Profile cover photo
Profile photo
Gary Boone
Technology. Machine Learning. Google.
Technology. Machine Learning. Google.
About
Posts

Post has shared content
In her debut appearance today at a Senate Banking Committee hearing, Sen. Elizabeth Warren from Massachusetts made federal regulators uncomfortable when she asked a simple question: When was the last time you took a big Wall Street bank all the way to trial?

Masslive.com reports:

"Elisse B. Walter, chair of the SEC, said that although they can take financial institutions to trial, they typically do not go that route.
"'As you know, among our remedies are penalties but the penalties we can get are limited,' Walter said. 'When we look at these issues, and we truly believe we have a very vigorous enforcement program, we look at the distinction between what we could get if we go to trial and what we could get if we don't.'"
"We have not had to do it as a practical matter to achieve our supervisory goals," Thomas Curry, the Comptroller of the Currency, which regulates national banks, added.

Warren said that she was worried that banks are simply paying fines from the profits they earned breaking regulatory rules.

"I want to note that there are district attorneys and U.S. attorneys who are out there everyday squeezing ordinary citizens on sometimes very thin grounds and taking them to trial to 'make an example,' as they put it," Warren said. "I am really concerned that too-big-to-fail has become too-big-for-trial."

Warren is not the first to express concern that the SEC is settling too many cases before going to trial.

U.S. District Judge Jed S. Rakoff, based out of Manhattan, has thrown out two huge settlements between the SEC and Citigroup and the SEC and Bank of America.

Rakoff has argued that settling without a trial was not in the public interest. The SEC, he said, "has a duty, inherent in its statutory mission, to see that the truth emerges." And that means taking some cases to trial.
Add a comment...


Ok, now this is fun. Seriously geeky, Internet engineer, insider fun. To appreciate it, I'll show you a little bit about how the Internet works. 

We talk about Internet locations in ways that make sense to humans, such as "news.google.com". But the computers use a numbering scheme for addresses, typically formatted as four numbers separated by dots, like "74.125.224.64". It's easy to look up addresses; just use a tool called "dig". If you have a Mac, open a terminal (⌘-space, type "terminal"), then type "dig google.com" and hit return. You'll see a list of addresses. You can even paste one of these into your browser address bar---guess where you'll go.

So thats's how to turn a domain name into an IP address. Next, you can see how the Internet routes from your machine to that address using a tool called traceroute. In the terminal, type "traceroute 74.125.224.64" and hit return. You'll see a series of computers by their names and IP addresses--that's the route to Google--and you can see that each one adds a little bit of time to the total along the way. Each machine in that list is contacted and forwards your request toward a machine that can return the requested Google home page. There are even applications you can install that will show the route on a map.

Now that you know what traceroute does, try "traceroute 216.81.59.173" and read the results.

Heh. That's awesome! What happened?

The first few hops are standard Internet routing. But evenutally, the routing computers forwarded your request to routers controlled by Ryan Werber. He then bounced them around some virtual routers along a path to which he gave creative, epically story-telling names. 

Brilliant!
Add a comment...

Post has attachment

What's so utterly, astoundingly, amazing about the Mandelbrot Set isn't its beautiful visual complexity. Ok, that's a big part of it. It's the fact that the complexity is emergent from such a simple formula. 

Look, all you do is make a sequence x <= x^2 + c and see if it converges for different values of c in a complex plane. Pick a point c on your screen. Start with x=0. So x1 = 0^2+c, x2 = x1^2+c, x3 = x2^2+c. Color the point c on the screen according to how many steps it takes for the magnitude of x to exceed 2. It's that simple.

With that description, what would you expect to happen? Given the simplicity of the process, you might think nothing particularly interesting. The Mandelbrot set's complexity shows quite the opposite. It's fantastically complicated. 

But to really blow your mind, watch what happens when you zoom in. It's even more complex. And you find copies of the original Mandelbrot shape recurring. And if you keep zooming in you find more and more complexity. If you watch this video to the end, how far have you zoomed? Well, you'll be zoomed in so far that the original image you started at, if drawn full size, would be 10^261 times the estimated size of the known universe.

mandelbrot fractal deep zoom 15 2^969 (HD)

How I calculated that:
1. The floating point resolution of the zoom endpoint is 2^969, according to the video.
2. Assuming you're watching screen with say 1024 resolution, you're using 10^3 bits for the screen, or 2^10 bits.
3. Let's say your screen is a 15" laptop screen. Say 36" in a meter.
4. The universe is about 10^27 meters in size.
5. 2^969 / 2^10 * 15 / 36  / 10^27 = 2 * 10^261 universes
Add a comment...

Post has attachment
Add a comment...

Post has attachment
So take a paper strip and fold it half, then again. What pattern of folds do you expect? All valleys? Not so--once folded in half, the next fold is paired: one up one down. What happens as you add more folds always in the same direction? The answer may surprise you. 

What's so amazing about this is that such simple processes creates such fantastically complex patterns. And what's even more amazing--really this must blow your mind--is that at the end, the complex pattern that emerges from rotating folds fits together with itself rotated! What?! Cannot compute!!

Dragon Curve - Numberphile
Add a comment...

Post has attachment
Ask your friend or SO these questions:

1. How do you pronounce P-O-L-K?
   They'll say "poke."
   What do call the white part of an egg?
   They'll say "a yolk."

2. How do you pronounce G-R-E-N?
   They'll say "gren."
   What color is the ocean?
   They'll say "green."

2. How do you pronounce S-T-O-O-P?
   They'll say "stoop."
   What do you do when you come to a green light?
   They'll say "stop."

Now you can remind them that they just called an egg white a yolk, said that the blue ocean is green, and that they stop at a green light instead of going.

In these examples, the first questions engage the system 2, deliberative thinking, whereas the second question engages the system 1 fast thinking. Woops.

You didn't consider yourself rational, did you?

Brain Tricks - This Is How Your Brain Works
Add a comment...

Post has attachment
Maybe a little too simplified, but at least shows that the funny squiggles we call 'letters' aren't random, but carry a long cultural history.

http://i.imgur.com/aJA4o.gif
Add a comment...

Post has attachment
Now this is awesome! Remember the strange noises original telephone modems use to make when connecting to BBSs, Compuserve, Prodigy, and America on Line? Here's an explanation of what the modems were doing. In addition to negotiating a protocol, they actually analyzed and conditioned the telephone connection. And you can actually hear it: as they use more bandwidth, it sounds more like noise. 

http://www.adafruit.com/blog/2013/01/30/breaking-down-the-handshake-sound-anatomy-of-the-modem-connection/
Add a comment...

Post has attachment
Add a comment...

Post has attachment
I suppose we've known since Freud that we're irrational creatures governed by emotional and perceptual forces we're unaware of. Still, I don't like being robbed.

Colour Changing Card Trick
Add a comment...
Wait while more posts are being loaded