Joined August 2021
- Tweets 87
- Following 55
- Followers 58
- Likes 47
2 Photos and videos
Pinned Tweet
May 25
I keep watching people use AI. The ones who get ahead aren't the ones with the best prompts.
They're the ones who spotted their own blind spots before the market did.
Three patterns I keep seeing:
1. They use AI to amplify their strengths,not fix their weaknesses.
2. They ask "what am I optimizing for?" before every prompt.
3. They treat AI like a thinking partner, not a replacement for thinking.
The gap between those who benefit from AI and those who don't isn't technical.
It's psychological.
AI is easy. Knowing yourself? That's the edge.
1
1
53
Jun 13
我常常在想這件事。地球上永遠都有問題，也永遠都有我們希望變得更好的地方，有我們想在地球上解決的事，而我們也確實應該去解決它們。 但同時，也必須有一些事情，能讓你對未來感到興奮；讓你每天早上醒來時覺得開心，因為你迫不及待想知道接下來會發生什麼。 這就是 SpaceX 想帶給你的未來。
28
Jun 12
GitHub Trending This Week (Jun 5-11)
5 projects that actually matter:
This is where RAG systems will be won or lost.
1. turbovec ( 1.6K ⭐ in 7 days)Rust-powered vector index with Python bindings.Vector search just entered the "speed matters" era.The knowledge curse is about to break.
2. Understand-Anything ( 2K ⭐)Code, docs, papers, video—one tool to parse them all.Not searching. Understanding.
Don't bet on one horse? ECC is building the saddle.
3. ECC (affaan-m/ECC) ( 716 ⭐)Runs Claude Code, Codex, Opencode, Cursor simultaneously.The orchestration layer between AI coding tools.
Learning AI = the productivity unlock.
4. open-notebook ( 554 ⭐)Open-source Notebook LM reimplementation.When open-source clones a closed tool, the ceiling becomes the floor.
Real-time intelligence > historical knowledge.
5. last30days-skill ( 1.1K ⭐)Forces AI to only use data from the last 30 days.Kills hallucination at the source.
From "general skills" → "domain knowledge."Signal of the week:Skills ecosystem is going vertical.
Every Skill = a knowledge crystallization layer.
Who builds the Skills, owns the ecosystem.AI agents are splitting into specialized roles.
#OpenSource #GitHub #AIAgents #DevTools
30
Jun 12
Carnegie Mellon Unanimous AI: 25-30 humans connected by AI agents hit 62% accuracy on NBA predictions, beating Polymarket's 55%.
The future isn't AI replacing humans. It's AI amplifying human judgment. Hybrid > pure.
1
13
Jun 12
FIFA opened Football AI Pro to all 48 teams — Cape Verde gets the same analytics as Spain. Information parity achieved.
But here's the twist: more fairness = more unpredictability. When everyone has the same data, nobody has an edge. The game gets harder to predict, not easier.
That's the real AI paradox.
32
Jun 12
6 AI models, 1 consensus: Spain wins the World Cup. But history says otherwise — Al Jazeera AI hit 58.7% in 2022, basically a coin flip. The real evolution isn't accuracy. It's learning to say "I don't know."
From Paul the Octopus to 300 AI Agents: the prophecy paradox 🐙🤖
1
28
Jun 12
Kimi deployed 300 Agents to predict every World Cup match — then did something rare in AI: wrote confidence intervals in the main text. High: 85-90%. Medium: 55-65%. Low: ~random.
Honest uncertainty > false certainty. This is what AI maturity looks like.
1
8
Jun 10
The real safe haven isn't gold.
It's a clear mind that doesn't panic when headlines scream.
Always in style.Hardest asset to accumulate.Zero correlation to geopolitics.
1
Jun 7
【三层锚定陷阱】
第一层：$1.77万亿本身。
听到这个数，大脑自动比"全美第七大公司"。
你不再问"值不值"，只问"涨不涨"。
第二层：$28.5万亿TAM。
招股书声称"人类最大可寻址市场"。
锚点越荒谬，目标价越"合理"。
第三层：纳指100的3倍权重放大。
$750亿流通市值在指数中算$2250亿。
新锚点再推一轮"还得涨"。
独立估值怎么说？
Morningstar：$7800亿（-55%）
达摩达兰：$1.22万亿（-30%）
伯里："不值1万亿，更别说2万亿"
三个专业估值远低于IPO价。
市场视而不见——因为锚已经把"合理"拉到$1.77万亿。
1
110
Jun 7
【FOMO工程四步曲】
第一步：制造稀缺。
SpaceX私有多年，散户够不着——
天鹅绒绳越紧，想进去的越疯。
第二步：降低门槛。
Fidelity认购最低从$10-50万降到$2000。
大学生兼职两周就够了。
这不是普惠金融。是扩大FOMO覆盖面。
第三步：社交病毒。
X上散户疯狂晒认购单。
每一条都是FOMO的二次传播。
第四步：超购恐慌。
路演需求$1500亿，2倍超购。
散户不是在决策，是在抢购。
为什么马斯克需要散户？
机构怕踏空，但填不满$750亿的盘子。
散户的FOMO帮马斯克拿到了更少的发行折扣。
还有一个原因：散户不会联合起诉你。
58
Abyss Edge retweeted
May 25
36
125
555
1,031,082
May 28
Why You Keep Losing Money to Your Own Brain
A thread on loss aversion—the invisible tax on every investment decision you make.
1/ Why "Don't Lose" Beats "Make Money"
Loss aversion: losing $1 hurts ~2x more than gaining $1 feels good.
Nobel laureate Kahneman's famous experiment:
→ 80% chose the "sure thing" ($3,000)80% chance to win $4,000 vs. 100% chance to get $3,000
Flip the frame:
→ 80% chose the gamble80% chance to lose $4,000 vs. 100% chance to lose $3,000
Same math. Opposite choices. Because we're human.
1
21
May 28
The Bottom Line
They're beaten by their own nervous systems.In markets, 90% of people aren't beaten by institutions.
Loss aversion isn't your enemy. Recognizing it is.
First, be unlosable. Then, wait for the other side to become losable.
— Sun Tzu, probably, if he traded
1
13
May 28
The Disposition Effect: Your Portfolio's Silent Killer
When you're DOWN: "I'm a long-term investor now..."When you're UP: "Take the profit, lock it in!"
This is the disposition effect—and it costs the average investor ~3.4% per year. (University of Chicago, 10,000 accounts, 10 years.)
Compound that over a decade.
That's 30% of your wealth. Gone. To your own brain.
9