Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
T_S_
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
53 ms
·
481.
▲
by
T_S_
16y ago
R and S-Plus are very similar. Curious when you would prefer S-plus over R?
482.
▲
by
T_S_
16y ago
Good stuff.
483.
▲
by
T_S_
16y ago
Foreign function calls are a good thing to support. R does this well and lots of R routines already call C or Fortran libraries. Domain specific languages are good (and fun). Interacting with open source developers can be an awesome expersi
484.
▲
by
T_S_
16y ago
The article provides a very good example in support of the argument domain specific languages to help you crank out domain specific code. I like R and use it. But I also code my own methods (in haskell these days, but that's not the main po
485.
▲
by
T_S_
16y ago
There is an overwhelming desire by non-programming business people to reduce programming to something they understand better. Not knowing how to program, they have to grasp for analogies like building a house, a bridge or some other large
486.
▲
by
T_S_
16y ago
Those two firms are ok, from people I have spoken to there. There are a bunch of small prop shops around, throwing spaghetti at the wall, hoping something sticks. Like many other fields, if you can get in with a quality crew it can be a goo
487.
▲
by
T_S_
16y ago
My pithiest comments never get enough upvotes either. Solution? More pith. Gave this article a sympathy upvote. Hope that helps a little.
488.
▲
by
T_S_
16y ago
If you of education as teaching you how to get what you want , then go ahead and run the numbers to decide. I believe it's better to think of it as helping you learn what to want . In that case the numbers won't help you much and that the
489.
▲
by
T_S_
16y ago
Give me Excel 2003. That was the most usable version for me. (Actually I was the most productive by far in Lotus 1-2-3. The reason I believe was the lack of a mouse.)
490.
▲
by
T_S_
16y ago
Sure, here's one. ppt files with elements that don't work the same on mac and PC. Fonts changing size, elements moving around. These problems make the Mac and PC Office users avoid collaborating. Also, because the releases are are out of sy
491.
▲
by
T_S_
16y ago
Unnecessary rock thrown at the iPhone (almost literally). Still a very good presentation. Since Office on the Mac sucks and has subtle compatibility issues whenever I share files with PC users, I am considering giving the web app a try as m
492.
▲
by
T_S_
16y ago
You mean AA?
493.
▲
by
T_S_
16y ago
You can borrow money from a friend. Right? You can borrow stock from anyone who holds it. If your friend wants his money back, you can pay him cash and the bills don't have to have the same serial numbers as the ones you borrowed. Same with
494.
▲
by
T_S_
16y ago
Clever but I don't see it catching on in the U.S. Looks like a potential lawsuit coming from burglars injured on the job. And if he can't get out, it's a mantrap, which I believe is illegal.
495.
▲
by
T_S_
16y ago
We'll never be able to exit now. Gulf oil and Afghan lithium. Screwed either way. Is it just me, or is anyone else tired of perpetual war?
496.
▲
by
T_S_
16y ago
One way: take in new money at a lower value than the founder would like. There is always a reasonable difference of opinion and majority shareholders can exploit this to dilute the value. They don't even have to take risk. Just make the com
497.
▲
by
T_S_
16y ago
I am banning the phrase "reach out". Call or email me if you object. Tweeting is ok too, if I am following you (unlikely), but please don't reach out to me. edit grammar
498.
▲
by
T_S_
16y ago
Too harsh, for the reason that many of us are not better than this. For lack of a better term, most people need "coaching" in some aspect of their life. Often they either aren't aware, aren't willing to admit it, or don't know where to ge
499.
▲
by
T_S_
16y ago
Here's what I have learned personally, mostly by accident. For intellectual tasks: 1) What seemed hard last night always seems easier the next morning. (Sleep learning? Not sure.) 2) Verbalizing a problem, even to someone that has no chance
500.
▲
by
T_S_
16y ago
Which thing?
501.
▲
by
T_S_
16y ago
If he was truly insane, he never would have told anyone he solved the problem. Lucky for us.
502.
▲
by
T_S_
16y ago
I don't read reluctance into those comments at all. Just look at the comments on personnel management. It's clear he likes it. The early remarks about the "stupid decisions" being made at the first company are a clear sign. Most business le
503.
▲
by
T_S_
16y ago
I'm guessing this fellow is just a little different, and very very stubborn. That might also be the reason he solved the problem.
504.
▲
by
T_S_
16y ago
That was a priceless post. Turns out Feynman was wrong on one thing though. Wolfram does like being the boss. Possibly very much.
505.
▲
by
T_S_
16y ago
You can solve the friends and family issue quickly, at least partially, by giving your business a name and having a few business cards to hand out. Confidence is contagious.
506.
▲
by
T_S_
16y ago
Let's not be snobby about ads. Relevant ads actually help people. I l-l-l-like the ads on google. There I said it. Why was that so difficult? Simply because my opinion of ads is colored by mass marketed ones that are not relevant the vast m
507.
▲
by
T_S_
16y ago
This article mentions nothing about how they measure alertness. How objectively can this possibly be measured? It seems to me that the measure of alertness may only be indirectly measured in by task specific performance. So there may be dri
508.
▲
by
T_S_
16y ago
There have always been news junkies. It's just that the price of crack has dropped. Anything done to excess can harm you, and even if not done to excess, give you cancer.
509.
▲
by
T_S_
16y ago
Let's fix this hunter-gatherer analogy up a bit. When I find some good information to "eat" it never runs out. I can share it with the tribe endlessly. I think a better analogy is that we are all on watch, looking out for the tribe, ready t
510.
▲
by
T_S_
16y ago
I think haskell is worth watching. If projects like repa and dph are able to deliver parallel performance speedups, then haskell could become a top choice for machine learning and related algorithms. Personally, I have found haskell to be
More ›