Sabitlenmiş Tweet
@0xfdf The same thing happens in trading: someone from a top shop can be very expensive to hire, but the chances that they understand something useful deeply enough are often low. The market has adverse selection
English
Georgy Khalin
483 posts
Georgy Khalin
@gkhalin
Quant dev lead at https://t.co/HuN94SnxTi. Deep understanding is the only real leverage
Amsterdam, The Netherlands Katılım Mart 2012
134 Takip Edilen188 Takipçiler
Why do people tend to judge others after meeting them once or twice? Real values can only be tested over time
English
@swyx I would prefer to manage 5-10 people, each managing 10 agents
English
@drgurner That was high impact advice for me
I have a huge bias toward action. Sometimes I forget to take more time to think
English
Build in 30 min of space every day to think. If you are always back-to-back busy, you'll always stay smaller than you could be.
You'll miss opportunities.
The people who have leverage, have space. They can think, see the field, be deliberate, and strike with intention.
English
@systematicls Some people phrase things so they’re right in almost any context. They also make sure you know how right they are. Very annoying
English
Don’t blame agents for poor work if you give them no way to verify it
English
Useful knowledge is something you keep referring back to in future experiments. Extracting that kind of knowledge is surprisingly hard
Jack Morris@jxmnop
my definition of science: the continual process of (a) generating artifacts that are surprising or useful and (b) explaining them these days there's exponential growth in papers, and in useful artifacts, but we don't often discover good explanations so i was happy to stumble upon this extremely thorough paper about how RL surfaces useful capabilities from base models. clear experiments w great explanations are hard to find these days
English
@drgurner That’s why you can’t fully trust your judgment about someone until you’ve seen them under real pressure. It takes time
English
Never believe what people tell you, watch what they do.
Look at the evidence.
This is just as true in relationships, as it is in business.
People say a lot of things, but consistent actions (over a long period of time) are what always tells the tale.
English
@signulll Like in real life, sometimes it’s better to say “I don’t know” or “I don’t have an opinion”
English
podcasts reward authenticity, but they’re also long form environments where inexperience & hesitancy compounds like crazy.
unlike a 10 minute cnbc planned hit, two hours of pseudo unscripted conversation will expose whether someone has actually thought through their worldview or not.
if you’re a ceo or you wanna be one, you’d better be seasoned as hell at this. it’s a true test of eloquence, depth, & articulation ability in close to real time.
English
Stress helps exploitation, but harms exploration. It’s hard to come up with creative ideas when you’re anxious and in a rush. Startup grind culture works because most of the work is exploitation
English
@__paleologo @ValedictorianZZ We should apply to find out the interview process 😂
English
@ValedictorianZZ If it helps, I don’t ask brain teasers. Just relatively basic questions on topics mentioned in the candidate’s resume.
English
I remember Jim Simons saying in some interview or fireside chat (it’s a blur) “I wasn’t the fastest at solving problems, but I was deep”. How nice that another geometer (in a different area) was made of the same stuff.
Michael Guo@Michaelzsguo
国内和国外学校在王虹获奖后给出的评价可谓是截然不同。 北大教授/同学对王虹的评价： 不是竞赛生，不是最亮眼的，班上小透明，但是及其自律，勤奋，肯下苦功，靠长期坚持实现逆袭。 MIT博导评价王虹: 具有慢而深的研究特质，不被传统束缚，能创新方法找到前人未发现的突破口 巴黎理工学院老师评价王虹：直觉罕见，远超同年级所有学生，快速攻克超纲难题 是及其罕见的天赋，韧性，思维深度拉满的学生。 北大的教授，评价王虹仍然是从她没有竞赛背景出发，没有天赋，强调她是靠勤奋实现“逆袭” 美国和法国的导师都盛赞王虹的天赋异禀，思维深度，直觉，和韧性。
English
@staysaasy We play it in our company, including management, and it has worked out well so far
I wouldn’t work with people I don’t trust
If they can’t handle my secrets, they can fire me
English
My heart goes out to this guy, but I thought that everyone with more than 5 years of work experience knew that when the trust exercises start you just make up some boring vulnerability like how you think labradoodles are overrated or cried when you saw the movie Up
English
@fchollet I’d rather see the conclusions than the raw results. Most experiments fail anyway, and processing all the outputs takes too much time
English
In science, you have to report the experiments that didn't work, not just the ones that did. Same with AI. (I wish.)
English
@staysaasy Remember the last time you met your high school or university classmates
The outcomes are often surprising
English
My dad has five kids and we always had a million friends around our house. We also lived in the same place for 30 years so you could see where people ended up.
He said you could see who someone essentially was by the time they were like five years old. Doesn’t mean they don’t surprise you with where that person ends up in the world, but their core self is baked in from the get go.
Paul Graham@paulg
We have multiple reasons for wanting to believe nurture is more important than nature and none for wanting to believe the opposite (unless we = the far right, which we ≠), so it's not surprising we overestimate its effect. But man are kids' personalities inborn.
English
@craigweiss Was the dot com bubble really the opportunity window?
Many of the iconic companies were founded during it, but a lot of the value was created after the crash
English
what you choose to do now will meaningfully change the trajectory of your future and your offspring’s future
Craig Weiss@craigweiss
99% of population is clueless, yet we’re in the singularity
English
@kunalvg Killing ideas fast has always been the moat for making progress
AI helps you execute ideas much faster
It doesn’t help nearly as much with generating high-quality ones yet
English
Become a self-sovereign individual.
Acquire skills at speed. Build wealth outside the salary cheque. Own your mind and body.
AI will render most ways of making money in tech useless.
The skill that matters now is making money with AI in the loop. Employed or not.
English
@signulll Judgement quality doesn’t seem to be improving that fast for creative work like hypothesis generation
English
my current general principle of ai:
ai gets you roughly 2/3rds of the way to a good result on almost any cognitive task.
that final third is where taste, judgment, iteration, domain expertise, & originality compound.
today the last third is where almost all the value lives. as the models get better & better, the last third will shrink to ~zero & then possibly negative.
Josh Elman@joshelman
It has never been easier to get first draft quality work. Almost no effort. This means that taste and final polish will matter more than ever to rise above the slop. We will see this in every category.
English
Benchmarks are hard. What makes you think that offline metric reflects long-term success? If production PnL is deeply negative, is it market impact, noise or a bug? Trust metrics, but ground truth is noisy too
English
@nickcammarata Same if you ask it to come up with 3 hypotheses. It lacks judgement
English
