martincmartin

9 ms·

61.
by martincmartin 3y ago
correlation ... strongly positive. A positive correlation just means better than chance. "Strongly" is vague, and might not be much better than chance.
62.
by martincmartin 3y ago
The people at Twitter who understood the system and could predict the side effects were all fired or left. My guess is Elon said "the site's too slow!" Engineers noticed that the home feed request was slow. They didn'
63.
Downtown San Francisco is at a tipping-point(economist.com)
80 points by martincmartin 3y ago|196 comments
64.
What performance-enhancing stimulants mean for economic growth(economist.com)
1 points by martincmartin 3y ago|0 comments
65.
by martincmartin 3y ago
The quantum mechanics stuff misses EPR & Bell's theorem, which roughly says that the probabilities in quantum mechanics are essential. Experimental confirmation of this won the Noble Prize last year. Feynman famously counseled on
66.
by martincmartin 3y ago
Many of these things are at the core of the book Java Puzzlers. http://www.javapuzzlers.com/
67.
The promise of crypto has not lived up to its initial excitement(economist.com)
53 points by martincmartin 3y ago|128 comments
68.
AI Like ChatGPT Are No Good at 'Not'(en.wikipedia.org)
1 points by martincmartin 3y ago|2 comments
69.
by martincmartin 3y ago
I'm calling out lazy security that cuts corners because doing it well takes time and effort. Time and effort cost money. It's possible that the value gained is less than the cost, and so therefore isn't worth it. Of cours
70.
by martincmartin 3y ago
Yes, these are the same issues faced by databases on spinning rust disks. The solution was to figure out how many keys fit in a disk block / cache line, let's call it k, then have a k-ary tree instead of a binary tree. This is c
71.
by martincmartin 4y ago
Yes, but Australia is " the land mass that makes up the entirety of the mainland" of Oceania.
72.
by martincmartin 4y ago
The experiments are now so sensitive that if an electron were the size of Earth, they could detect a bump on the North Pole the height of a single sugar molecule. This new bound is sensitive to energies above roughly 10^13 electron-volts
73.
The electron is so round that it’s ruling out potential new particles(quantamagazine.org)
179 points by martincmartin 4y ago|139 comments
74.
by martincmartin 4y ago
There's a sense in which worm holes allow information to travel faster than light. However, since the worm hole itself has a reference frame, you can't use it to create a grandfather paradox.
75.
by martincmartin 4y ago
Doesn't work with North Korea, Venezuela or lots of other dictatorships around the world. Regime change is hard. Creating a functioning, not corrupt state in the aftermath is harder.
76.
by martincmartin 4y ago
The barrier to outsiders invading and fixing it is nuclear weapons. If the west's troops invaded Russian, it's quite possible Russia would launch nukes. Even without that, the West doesn't want to put boots on the ground in
77.
by martincmartin 4y ago
Anyone have a tl;dr for this?
78.
How TikTok Broke the Ad Business(economist.com)
42 points by martincmartin 4y ago|75 comments
79.
by martincmartin 4y ago
I read a weekly news magazine (think Time, Newsweek, The Economist) for that reason. No need for AI, actual people can give you a weekly update!
80.
by martincmartin 4y ago
When it makes an illegal move, and is told "that's an illegal move, choose a different move," if it can then come up with a different move for the same board, I'd say it plays chess.
81.
by martincmartin 4y ago
As the OP says, yes. The CEO then needs to sell their house and use the money to pay wages.
82.
by martincmartin 4y ago
WhatsApp stopped charging users, and doesn't run ads. It has no revenue.
83.
by martincmartin 4y ago
I agree the tools are not ready because, as you say, the foundational concepts are not yet encoded. But for OP, the point isn't having humans understand. It's having the computer check the proof. You still use traditional journa
84.
by martincmartin 4y ago
The paper says code is available as part of the Sux project at: http: //sux.di.unimi.it/
85.
by martincmartin 4y ago
Could a large language model help with this? Like, train it on papers, then feed it your rejected draft and ask it to generate an improved draft that sounded more like the training data? The idea would be to use the expected phrasings and
86.
There is a worrying amount of fraud in medical research(economist.com)
296 points by martincmartin 4y ago|108 comments
87.
by martincmartin 4y ago
Suppose you do this, try some different models, get the error down on the training sample. Then you apply to the test sample and the results are ... about the same as the well known results. The best practice says you should now quit and
88.
by martincmartin 4y ago
How's your work-life balance? I worked in the games industry a couple times, at Activision in the 1990s and later at Rockstar, and like almost the entire industry, everybody was in crunch mode almost all the time. I remember when the
89.
by martincmartin 4y ago
Inside the star, or even on the surface, there is a lot of energy, so you won't see the light that specifically comes from a single electron, in the first excited state with a well defined energy, that then decays to the ground state.
90.
Deciphering My Dead Mother’s Cipher(martincmartin.com)
2 points by martincmartin 4y ago|0 comments

More ›