Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
tmsh
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
44 ms
·
391.
▲
by
tmsh
16y ago
This is Sparta. class SpartanList(list): def foot(self): return self[0] assert l.foot() return l.foot()
392.
▲
by
tmsh
16y ago
He does say Plato's Socratic dialogues. The Republic, which he has problems with (e.g., http://www.capitalismmagazine.com/culture/philosophy/5393-Je... -- thanks for mentioning that correspondence -- had no idea, quite fascinating) we of
393.
▲
by
tmsh
16y ago
Apple run loops are more flexible and adaptable than this http://android.git.kernel.org/?p=platform/frameworks/base.gi... via http://stackoverflow.com/questions/3321587/anatomy-of-an-and... but since Looper is OSS it should be improved.
394.
▲
A programmer's modifications for efficiently operating a Mac
(tmsh.posterous.com)
4 points
by
tmsh
16y ago
|
3 comments
395.
▲
by
tmsh
16y ago
Negative capability http://en.wikipedia.org/wiki/Negative_capability isn't quite the same thing, but it may be related. Flexibility with ignorance can lead to better understanding of one's ignorance and faster adaptiveness overall. Per pr
396.
▲
by
tmsh
16y ago
I'm surprised image filtering technology doesn't have a way to just view a silhouette of the skin. I.e., establish the skin color gradients and then remove them...
397.
▲
by
tmsh
16y ago
Just thinking about what forces might prevent the industry from being disrupted (regardless of whether/how this is good for the public): I think the biggest innovation is the drivers rating the customers. It seems like a joke, but if you th
398.
▲
Foreign Policy: The Mobile Banking Revolution
(npr.org)
1 points
by
tmsh
16y ago
|
0 comments
399.
▲
by
tmsh
16y ago
Actually, I think you have to go back further. My grandfather, a poor kid from west LA, went to UCLA and then UC Berkeley as an engineering student back in the late 1930s / early 1940s. He learned Latin in high school (think inner city kid
400.
▲
A critique of Hacker culture
(tmsh.posterous.com)
1 points
by
tmsh
16y ago
|
0 comments
401.
▲
by
tmsh
16y ago
The Method of Loci sounds very interesting (as well as the Sapir-Whorf hypothesis). But if we're still presuming that speech is inherently linear, I think we're mistaking how the brain abstracts those sequences of texts. I haven't read the
402.
▲
by
tmsh
16y ago
Again, this would be a valid criticism if I had simply said: 'I expect to be downvoted.' Or if I were trying to game the system or otherwise accumulate points. As odd as it is to imagine though, that's not actually my point. My point, to be
403.
▲
by
tmsh
16y ago
Ok. Linearity isn't my term -- it's jgc's. But I presume he and Carr mean something like a sustained engagement via a body of text. In essay writing, this usually involves multiple revisions, again and again, across the same space. In essay
404.
▲
by
tmsh
16y ago
I do speak in very, very general terms here. But I think you can still extract valuable insights if you're careful and you qualify these abstractions. It could be useful to analyze what the nature of an HN signal is? Could be very-low on ar
405.
▲
by
tmsh
16y ago
I suppose my point was to challenge your signal-sensing apparatus. The problem of putting signal v. noise on a pedestal is that we take for granted what the shape of good signals looks like.
406.
▲
by
tmsh
16y ago
I expect to be downvoted as the signal that I would indicate is not exactly on the same wavelength as most -- and so you'll probably interpret it as noise. However, I think Nicholas Carr's points are not completely without value. The proble
407.
▲
by
tmsh
16y ago
BSD's problem is that most progress gets entombed in proprietary forks. It was inevitable that a GPL'd kernel, being immune to that problem, would catch up and displace it. How I read it, fwiw, is that kernel development involves a lot of
408.
▲
by
tmsh
16y ago
I'm just saying: it's questionable whether web application would've succeeded this quickly if Microsoft application/desktop development hadn't been impeded. Perhaps it would've. But it's also questionable whether Linux would've supplanted B
409.
▲
by
tmsh
16y ago
Software patents are a spur to innovation by being an antispur to greed-touching systems. The best thing that happened to Linux was the BSD fight. The best thing that happened to the web was the Microsoft antitrust fight. Just sayin'.
410.
▲
by
tmsh
16y ago
For Arabic, check out Al-Kitaab (and its predecessor book) at Amazon.com (worth reading the reviews, imho). Part of a three year course. Still I think forms the basis for most university programs. May be available on the internets. Some day
411.
▲
A common, deep mistake in polymorphism
(tmsh.posterous.com)
1 points
by
tmsh
16y ago
|
1 comments
412.
▲
by
tmsh
16y ago
Personally, I am for total net neutrality (wireless, etc.). But speaking pure strategy, you'd think after Microsoft antitrust, USL/BSDi, etc., that large tech companies would learn to avoid going anywhere near debates about the law. Legal d
413.
▲
by
tmsh
16y ago
Overlay sounds so much better than popup...
414.
▲
by
tmsh
16y ago
I think this gets at the distance between US culture, as it were, and the realities of violence. War is not the only source of violence. But it is a big one, which affects the entire culture pretty significantly. So the 70s correspond to Vi
415.
▲
Suggestion for betting on the union between social / interest graphs and search
(tmsh.posterous.com)
2 points
by
tmsh
16y ago
|
0 comments
416.
▲
by
tmsh
16y ago
Tim O'Reilly warned us: http://www.google.com/buzz/timoreilly/j61qZ42h6rB/Frustrated... But seriously, that's strike two. One more strike, and personally I won't be reading the NYTimes for anything technology-related. Even via a link. Or
417.
▲
Debugging GLSL with Vim
(tmsh.posterous.com)
2 points
by
tmsh
16y ago
|
0 comments
418.
▲
by
tmsh
16y ago
Ah yes, I find myself wondering: how to explain to younglings that even more important than the realization that one needn't conform to the status quo (and indeed one might reject one's entire frame of reference) is that the truth -- reject
419.
▲
by
tmsh
16y ago
I think the largest fallacy is that HFT is about making money solely via high-frequency. However, HFT is rather a group of market making strategies that operate at high speeds. They'd still work at lower speeds (above a certain threshold).
420.
▲
by
tmsh
16y ago
You ain't kidding. http://news.google.com/news/search?aq=f&pz=1&cf=all&...
More ›