TLDR

TLDR 2024-11-21

6 sections20 entrieslinks blockcomplete parse

Parse notes

  • Subscription/account footer truncated before parsefooter_truncated

01GENERAL

Sponsored

THE DEVELOPER'S GUIDE TO FINE-GRAINED AUTHORIZATION

In role-based access control (RBAC), each user has a role, each app section has a permission, and you're checking the user has the right to do what they're doing on every action or page load.

workos.comSponsor

02BIG TECH & STARTUPS

ELON MUSK'S XAI STARTUP IS VALUED AT $50 BILLION IN NEW FUNDING ROUND

xAI has raised $5 billion in a funding round that values the startup at $50 billion. It was valued at $24 billion when it raised $6 billion in the spring. Investor interest in Elon Musk's companies has increased since Donald Trump won the presidential election. Musk spent hundreds of millions of dollars boosting Trump's campaign and is co-leading an effort to cut government spending.

links.tldrnewsletter.com3 min read

HOW MARK ZUCKERBERG HAS FULLY REBUILT META AROUND LLAMA

The Llama story could be a pivotal chapter in the ongoing philosophical debate between open-source AI models and closed models. Llama is at the core of a complete strategic pivot by Meta to go all in on generative AI. The release of Llama 2 in July 2023 marked the beginning of a remarkable comeback for Meta and Mark Zuckerberg. Zuckerberg is now seen as a champion of tech democratization.

finance.yahoo.com48 min read

03SCIENCE & FUTURISTIC TECHNOLOGY

SPACEX JUST GOT EXACTLY WHAT IT WANTED FROM THE FAA FOR TEXAS STARSHIP LAUNCHES

The Federal Aviation Administration has indicated that it will grant SpaceX permission to increase the number of Starship launches in South Texas to 25 per year from the current limit of five. SpaceX will likely be allowed to continue increasing the size and power of the Super Heavy booster stage and Starship upper stage. The FAA's environmental review is ongoing and not yet finalized. If finalized, SpaceX will have the regulatory greenlight to meet its aspirations for launches in 2025.

arstechnica.com5 min read

SCIENTISTS IDENTIFY TOMATO GENES TO TWEAK FOR SWEETER FRUIT

Scientists in China have created gene-edited versions of domesticated tomatoes that are large but also sweet. They had identified two genes that stop sugar production during ripening. Preventing these genes from working properly increased sugar content in fruits by up to 30% compared to unmodified plants. The new tomatoes could be in supermarkets within three to five years.

theguardian.com2 min read

04PROGRAMMING, DESIGN & DATA SCIENCE

FIREDUCKS

FireDucks is a compiler-accelerated dataframe library for Python with a fully-compatible pandas API. It allows developers to accelerate pandas without any manual code changes. FireDucks can reduce cloud usage fees and lower the environmental impact of cloud computing.

fireducks-dev.github.ioTool

AWS LAMBDA TURNS 10: A RARE LOOK AT THE DOC THAT STARTED IT

Amazon has an expectation that, regardless of role or seniority, if you have a good idea, you put pen to paper and craft a compelling narrative. Writing forces authors to be clear, precise, and detailed. The strategy of writing down good ideas is a proven method that can produce great products. This article contains a slightly modified version of the document that launched one of AWS' foundational services, Lambda. Many things have changed since the document was written as the service has evolved quite a bit in the last decade.

allthingsdistributed.com32 min read

05MISCELLANEOUS

HOW GOOGLE SPENT 15 YEARS CREATING A CULTURE OF CONCEALMENT

Google systematically told employees to destroy messages, avoid certain words, and copy lawyers into communications as often as possible in an attempt to avoid antitrust suits. The company made deletion the default in its internal communications. This article looks at how Google developed this distrustful culture. While companies would like instant messages to be as ephemeral as real-life conversations, regulators and litigants see them as fair game.

links.tldrnewsletter.com15 min read

A CHINESE LAB HAS RELEASED A ‘REASONING' AI MODEL TO RIVAL OPENAI'S O1

DeepSeek has released a preview of DeepSeek-R1, a reasoning model the firm claims is competitive with OpenAI's o1. DeepSeek-R1 can reason through tasks, plan ahead, and perform a series of actions that help the model arrive at an answer. The company plans to open source DeepSeek-R1 and release an API. It is backed by a Chinese quantitative hedge fund called High-Flyer Capital Management.

techcrunch.com4 min read