惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

D
Docker
Simon Willison's Weblog
Simon Willison's Weblog
H
Help Net Security
F
Fortinet All Blogs
H
Heimdal Security Blog
S
Schneier on Security
L
LangChain Blog
博客园 - Franky
酷 壳 – CoolShell
酷 壳 – CoolShell
NISL@THU
NISL@THU
P
Palo Alto Networks Blog
J
Java Code Geeks
博客园 - 【当耐特】
The Last Watchdog
The Last Watchdog
W
WeLiveSecurity
www.infosecurity-magazine.com
www.infosecurity-magazine.com
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
V
Vulnerabilities – Threatpost
I
InfoQ
Recorded Future
Recorded Future
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
C
CERT Recently Published Vulnerability Notes
T
Tenable Blog
腾讯CDC
C
Check Point Blog
量子位
M
MIT News - Artificial intelligence
GbyAI
GbyAI
罗磊的独立博客
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
B
Blog
小众软件
小众软件
cs.AI updates on arXiv.org
cs.AI updates on arXiv.org
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
C
CXSECURITY Database RSS Feed - CXSecurity.com
Stack Overflow Blog
Stack Overflow Blog
P
Proofpoint News Feed
P
Privacy & Cybersecurity Law Blog
V2EX - 技术
V2EX - 技术
T
Threatpost
Engineering at Meta
Engineering at Meta
Attack and Defense Labs
Attack and Defense Labs
T
Tailwind CSS Blog
S
Securelist
The Cloudflare Blog
博客园 - 叶小钗
L
LINUX DO - 最新话题
T
Troy Hunt's Blog
C
Cyber Attacks, Cyber Crime and Cyber Security
爱范儿
爱范儿

est の 输入 输出和出入

gitweets改版,复刻微信「朋友圈」 MiMoCode 干完活儿发通知 写作能力和 locate cost grep vs sqlite 谁更适合微信聊天记录? [AI] curl -NT. 导致100% CPU原因 或许「数羊」真是个有效的入睡法 唯物主义「天命」论 我的 Vibe Coding 最佳实践——ADR文档 MacOS 快速插入当前时间 locate cost 基于 git 的零拷贝静态web服务器 AI和柜台费 Sutton 论 discovery Elon Musk 五步工作法 Playlet:DLNA听歌神器 免安装app播放NAS里的歌 不修改nginx接收websocket AI 硬伤 理解LLM的范式——它就是个差分机? 路径依赖 AI 流式接口的pattern FSRS核心字段 SVG 时钟 AI的自我觉醒是必然 唐宋之变 Louis Alexander 谈英语学习 Predict Next Word Instruct 的意义 Sentry 替代品 Bugsink 安装配置 浏览器通过WebGPU上做AI推理 Indent Is All You Need 语文学习和考试 失落的学习兴趣 如何跟孩子讲学习的意义 物权 Bonsai 在 M2 安装 首页和404更新 无用之用——或许LLM真的还不是AGI 发明后训练的人真是天才 看好 Taalas 码奸 claw会代替员工? 如果拿 token 抵工资会发生什么? tmux enables AIs to operate servers safely chat.est.im launches at 3.14 2026 股灾? AI 弑父 Configurable Intelligence 巴别塔,Vibe Coding和Lisp之咒 Citrini的2028智力危机 中文不是唯一的意译语言 观星有感 体细胞阶级论——当繁衍权被剥夺
The Porsche Diffusion
2026-03-29 · via est の 输入 输出和出入

Social media and smartphones are sterializing humanity.

There is an absurd claim I saw the other day. It's a popular call out "The Porsche diffusion" on Chinese interwebs. It goes like this

If one woman dates a guy who drives a Porsche, she’s unlikely to settle for less afterward.
Then nine of her BFFs think, “If she can get a Porsche guy, why can’t I?”
Now you have ten women who won’t consider non-Porsche men.
That’s the Porsche Diffusion Law.

I initially dismissed this as a blatantly misogynistic take on hypergamy. It felt like one of those cynical internet takes trying to reduce complex human behavior into a cheap punchline. Then someone did a bit of napkin math:

Imagine a mega city with 10 million people. Say there are about 20,000 people who can visibly signal high status - not just wealth, but performative wealth. Think luxury cars, curated lifestyles, Instagram-ready relationships. Call them “Porsche guy”

That’s 0.2% of the population.

The question isn’t “how many Porsche guys exist from commoner's perception?”

To translate into Math: what’s the probability you’ve seen at least one Porsche guy around you?

Consider a typical social graph. Between friends, coworkers, friends-of-friends, and social media exposure, it’s not unreasonable for someone to be indirectly exposed to a few hundred distinct individuals.

Let’s say 300.

What’s the probability that none of those 300 people are connected (directly or indirectly) to someone in that top 0.2%?

Roughly:

(1 − 0.002)³⁰⁰ ≈ 55%

So there’s about a 45% chance you will encounter at least one “Porsche-level” signal within your immediate social horizon.

Now add recommender algorithms.

You are no longer sampling randomly from 300 people. You are sampling from a biased, algorithmically amplified feed - one that disproportionately surfaces high-status, high-engagement content.

Your effective exposure probability is no longer 45%. It’s closer to saturation.

At that point, the system changes character.


I actually went a bit further into this topic. Checkout my post few weeks back in Chinese.

Thinking about this made me realize something a bit unsettling.

In early development of a human body, every cell starts out basically identical. Same DNA, same potential. Then some local chemicals kick in, and only a tiny fraction become germ cells. The rest become somatic-functional, necessary, but no longer part of reproduction.

No cell is forced. It’s just the gradients.

The system ends up with massive scale, high efficiency-and very few cells actually reproducing.

If you squint, the pattern doesn’t feel entirely alien.


I wrote this blog because of an HN thread. Praise the unholy AI trinity of 搜广推 business (Another Chinese connotation which stands for Search, Ads, and Recommendation engines in case you are wondering).