite the difficulties, the author appreciates the availability of different software options and advises managing expectations when working with OpenID Connect.
Top 2 Comment Summary
The article discusses the misuse and complexity of OIDC (OpenID Connect) technology. While OIDC can be initially easy to integrate into an app, it significantly increases the complexity by requiring handling of redirects, callbacks, cookies, return URLs, logouts, profiles, and more. The author questions the idea of using a central OIDC solution as the primary login for an app, as it introduces many potential failure points and debugging challenges. The author shares that they removed OIDC as the primary login in their company and replaced it with a different solution, which resulted in fewer support tickets and a faster and more intuitive user experience. The author warns against nesting OIDC within OIDC and advises using OIDC only for social logins. They mention that solutions like Zitadel and Auth0 can become difficult to support and mature.
10. Why Did These LEDs Light Up? ESP32-S3 – Which Pins Are Safe to Use? [video]
Total comment counts : 7
Summary
error
Top 1 Comment Summary
The article discusses the frustration and time-consuming process of working with an ESP32-based board. The author shares their experience of debugging code, cutting traces, soldering bridge wires, and using a logic analyzer. They found out that some of the pins on the ESP32 are strapping pins, which configure settings like voltage for the flash memory. The author mentions that blowing certain fuses can hardcode some of the strapping pin settings. Additionally, the default behavior of certain pins may be set to JTAG, requiring explicit reset to use them as input/output pins. The lesson from this experience is to check for strapping and default pin behavior if there is an issue with a microcontroller.
Top 2 Comment Summary
The author finds it interesting that someone claims USB flashing cannot be trusted. They have been using the Samsung Galaxy S3 for over a year and have found it to be reliable, even when flashing it using OpenOCD inside a Docker container within WSL2 with usbipd.