gubbrora

15 ms·

121.
by gubbrora 8y ago
One way could be spectroscopy
122.
by gubbrora 8y ago
It's bigger than "just one company". You say it's the company's free choice but that is not true. There is a large amount of pressure on companies to conform. It's a part of a broader movement to shut down diss
123.
by gubbrora 8y ago
It's a made up problem. There has always been lots of bad information. People are better informed than ever. Just take religion and magic and superstition which have declined a lot. It's a made up problem to justify authoritarian
124.
by gubbrora 8y ago
> could have been caught by a run time bounds check And here I thought rust was all about zero cost abstractions.
125.
by gubbrora 8y ago
I don't love it because let's say I really did want to change it, it doesn't tell me how much effort has previously gone into it and what happened. Maybe it means 10h have been spent in vain maybe 10000h. If I want to try my
126.
by gubbrora 8y ago
> And my likely future is so short that there's no point in long-term planning. I've enjoyed your hn posts so thank you for that. I hope you can make the most of the time you have left.
127.
by gubbrora 8y ago
> GEB indicates that your self (your consciousness of being someone) is a swirling self-referential symbolic process It advances the hypothesis. > Dawkins indicates that your self is a kind of evolved meme whose function in nature is
128.
by gubbrora 8y ago
I find dark mode comfortable. I find light mode comfortable. Using both in different programs is very uncomfortable.
129.
by gubbrora 8y ago
Slow code review is really bad for productivity. Both cost of task switching as you wait, and of merging upstream changes into your branch.
130.
by gubbrora 8y ago
Imagine if instead of holding the budget hostage people held law hostage.
131.
by gubbrora 8y ago
School / company network with content blocker?
132.
by gubbrora 8y ago
There are faster algorithms than 3 nested for loops.
133.
by gubbrora 8y ago
On further thought I think this is wrong. Large rotations can cause coefficients to become negative, so there are constraints.
134.
by gubbrora 8y ago
From my understanding the vectors don't really matter so much. We care more about the subspace they span (because we want to project our data into it). So any rotation is as fine as any other.
135.
by gubbrora 8y ago
Back before deep learning was all the rage textbooks used to argue that two layers were enough because a two layer network can approximate anything arbitrarily well. They were right and wrong. Two layer networks CAN indeed approximate any f
136.
by gubbrora 8y ago
Float/doubles are finite. They only approximately model the real. And that's key. Because the fact that approximation works for a problem means it must be reasonably smooth, which means everything will be more well behaved and ru
137.
by gubbrora 8y ago
5% profit margin doesn't mean I will lose 5% on average. My risk will be pooled with reckless people and scammers. I know I am neither but my insurance company doesn't. Then there are the sales and administration costs that need