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

F
Full Disclosure
WordPress大学
WordPress大学
小众软件
小众软件
Cloudbric
Cloudbric
AWS News Blog
AWS News Blog
腾讯CDC
量子位
人人都是产品经理
人人都是产品经理
大猫的无限游戏
大猫的无限游戏
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
V
Vulnerabilities – Threatpost
Scott Helme
Scott Helme
Hugging Face - Blog
Hugging Face - Blog
博客园_首页
C
CXSECURITY Database RSS Feed - CXSecurity.com
The Hacker News
The Hacker News
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
IT之家
IT之家
Jina AI
Jina AI
Attack and Defense Labs
Attack and Defense Labs
S
SegmentFault 最新的问题
Simon Willison's Weblog
Simon Willison's Weblog
The Cloudflare Blog
阮一峰的网络日志
阮一峰的网络日志
T
Tailwind CSS Blog
Last Week in AI
Last Week in AI
博客园 - 【当耐特】
Google Online Security Blog
Google Online Security Blog
美团技术团队
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
V
Visual Studio Blog
罗磊的独立博客
L
LINUX DO - 最新话题
博客园 - Franky
博客园 - 叶小钗
Apple Machine Learning Research
Apple Machine Learning Research
The Last Watchdog
The Last Watchdog
J
Java Code Geeks
AI
AI
C
Cisco Blogs
酷 壳 – CoolShell
酷 壳 – CoolShell
C
Cyber Attacks, Cyber Crime and Cyber Security
Cisco Talos Blog
Cisco Talos Blog
博客园 - 三生石上(FineUI控件)
雷峰网
雷峰网
Help Net Security
Help Net Security
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
云风的 BLOG
云风的 BLOG
I
Intezer
S
Securelist

Pierce Freeman

A browser for agents | Pierce Freeman The grey market of podcast appearances The way I travel | Pierce Freeman Fixing slow AWS uploads | Pierce Freeman Local tools should still use vaults We solved scratch content first Starting a podcast in 2025 Automating our home video imports Adding my parents to tailscale A deep dive on agent sandboxes Language servers for AI | Pierce Freeman My simple home podcast studio We need centralized infrastructure | Pierce Freeman Coercing agents to follow conventions using AST validation My unified theory of social selling My personal backup strategy | Pierce Freeman July updates to the homelab How the KV Cache works httpx is the right way to do web requests in Python Reputation is becoming everything | Pierce Freeman Building a (kind of) invisible mac app Updated knowledge in language models Making an ascii animation | Pierce Freeman How speculative decoding works | Pierce Freeman Under the hood of Claude Code Doing things because they're easy, not hard Speeding up sideeffects with JIT in mountaineer Firehot for hot reloading in Python Misadventures in Python hot reloading How text diffusion works | Pierce Freeman The tenacity of modern LLMs The ergonomics of rails | Pierce Freeman How language servers work | Pierce Freeman Just add eggs | Pierce Freeman Unfortunately SEO still matters | Pierce Freeman The futility of human-only web requirements Setting up Input Leap | Pierce Freeman Checking in on Waymo | Pierce Freeman The react revolution | Pierce Freeman Speeding up many small transfers to a unifi nas Quick notes on swift libraries AI engineering is a different animal San Francisco | Pierce Freeman Debugging a mountaineer rendering segfault Local network config on macOS Building our home network | Pierce Freeman Introducing Envelope.dev Legacy code and AI copilots Typehinting from day-zero | Pierce Freeman Generating database migrations with acyclic graphs Lofoten | Pierce Freeman Mountaineer v0.1: Webapps in Python and React Constraining LLM Outputs | Pierce Freeman Passthrough above all | Pierce Freeman Accuracy in kudos | Pierce Freeman How quick we are to adapt The curious case of LM repetition Costa Rica | Pierce Freeman Debugging chrome extensions with system-level logging Speeding up runpod | Pierce Freeman Inline footnotes with html templates Parsing Common Crawl in a day for $60 An era of rich CLI All or nothing with remote work The Next 10 Years | Pierce Freeman Adding wheels to flash-attention | Pierce Freeman LLMs as interdisciplinary agents | Pierce Freeman New Zealand | Pierce Freeman Representations in autoregressive models | Pierce Freeman Let's talk about Siri | Pierce Freeman Minimum viable public infrastructure | Pierce Freeman Reasoning vs. Memorization in LLMs Automatically migrate enums in alembic Greater sequence lengths will set us free On learning to ski | Pierce Freeman Dolomites | Pierce Freeman Using grpc with node and typescript Opportunity years | Pierce Freeman Buzzword peaks and valleys | Pierce Freeman Buenos Aires | Pierce Freeman Network routing interaction on MacOS Independent work: November recap Debugging slow pytorch training performance The provenance of copy and paste Debugging tips for neural network training Patagonia | Pierce Freeman Santiago | Pierce Freeman My 2022 digital travel kit AWS vs GCP - GPU Availability V2 Independent work: October recap | Pierce Freeman Planning Patagonia Relationship modeling | Pierce Freeman The power of status updates A new chapter | Pierce Freeman Give my library a coffee shop AWS vs GCP - GPU Availability V1 Switzerland | Pierce Freeman Headfull browsers beat headless | Pierce Freeman Webcrawling tradeoffs | Pierce Freeman Copenhagen | Pierce Freeman
Being late but still being early
2025-11-25 · via Pierce Freeman

I heard about Bitcoin for the first time in 2011. I was on a trip to Tahoe with my dad when highway patrol closed I-80. During the winter it's the only road connecting the Bay Area and the Sierra Nevada, so we had to spend the night in a motel in Auburn. We turned on the little TV to a news reporter losing their mind that this internet currency — totally made up, exclusively digital — had just crossed $5 in value. It was named Bitcoin, of all things. My dad shook his head and said something about Dutch tulips. It's currently at $88,000, down from $100k earlier this year.

Fast forward a few years to university in 2015. I became intimately familiar with Nvidia when we were using their chips for ML training jobs1. Split-adjusted, the stock was around $0.70, but it was trading at an unbelievably high P/E ratio. We all figured we were in a bubble and put our internship profits into the other "more diversified" FAANG stocks instead. The stock is up 200x since then.

I was objectively early. But I felt late both times. Everyone that I knew had already heard of them, which conventional wisdom would make you think that the market has already priced in the proper value.

In the frame of technical risk versus market risk, Nvidia and Bitcoin had to contend with both. In the early days they were tackling unsolved problems in game theory and matrix parallelization. Working with them on the ground, I knew they were succeeding. Which I probably should have taken as a good signal of potential upside, since it spoke to the de-risking of technical risk. And yet I hesitated. I didn't know how to price their current value. Is this price high or low? Will there actually be market adoption? It was the market risk that concerned me.

I suspect technical people are biased towards this failure mode. Our "missing out on bitcoin" stories are rarely about a lack of awareness. They're more often about our lack of conviction where tech meets capital markets. Interestingly, I don't see many of my financier friends share the same fear. They're quite open about having no idea about the tech, but they feel confident in the charts.

In little ways and big ways, we're always trying to do some estimate of where we're falling on the adoption curve. It influences the value of your stock options, obviously, but I also see it crop up in more subtle ways.

It's easy to feel like you're late to a trend when you're living in San Francisco. You probably aren't the first to hear about something. You might not even be in the first hundred thousand. But relative to market adoption, you might as well be the first one to know. Your reference group is distorted: you're comparing yourself to the small slice of people who heard before you instead of the 99.99% who haven't heard at all.

I can admit I don't have a native skill in accurately assessing future market conditions. To be honest, I don't think almost anyone does.2 But I would rather take the opportunity for asymmetric reward by understanding the tech, than by trying to figure out whether we were early or late.

Maybe that's a resolution going into the new year. Put some money where your adoption curve is.

  1. Back then TensorFlow was really the only computational graph game in town. ↩

  2. There's a reason why VCs' hit rate is usually only 1/100 within a fund's vintage. This also reminds me of the story of McKinsey trying to price the cellphone market size back in 1980. They guessed 900k by 2000. The answer was actually 109 million. Oops. ↩