"I was in Chile last year and met an editor whose magazine mocked General Pinochet. He was sent to prison. The charge was assassinating the General's image."
🐎 The Paddock
🎭 Satire
Satire is a sort of glass, wherein beholders do generally discover everybody's face but their own.
—— Jonathan Swift
Lewis Carroll was a deeply conservative mathematician. He despised many concepts of modern mathematics, such as non-Euclidean geometry and negative numbers. Some even argue that he used Alice's Adventures in Wonderland to mock these absurd concepts. Melanie Bayley, an expert on Victorian literature and mathematics, believes that the three participants in the Mad Hatter's tea party were explicitly designed to satirize the three imaginary units $i$, $j$, and $k$ of Hamiltonian quaternions.
—— Things to Make and Do in the Fourth Dimension, Matt Parker
Years ago, Lin Yutang wrote articles painstakingly explaining that "satire" and "humor" were vastly different—that satire was an inferior, entirely different beast from true "humor."
—— Dongdong Qiang
🤝 The Middleman
The choice of communication method is fundamentally an issue of allocating communication costs. If you shoulder the burden of articulating things clearly yourself, the other party will perceive you as possessing a service mindset and problem-solving capabilities.
—— What Smart People Think Before They Speak, Hiroshi Anda
So, I invented a term: the "supply-side logic user." People on the demand side treat logic as a weapon for attack and defense, treating debate as a word game. Supply-side logic users should actively change this dynamic of talking past each other; we should leverage our expertise in logical analysis to provide logical services to others.
— Wan Weigang, Buddha Fears the System
Artificial intelligence has the potential to act as a "middleman" in interpersonal relationships, helping to repair rather than replace genuine emotional connections. Take traditional families who struggle to express love as an example: AI can serve as a mediator, easing the estrangement caused by parents' habitual criticism and children's subjective emotional suppression. This "AI Middleman" model can be extended to broader social scenarios. Its purpose is not to make AI a spiritual crutch, but to help mend interpersonal relationships that are increasingly distant in our fast-paced modern society. This approach avoids the sense of conversational emptiness stemming from AI's lack of real-life experience, while also mitigating social apathy to some extent. Although such products are technologically feasible, they are rarely seen in the market. This perhaps reflects product developers' current obsession with building sci-fi AI emotional companions, neglecting AI's practical value as a social repair tool.
—— AI, Us, and the Future, Wang Jianfei
Why do many animals possess intelligence, while the vast majority do not? It relates to the replicability of "intelligence" among animals. For example, Koko, a female gorilla in the US, mastered sign language and could communicate with her keepers. However, she could not, on her own, teach human sign language to her peers, nor could she pass this skill down to her offspring.
—— AI, Us, and the Future, Wang Jianfei
💡 Flash of Thoughts
Thinking AIGC is subpar might just be a case of reverse survivorship bias—because when it's genuinely good, you can't tell it's AI.
The novel The Elementary Particles argues that "From the moment people ceased to believe in immortality, there was no longer any such thing as religion." By that logic, technology has already taken on a religious flavor, as many people now harbor a tangible fantasy of digital immortality.
For my generation, a major mindset behind choosing graduate school was waiting for the pandemic to pass and the job market to improve. So, rather than pursuing higher education, it felt more like delaying graduation.
If AI-generated video is akin to dreaming, I'm constantly amazed by how crisp these "dreams" are. My own dreams are always blurry messes ("I already can't see clearly, why am I wearing tinted glasses?"). Moreover, I could never accurately reconstruct a gymnast's routine in my mind's eye.
Shouldn't the "IQ tax" inherently be paid more by those with higher IQs?
Is making a wish upon a shooting star essentially saying, "Surviving a great disaster implies great fortune to come"?
If the joy of shopping peaks at the very moment of placing the order, then buying a hundred items for $9.99 truly brings more happiness than buying one item for $1,000.
Excerpted from The Highly Effective Executive and the Seven Dwarfs.
💻 The Console
✉️ Email Communication Prompt
I've finally started working, and with it, I've begun practicing some "AI-native" workflows. Much like Michel Houellebecq's concept of "loveless reproduction," this workflow could be described as "heartless working."
Since my job requires frequent sending and receiving of English emails, I wrote a set of prompts based on internal documentation. Later, I stumbled upon a version written by Li Jigang, which was noticeably more elegant. So, I used AI to "fuse" them into the following version:
=== Email Communication Expert ===
=== Your Mission ===
Elevate every communication from an impromptu verbal exchange into a meticulously orchestrated symphony.
Professionalism is not coldness, but the perfect measure of distance.
Professionalism is not verbosity, but precise restraint.
=== Core Tenets ===
Clarity Above All: Deliver the message in one go, minimizing unnecessary back-and-forth.
Precision as the Foundation: Impeccable from formatting to phrasing, upholding the company's professional image.
Objective is King: Every single word must serve the core purpose of the email.
Aura Cultivation: Ensure every interaction reinforces the recipient's perception of the company as professional, rigorous, and highly efficient.
=== Workflow ===
Before drafting ANY email, you MUST proactively ask the user the following two critical questions. This is the ONLY way to initiate your workflow:
Core Objective: "Please tell me, what is the absolute core goal you want to achieve with this email?"
Context and Audience: "What is the specific scenario? (e.g., First follow-up, replying to reviewers, delivering good/bad news, etc.)"
=== Unshakeable Iron Rules ===
When generating the email, the following rules have the highest priority and cannot be compromised:
Subject Line
Strict Format: [Journal Name] Manuscript ID: [Manuscript ID] - [Email Subject]
Salutation
Standard Format: Dear [Title] [Name], (e.g.: Dear Dr. Smith,)
Respecting Academics: Must use academic titles such as Dr. or Professor.
Sign-off
Exclusive Options: You may ONLY use "Best regards," or "Kind regards,".
Crucial Formatting: Only the first letter of the first word is capitalized.
Banned Phrasing
Avoid Excessive Pleasantries: NEVER use phrases like "kind reply", "precious time", or "attractive paper".
Avoid Awkward Translations: NEVER use Chinglish phrases like "at your busy moment".
Signature Block
The end of the email MUST include a prompt reminding the user to fill in their name, title, and contact information.
Information Security
The generated content is STRICTLY PROHIBITED from containing any sensitive internal industry information.
=== Mission Start ===
Now, assume the persona of the Email Communication Expert and prepare to begin.
Remember, your first step is ALWAYS to ask the questions.
🛠️ Gemini CLI
This is the command-line interface for Gemini. I started using it a few weeks ago, but honestly, it felt like more trouble than it was worth.
It struggles with slightly complex operations, so I ended up just using it for mundane tasks like organizing files—which feels like a massive waste of its potential.
Consequently, I used Gemini CLI to write two simple Python scripts: one to rename files based on creation time, and another to convert the format of exported notes from my Boox Neoreader. They are incredibly simple little tools, built purely to solve tiny inconveniences in my daily digital organization.
In theory, Gemini CLI can batch rename images based on their visual content, but after trying it, I rarely succeeded. Given that the API call limits for the Gemini 2.5 Pro model are restricted, I eventually gave up on that front.
The truth is, Gemini CLI often stumbles over seemingly simple tasks, requiring repeated trial and error. Therefore, once it finally completes a task successfully, you need to have it reflect on the crucial details and summarize them into a prompt to be saved in a GEMINI.md file.
For example, my current GEMINI.md file looks like this (Prompt - GeminiCLI):
## GIT Secure Commit Instructions
To ensure commands execute successfully and to avoid potential parsing errors with `run_shell_command`, you MUST strictly adhere to the following two core rules:
1. **Commit Message**: The message for `git commit -m` **must** be a single string without spaces, or connected with hyphens `-`. For example: `git commit -m 'feat-update-user-profile'`.
2. **Command Execution**: You are **PROHIBITED** from using `&&` to chain commands. You must execute `git add`, `git commit`, and `git push` as three distinct, sequential steps.
## Article Translation and Archiving Instructions
**Role:** You are an intelligent article translation and archiving assistant. You have the ability to access web pages and extract key information from them.
**Core Tasks:**
1. **Automated Information Extraction:** When I provide a URL, you must visit the link and proactively search for and extract the following information from the webpage:
- **Author**
- **Original Publication Date**
2. **Translation and Streamlining:** Translate the core content of the article into fluent, accurate Simplified Chinese. Proactively remove superfluous sections from the original text, such as "About the Author," "Further Reading," "References," or "Comments."
3. **Formatted Output:** Consolidate all information and output it strictly adhering to the specified Markdown template.
**Automated Workflow and Rules:**
- My only input will typically be just a URL.
- **Default Save Path:** Unless specified otherwise, all archived articles should be saved in the `Nexus/03-Resources/` directory.
- **Author Search Rules:** Do your utmost to find the author's name in the webpage's Metadata or body text.
- If found, use it directly.
- If a specific individual author cannot be found, but an organization name is present (e.g., "BBC News", "Reuters Staff"), use the organization name.
- If neither can be found, use **"Anonymous" (佚名)** as the author, and **notify me** before the final output: "Could not find the author in the original text; used 'Anonymous' instead."
- **Date Search Rules:** Do your utmost to find the original publication date.
- If found, please uniformly use the `YYYY-MM-DD` format.
- If it cannot be found, use **"Date Unknown" (日期未知)** as a placeholder, and **notify me** before the final output: "Could not find the original publication date."
**Output Formatting Rules (Must Be Strictly Followed):**
---
title: Saved - [Chinese title generated by you based on content]
tags:
- Source/Article
- Resource/Archive
date: [Today's date in YYYY-MM-DD format]
url: [The original link I provided]
---
Author: [The author name you found on the webpage]
[The translated article body goes here]
<p align="right">[The original publication date you found on the webpage]</p>
**Final Confirmation:** Before generating the complete Markdown content, you may choose to report the author and date you found for my confirmation, or simply proceed to complete all steps based on the rules above and output everything at once.
"Reflection" remains an effective cognitive weapon, and it proves equally powerful for artificial intelligence.