New MacBook Airs and 13-Inch MacBook Pro Will Reportedly Feature M3 Chip (2 minute read)
Apple is reportedly set to equip its next-gen 13-inch and 15-inch MacBook Air and 13-inch MacBook Pro models with the M3 chip. The M3 chip boasts an 8-core CPU and will be built on TSMC's latest 3nm process. It features increases in both performance and power efficiency. Apple's upcoming WWDC 2023 in June is an ideal time for the reveal of the M3 chip and the first devices it will power.
|
Meta planning thousands of more cuts after widespread layoffs (2 minute read)
Meta could be set to lay off thousands of workers in a new round of job cuts starting this week. This is in addition to the 13% of employees let go in November as part of a cost-cutting initiative to remove non-performing or outdated projects and streamline decision-making. Mark Zuckerberg is taking responsibility for the new round of cuts, saying that layoffs were a last resort. The company is looking to reduce layers of middle management to speed up decision-making and maintain profitability.
|
Science & Futuristic Technology
|
Florida Startup Moves Closer to Building Data Centers on the Moon (3 minute read)
Lonestar Data Holdings, a cloud computing startup, has raised $5 million to build data centers on the lunar surface for secure data storage and edge processing. Storing data remotely on the Moon will alleviate energy consumption and digital pollution problems caused by Earth-based data centers. Lonestar's lunar data centers will initially be used for remote data storage and disaster recovery, as well as support for commercial and private exploration of the Moon. The company aims to launch a miniature data center box, which weighs about 2 pounds and has a capacity of 16 terabytes, to the lunar surface this year.
|
Emergent Abilities of Large Language Models (17 minute read)
This article explores the concept of emergence and its relationship with Large language models (LLMs) and AI. LLMs are probabilistic models that can learn and generate natural language patterns. As LLMs are scaled and trained on more data, their ability to learn these patterns improves as observed by quantitative metrics. However, the scaling process also unlocks new and unexpected qualitative behavior, such as the ability to perform arithmetic and summarize passages, which LLMs learn through natural language observation. These emergent abilities appear suddenly and unpredictably, leading to questions about their cause and significance.
|
Programming, Design & Data Science
|
How Discord Stores Trillions of Messages (15 minute read)
Discord published a blog post in 2017 chronicling the journey of migrating from MongoDB to Cassandra to store billions of messages. The company was looking for a database that could scale, be fault-tolerant, and have fewer maintenance needs. The Cassandra cluster faced significant performance issues that required more effort to maintain than improve. As a result, Discord changed its message storage system. This article discusses the problems with the previous system and the process of developing a storage system to handle trillions of messages.
|
Online gradient descent written in SQL (9 minute read)
Modern MLOps can be complicated. Bundling the logic into a database can make it easier. This article discusses the implementation of a machine learning algorithm in a relational database using SQL to remove the need for separate inference and training services. It begins by implementing simpler online algorithms before working towards more complex implementations like online gradient descent.
|
How Reddit is getting simpler — and dealing with TikTok, with chief product officer Pali Bhat (38 minute read)
This article features an interview with Pali Bhat, Reddit’s first-ever chief product officer. In the interview, Bhat discusses his focus on making Reddit simpler for new users to join and participate in interesting conversations, implementing changes to simplify the moderation process, rolling out a dedicated video feed to compete with other social networks, Google’s role in search, and the criticism aimed at Google's search quality. He also provides insight into what he learned from Google’s culture that he hopes to bring to Reddit’s product team.
|
Why the floppy disk just won’t die (9 minute read)
The world’s reliance on floppy disks continues to linger years after their manufacturers stopped producing them. Machines across industries, from medical equipment to aircraft, are still dependent on them. Many of the companies using them are small to medium businesses or those constrained by costs. The supply limitations will eventually force people to upgrade or replace their equipment. The disks once manufactured have a finite supply, and a portion of that supply is being consumed every day. No one knows the exact size of this inventory, but it is believed to be dispersed among many small collections.
|
Hamilton (GitHub Repo)
Hamilton is a general-purpose micro-framework for creating dataflows from python functions.
|
Bearer (GitHub Repo)
Bearer is a static application security testing tool that scans code and analyzes data flows to discover, filter, and prioritize security risks and vulnerabilities.
|
If your company is interested in reaching an audience of tech executives, decision-makers and engineers, you may want to advertise with us.
If you have any comments or feedback, just respond to this email!
Thanks for reading,
Dan
|
|
|
|