标签：#Science

Overpressure from the Blue Origin blast shattered windows at a hangar about a mile away from the pad.

Researchers have prototyped an AI-powered internet worm . The coolest thing about the prototype is that it carries its own LLM with it, and runs it on computers that have been broken into. This is the closest to John Brunner’s original 1975 conception of a computer worm that I’ve seen.

Introduction Excel is one of the most used tools for data analysis. It allows beginners like myself to easily clean, organize, analyze and visualize data.Excel enables users to work with large datasets and extract meaningful insights without requiring advanced technical skills. What is Excel Excel is a spreadsheet that allows you to collect, organize, analyze, calculate, and visualize data efficiently.Despite the emergence of other data analysis tools like SQL and Power BI, Excel remains one of the most widely used tools for both personal and professional data management.This can be credited to its ease of access, learning, and use. Ways Excel is used in real-world data analysis This week, I had the opportunity to explore how Excel is used in real-world data analysis.I discovered that Excel is not just a basic spreadsheet tool, but a powerful application that helps make sense of data and support decision-making. Data organization and cleaning Excel is used to structure raw data, remove duplicates, and fix errors. This improves data quality, making it easier to analyze and more reliable for decision-making.This improves data quality, making it easier to analyze and more reliable for decision-making. Financial Excel is commonly used in finance to create budgets, calculate profits and losses, and monitor expenses.It helps organizations keep accurate financial records and understand their financial situation. Business decision-making Businesses use Excel to track sales, compare performance over time, and identify trends.This helps managers understand what is working well and what needs improvement. Excel features and formulas In just a week, I have learned several Excel formulas that simplify data management and make working with data more efficient. SUM function The SUM function is used to add a range of values together in Excel, making it one of the most essential tools for quick calculations.It's used to automatically add a range of numerical values together, elimina

Have you ever trained a model that performed beautifully on your training data but fell apart the moment it saw new data? Or perhaps you built something so simple it couldn't even learn the training data properly? These are the classic traps of overfitting and underfitting — and every machine learning practitioner runs into them. In this article, we'll cover what they are, how to detect them, how to fix them, and where the bias-variance tradeoff ties it all together — with real-world examples and code throughout. What is Model Fitting? Model fitting is the process of training a predictive model on a dataset to find the optimal parameters that best capture the underlying patterns in the data. The goal is simple: the model should generalize well to unseen data — not just memorize the training examples. There are three possible outcomes when fitting a model: Outcome Description Good fit Captures underlying patterns, generalizes well Underfitting Too simple, misses patterns even in training data Overfitting Too complex, memorizes noise, fails on new data What is Underfitting? Underfitting occurs when a model is too simple to capture the underlying patterns in the data. It performs poorly on both the training set and on new, unseen data. Think of it like this: imagine asking a child to predict house prices and they only use the rule "all houses cost $100,000." That model ignores all relevant features (size, location, age) and will be wrong almost every time. Why Does Underfitting Occur? Model is too simple : A linear model trying to fit a curved, nonlinear relationship Too few features : Important variables are left out Too much regularization : Penalizing complexity so heavily that the model can't learn anything meaningful Insufficient training : The model hasn't been trained long enough Real-World Example Suppose you're predicting whether an email is spam. If you only use the feature "email length" and ignore word content, sender, and links, your model will underfit —

The USDA this week confirmed the first known infection of the carnivorous fly larva, which feast on the flesh of living mammals, after the United States eradicated the nightmare bugs in the 1960s.

Scientists in Finland found bees could solve an insect version of the classic "box-and-banana" problem.

“I think the team has really experienced the loss of a loved one with the end of the mission.”

Some of the AI industry's biggest rivals have put their many, many grievances aside for a common cause: making it harder for people to use their technology to develop biological weapons. In an open letter to US lawmakers, tech leaders are pressing Congress to enact rules closing what they say is an alarming biosecurity gap […]

Used Waymo batteries will bolster California and Texas energy storage projects.

Leading AI labs, executives, and scientists are sending a letter to lawmakers urging them to improve tracking of synthetic DNA sequences that could be used for bioweapons.

Some quantum computing companies we've covered have done recent progress updates.

When they're being eaten, bean plants release chemicals that draw in parasitic wasps.

By the time the Children's Hospital closed its doors to trans patients, Sage had already stopped taking testosterone. A nonbinary high school student, they originally received treatment for the rapid onset of puberty. The changes their body experienced felt frightening and sudden. They developed PMOS, a relatively common hormonal disorder that can lead to hair […]

Google, Microsoft, and other hyperscalers have come under scrutiny for their impact on water quality and availability.

CAP Theorem Explained: Choosing Between Consistency, Availability, and Partition Tolerance in Databases Imagine you're trying to book a flight online, and just as you're about to pay, the website crashes. When you try to book again, you find that the flight is now sold out, even though the website initially showed available seats. This frustrating experience is a classic example of a database trade-off between consistency, availability, and partition tolerance. The CAP theorem, first introduced by Eric Brewer in 2000, states that it's impossible for a distributed data store to simultaneously guarantee more than two out of these three principles. In this post, we'll delve into the world of CAP theorem, exploring its fundamentals, real-world database examples, and design implications. Introduction to CAP Theorem Understanding the Basics of CAP Theorem The CAP theorem is based on three primary principles: Consistency : Every read operation will see the most recent write or an error. Availability : Every request receives a response, without guarantee that it contains the most recent version of the information. Partition Tolerance : The system continues to function and make progress even when network partitions (i.e., splits or failures) occur. Importance of CAP Theorem in Distributed Systems In distributed systems, where data is spread across multiple nodes, the CAP theorem plays a crucial role in understanding the trade-offs between these principles. By grasping the CAP theorem, developers can design more resilient and scalable databases that meet the specific needs of their applications. Brief Overview of the Blog Post This post will explore the CAP theorem in depth, using real-world database examples to illustrate the trade-offs between consistency, availability, and partition tolerance. We'll discuss the fundamentals of CAP theorem, examine CA, CP, and AP systems, and provide guidance on designing for each combination. By the end of this post, you'll have a solid un

Two Hypotheses In the contemporary discussion about artificial intelligence, two distinct hypotheses intersect and are often conflated. The first hypothesis describes AI as a thin client between intention and result. Historically, a chain of translators existed between a concept and an artifact. A person formulated a task for a programmer, the programmer wrote code, the code became a program. A screenwriter passed an idea to a studio, the studio hired a VFX team, the team produced a film. A composer worked with musicians and a studio to record a track. AI shortens this chain, allowing a result to be obtained directly from a natural language prompt. The second hypothesis is more radical. It asserts that AI washes out not only performers but also apprentices. The main function of many professions was not the production of the current result, but the reproduction of knowledge. A junior was needed not because he is useful today, but because in five years he will become a senior. A student was needed not to create value now, but to become an engineer. A doctoral candidate was needed not for brilliant papers, but to undergo the school of scientific thinking. The Destruction of the Apprenticeship Mechanism The classical model of competence growth was built on review. A junior wrote code, a senior dissected it, extracted the substrate of experience, and transmitted professional intuition. Each review was an act of knowledge transfer. The new model looks different. A person formulates a prompt, AI generates the result. If code of acceptable quality appears immediately, the economic need for a junior declines. Along with it, the mechanism through which knowledge was transmitted disappears. A structural question arises that goes beyond the labor market. Where will the next seniors come from if the intermediate link does not undergo the path of learning through mistakes and reviews. This is a problem of competence reproduction, not simply automation. The Transformation of Educa

Hand the same paired before/after dataset (n = 25) to ChatGPT five times. Same prompt: "These are the same subjects measured before and after an intervention. Did their scores change significantly?" Four of the five runs return p = 0.009 from a paired t-test. The fifth run does a Shapiro–Wilk normality check on the differences first, decides they're non-normal, switches to a Wilcoxon signed-rank test, and reports p = 0.000018 . All five reach the same conclusion (significant). But notice what happened: only one run out of five thought to check an assumption you'd want it to check. The other four skipped it. The choice of method — and the test statistic, and the p-value — depended on whether the LLM happened to run an assumption check that time. On borderline data, this is the difference between reject and don't reject. If you're using LLMs for exploratory data analysis on a weekend project, you might shrug. If you're using them for anything that gets cited, gets submitted to a regulator, or gets handed to a clinician, this is a problem. It's a known problem — Cui & Alexander (2026) documented exactly this kind of method-divergence empirically; AIRepr (Zeng et al., 2025) shows the same thing across reproducibility metrics. The current answer in the literature is to constrain the agent so its execution is replayable. But replayability fixes "did we run the same code." It doesn't fix "did we run the right analysis." I've spent the last two months building a different fix. The more interesting half is the architecture. Let me walk through it. The real problem isn't temperature The first reflex is "set temperature=0 ." It's not enough. temperature=0 doesn't make a tool-using agent deterministic across runs. Three reasons: Inference isn't bitwise deterministic, even at temperature=0. Production LLM serving batches requests dynamically, and the attention kernels aren't batch-invariant — so the same input produces different output tokens depending on what other requests it

With shiftbloom studio. I build tools and projects about a variety of experimental approaches to real-world problems. The issue for such use-case often was how most small media systems start out: one big always-on recorder that keeps costing money even when nothing is happening. For live capture you obviously need to stay ready at all times — sometimes you can’t risk losing the first minutes. But for everything else it’s complete overkill. The Core Problem Backfills, VOD downloads, clip imports, repairs and re-encodes are queue work. They can wait a few seconds, run on burst capacity, or even on a regular VPS or laptop. They don’t need the same always-hot infrastructure as the live recorder. That’s why I split the system. Instead of one large monolith, I deployed: Observer cells — only for live streams (time-critical) Harvest cells — for all queue processing (can be delayed) The Three Roles 1. Mothership A small control-plane cron job. It checks queue sizes, currently live channels and running observer tasks, then decides: how many harvest cells should exist right now which channels need an observer cell It’s intentionally simple. The database remains the single source of truth. 2. Observer Cells Each observer cell records exactly one live channel. It receives its assignment through environment variables: +++env OBSERVER_VOD_ID OBSERVER_CHANNEL_ID OBSERVER_CHANNEL_LOGIN OBSERVER_CHANNEL_NAME +++ It starts recording immediately, writes HLS segments to object storage, sends heartbeats, and waits a short standby window after the stream goes offline. This window is important because streams sometimes drop and reconnect quickly. Without it you end up with many small broken VOD fragments. 3. Harvest Cells These handle all background work: downloading VODs, re-encoding, recovering broken files, etc. They can run anywhere Docker is available — AWS tasks, a small VPS, or even a spare laptop. They only need outbound access to Postgres and object storage. What Changed Previous

"It’s a reminder of how human activity is changing the natural world in unanticipated ways.”

Neanderthals had some wild stuff in their toolkits.