WHY USHER IS PERFORMING AT THE SUPER BOWL FOR FREE
A breakdown of how and why some of the world’s most famous artists perform at the Super Bowl for $0.
Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
A breakdown of how and why some of the world’s most famous artists perform at the Super Bowl for $0.
Sponsored
POST/CON 24 [5] is the place to be if you’re looking to hear from industry leaders, gain new skills, or engage with the global API community. And there’s a 50% discount if you sign up early [5]!
This post walks through how to build a search engine from scratch using Python. The project was motivated by the desire to address the "Small Website Discoverability Crisis" and make small websites more visible. The post walks through the key components of the search engine, including the crawler, inverted index, ranker, and interface, providing both theoretical descriptions and practical implementation details. The project is helpful for deepening one’s understanding of search engine mechanisms.
pg_analytics is a Postgres extension that significantly improves the analytical performance of Postgres databases by up to 94 times compared to traditional methods. It addresses the challenges faced by developers dealing with slow query times and poor data compression, offering a drop-in solution within Postgres itself. This post explains the technical workings of pg_analytics, explaining the introduction of deltalake tables and integration with Apache Arrow and Apache DataFusion.
We traditionally use a "public" folder to serve files in Node.js applications. However, this only works for static assets and doesn’t work for files like profile pictures, private files, or generated files - these files are served from an API and loaded into memory. This isn’t scalable when there are a lot of users. Instead, it’s useful to use a stream to return large files in chunks. This article looks at a couple of ways to achieve this using both Next.js API routes and also the more generic Route handlers.
Many developers have experienced “flow state”: your ideas are flowing, time and space fade away, and the path ahead of you is clear. Achieving flow state is not only great for productivity, but also enhances dopamine and endorphins, resulting in an increased feeling of motivation, thinking ability, and general well-being. To better reach flow state, it’s helpful to optimize your environment, map out your work ahead of time, and find joy in the work that you’re doing.
Once your app gets into an undesirable state, the best thing you can do is reset it and start from a fresh, well-known, and correct state. Erlang’s "let it crash" philosophy suggests resetting parts of an application upon failure. While effective, it requires developers to manually track and restore application states. Another approach, "durable execution," allows computation to run uninterrupted by storing minimal data for state reconstruction, making local variables resilient even through restarts.
The phrase "wherever you find podcasts” highlights a stark contrast between podcasting and closed media platforms like YouTube or TikTok. Podcasting's open format empowers creators to control their content and audience relationships, enabling portability and resisting surveillance-based advertising. There are challenges in podcast economics, but podcasting is a beacon for the open web's renaissance and a model for reclaiming control in online spaces. Podcasting is similar to RSS feeds for blogs and newsletters.
SQLite Web is a web-based SQLite database browser written in Python. It lets developers add, drop, or edit tables, columns, rows, and indexes. SQLite Web works with both existing SQLite databases or can be used to create new ones. It allows for the import and export of both JSON and CSV files.
Concrete.css is a simple, classless CSS framework. It offers styling for typography, sections, buttons, images, lists, forms, tables, blockquotes, and progress bars.
Meow is a CLI app helper. It can parse arguments, convert flags to camel case, negate flags with “--no-” prefixes, and output helpful descriptions and help text. Meow has no dependencies.
While working at startups and moving fast, it’s important to use existing tools before adding new ones to avoid complexity. Software often changes fast at startups, so optimize software for change and focus on iteration, testing, and implementing a modular monolith approach. Store data in databases for easy querying, utilize decision logs, and employ continuous deployment practices for efficient software delivery.
The ideal size for small pull requests (PRs) is around 50 lines long. Smaller PRs are reviewed and merged faster, less likely to be reverted, and attract more review comments per line changed. Data from various repositories suggests that PRs between 25-100 lines offer optimal benefits across speed, review quality, and total code output. Extremely small PRs (under 25 lines) tend to have a higher revert rate, while very large PRs (over 10,000 lines) may receive less engagement and feedback.
Google Cloud’s localllm enables developers to utilize LLMs on local CPUs without GPUs. Developers can use quantized models and Cloud Workstations for cheaper LLM usage and better privacy controls compared to online models like GPT-4.
Git has an auto-correct flag that replaces typos with a suggested corrected command.
React components for resizable panels, supporting horizontal, vertical, and persistent layouts.
jQuery 4.0 beta drops support for IE10 and older, removes deprecated APIs, and is now importable with ES modules.
The latest ChatGPT-4 system prompt can be used to help you understand and prompt the LLM better.
to set and meet improvement goals that are backed by data.
Sponsored
Efficiency is still top of mind for most companies, and software teams are no exception. Engineering leaders need to know — and be able to show — that their teams are delivering operational excellence and driving business impact. Download this guide [5] to learn how to leverage metrics program data
WhatsApp has been working on an initiative for two years that will allow its users to send messages across different messaging applications using end-to-end encrypted methods. The initial phase of the interoperability rollout will enable basic functionalities such as text, image, voice message, video, and file sharing between individuals. Advanced features such as calls and group chats are expected to be introduced later. The feature will be opt-in. Third-party apps will have to encrypt their messages using Signal's encryption protocol and meet specific technical standards.
WeWork co-founder Adam Neumann, who stepped down in 2019, is attempting to buy the company out of bankruptcy. WeWork was once valued at $47 billion. It filed for bankruptcy last November, listing debts of over $18.6 billion at the time of filing. Neumann's current venture, Flow, is a residential real estate outfit focused on rentals. Flow received a $350 million investment from a16z in August 2022.
Bluesky is removing its invite system and opening its doors to anyone who wants to sign up. It will start letting outside developers host their own servers on its underlying AT Protocol later this month. Bluesky users will be able to opt into experiences that aren't run by Bluesky and bring their profiles with them to rival apps on the network. The company plans to make money by charging users for additional features in its app, taking a cut of purchases for things like custom feeds (which developers will be able to charge for), and through a service that will help others manage servers on the AT Protocol.
A series of papers detailing the US National Ignition Facility's 2022 achievement of a break-even fusion reaction have passed peer review. In the breakthrough experiment of December 2022, lasers fired 2.05 megajoules of energy into a fuel source, which resulted in 3.15 MJ being released. Researchers have fired more energetic lasers and produced even more energy in several experiments since. Links to the five peer-reviewed papers are available at the end of the article.
Mojo CSS is a framework for crafting UIs without shipping any CSS. It features multiple color themes, a small package size, compatibility with a wide range of technologies, and more. Developers can use Mojo CSS to build custom design systems. A short introduction video is available.
GitHub releases data on developer contributions quarterly as part of its Innovation Graph project. The most recent batch of data, released on January 18, shows a jump in the number of developers from Bangladesh, Nigeria, and Pakistan. The data only measures registered accounts rather than code commits, so it's heavily skewed to less active contributors.
visionOS suffers from several UI shortcomings. It is difficult to manage multiple windows on the screen and its eye tracking is not nearly as precise as it needs to be for productive window management. The Vision Pro's field of view is very narrow, making it difficult to have more than two or so apps open without encountering z-axis issues. These problems result in the device not being a great productivity tool - better hardware may not change this reality.
GFiber (formerly Google Fiber) is planning to raise external capital to expand. The company is currently entirely funded by Alphabet and customer revenue. GFiber is aiming to be independent from Alphabet, similarly to how Waymo and Verily have done over the years. External capital will enable GFiber to scale its technical leadership, expand its reach, and provide better internet access to more communities. The internet service provider is operating or has announced availability in 18 US states.
BuildKit takes configuration files and converts them into built artifacts using parallelizing build steps whenever possible.
Ahmad Abbas, who was previously a hardware engineer manager on Apple’s Vision Pro team, is now Midjourney's head of hardware.
The cost of writing tests should be lower than the cost of bugs.
Amazon bought PillPack in 2018 as an entry point into the online pharmacy market and acquired One Medical in 2022 for roughly $3.9 billion.
While the Metaverse is usually referred to as something we may one day reach, spatial computing has been here since Snow Crash was published.
Software offers humans a chance to create something magical, but this opportunity is squandered in a doomed quest for endless growth.
OpenAI is implementing watermarks in images generated by DALL-E 3, following standards from the Coalition for Content Provenance and Authenticity (C2PA), to include invisible metadata and a visible CR symbol for provenance verification.
Microsoft's AI-driven Copilot for Sales and Service is now generally available. It integrates with CRM systems like Salesforce to enhance productivity for sales and service professionals. The tools automate repetitive tasks and provide insights directly within Microsoft 365 apps, aiming to improve customer interactions and streamline business workflows. Early adopters like Avanade report significant time savings and better customer engagement using these AI capabilities.
The removal of large-scale datasets such as LAION, along with valid copyright concerns, makes training large-scale image models challenging. This work shows that you can train a powerful CLIP model with 30 million synthetically generated images.
This research explores how to make powerful language models more suitable for mobile devices by focusing on tiny models with fewer parameters.
Addressing the critical issue of hallucinations in Multimodal Large Language Models (MLLMs), researchers have developed MHaluBench, a new benchmark for evaluating hallucination detection methods.
Bria AI has open sourced (non-commercial) its background removal model and tools, including training data mixture and performance metrics. It is a significant improvement over alternative open models.
InteractiveVideo introduces a novel way of making videos where users can dynamically interact with the creation process. Unlike traditional methods, this user-friendly framework allows for real-time modifications using text, images, painting, and even drag-and-drop.
DeepSeekMath 7B is a language model specifically designed to tackle complex mathematical problems. It's been trained on a massive amount of math-focused content.
The new Meta Voice model is a small and powerful text-to-speech model that supports generation and voice cloning.
Advanced AIs like GPT-5 could have significant impacts across markets, real estate, labor, energy, national economies, geopolitics, and the economy's inflationary dynamics, leading to a reevaluation of value accrual, global power shifts, and changes in employment and economic structures.
As AI tooling and business use cases become more sophisticated, we’ll start to see more and more real-world applications of the tech. One lens to look at the opportunity here is how AI will make processes cheaper and more efficient whilst growing margins substantially. This might unlock a whole new way of doing things that previously didn’t make economic sense because margins were too thin. This article explores a few of these examples.