Category Archives: tech

CV Dazzle Updated

Adam Harvey has updated CV Dazzle
with all new patterns. (Previously.) The new patterns are aimed at defeating newly developed face detection algorithms from a variety of vendors including OpenCV, VeriLook, and Apple.

His main suggestions are:

  1. Use contrasting makeup, light colors on dark skin, and vice versa.
  2. Obscure the nose bridge.
  3. Partially obscure an eye.
  4. Modify the contrast, tonal gradients, and spatial relationships of light and dark areas of the face.
  5. Try obscuring the shape of the head.
  6. Make the face appear asymmetrical.

Suffering Robots

Via Slate:

According to Lance Gharavi, an associate professor of theater at Arizona State University, the question of free will rapidly resolves into a problem of desire. Steering the conversation into philosophical terrain, he observed that we can’t even say definitely whether humans have free will. But, he continued, if a robot has desires, even if those desires just involve the need to appropriately serve its master, then it can suffer. And if it can suffer, we have an ethical responsibility toward it. For Hartzog, on the other hand, the ethical stakes of human-like robots have more to do with the ways that we relate to humans.

Previously.

Hack The Zero Rating

When Paraguayan hacker Matias Insaurralde, found out that largest mobile provider in Paraguay, TIGO, had patterned with Facebook’s Internet.org to zero-rate Facebook, he decided that all the web should be accessible, not just Facebook. So he build IP over Facebook. While still in early development, his project routes network packets to a proxy via Facebook Messenger.

This is good. We need more of this.

Previously.

People for the Ethical Treatment of Robots

In a Twitter inspired story, CNN asks, “is it cruel to kick a robot dog?”

In what has become a become a regular demo by Boston Dynamics, engineers kicked the Spot the Robot Dog in order to push it off-balance and watch it stabilize itself. This time (and probably every time as well) number of people on Twitter said, “that poor robot!”, and so CNN rounded up the tweets and asked retired AI and robotics professor, Noel Sharkey about the ethics of kicking robots, who said, “The only way it’s unethical is if the robot could feel pain.” He then followed it up with the warning that because humans anthropomorphize things, we may become more likely to abuse things that can experience pain, because we’re used to it. He drew comparison to the philosophers who argued animals were “clockwork” (I’m not familiar with anyone using that term, but certainly the comment that animals are lesser than humans because they are soulless has been around for thousands of years), but none the less argued that animals should not be abused because it debased the abuser.

His comment that it wasn’t unethical because animals did not feel pain, got me thinking. Yes this is certainly true, that the robot does not have any sensors to indicate damage with respect to these kicks, and it does nothing other than regain its balance, but the idea that “It’s cool, it doesn’t feel pain,” strikes me as just a variation of the old thinking machine conundrum. We say computers don’t think, because we completely understand the rules that govern its behavior. We say robots don’t feel pain, because they’re not alive. But it seems to me, that with pain, we have a simpler Chinese Room. I know that when I’ve run computer programs integrated computer programs that were suffering from some sort of system fault and logging errors continuously as “being in pain.” No the programs weren’t alive, but what is pain other than a signal indicating damage or negative reinforcement? Certainly error counters and exceptions do that. In a sense, that’s pain, or at least a reasonable functional facsimile there of.

So is it wrong to kick Spot? I’m thinking it’s not, but at the same time, if you’re do it too much, and for enjoyment, maybe it is. Perhaps that’s not very satisfying, but isn’t it often the case, that the motivations of the actor the determining factor when deciding if something is moral or not?

When Charity Isn’t

Quartz recently reported that “Millions of Facebook users have no idea they’re using the [I]nternet”. Specifically, in southeast Asia and Africa, there are more self-identified “Facebook users” than there are “Internet users.” Most of the Internet basically just said “Ha! Ha! Look at those stupid people!”, but that misses what’s really going — and explained in the Quartz article.

In the developing world, most people don’t have computers, they have phones, and those phones aren’t even smartphones, they’re feature phones. When they browse Facebook, they aren’t using an app, or even m.facebook.com, they’re using what’s called “mbasic“. Additionally, the bandwidths are limited — say 2G — and the data rates are relatively expensive. In a very real sense, trying to surf the modern web is nigh-unusable. So Facebook experience is in certain sense not the Internet. The trouble comes in from the fact, that’s how Facebook wants it.

FB is a walled garden. It always has been. It’s a walled-garden with a billion users on a 7 billion person planet. There’s just not that much growth in the wired world available beyond population growth… unless you increase the number of potential users. And that’s where Internet.org comes in. Internet.org is a corporate partnership between Facebook and various feature phone manufactures that is “dedicated to making affordable internet access available to the two-thirds of the world not yet connected.” It sounds like a charity. It talks about the good works it does. But in reality, it’s just a ploy to increase the number of Facebook users. One of Internet.org’s main tactics is “zero-rating“. Zero-rating is a network-biased approach where telcoms or ISPs don’t charge customers for packets exchanged between them and the zero-rated site, thereby biasing user behavior.

There’s nothing altruistic with Internet.org, it’s all about increasing advertising revenue. That’s not necessarily a bad thing, that’s how the company makes money, and hell it pays my bills. And yeah, Facebook is primarily user generated content, and yes it is possible to do your own things on it, but the thing that bothers me about Internet.org is the whole mission from God rhetoric about it. “Is Internet connectivity a human right?” Well maybe, maybe not, but spouting this while promoting a walled garden is bullshit.

FBI Snatch Attack

From the Silk Road trial:

As soon as Ulbricht was distracted, another agent grabbed the open computer and gave it to Kiernan, who is an FBI computer specialist. Kiernan spent the next three hours doing “triage” on the machine. Without allowing it to go idle, and thus become encrypted, he took photographs, went through the browser history, and ultimately handed it off to another agent who imaged the hard drive.

So the FBI grabbed the laptop, shoved a mouse jiggler into the USB port and began to photograph the screens while simultaneously arresting Ulbricht. Good. Clever. Maybe b cause I was a a wannabe script kiddie (I never did anything) reading alt.2600 and alt.cypherpunks back in the day, but I read this and immediately thought about ways to make a deadman switch to foil the snatch attack.

As Quoted in the New York Times

Tech Giants Settle Antitrust Hiring Suit:

On Hacker News, the grumbling about the settlement was immediate. “I honestly do not understand why the plaintiffs would settle this case,” one poster wrote. But some saw another side: “What did the engineers risk with this lawsuit? Nothing. What did the law firm risk? Getting paid peanuts for hundreds of hours they spend on the case if they lose.”

The Only Gun The NRA Doesn’t Like

Meet the Armatix iP1. 22 LR calibre semiautomatic with a 10 round magazine. What’s not to like? How about the the RFID wristwatch that only allows the wearer to fire the gun?

Back in 2002, New Jersey passed a law mandating that all guns sold in the state had to have lockout technology built in, three years after such a gun went on sale somewhere in the United States. The gun has gone on sale &emdash; albeit very briefly &emdash; twice, once in Los Angeles, and then again in Rockville, Maryland (a place that it is strongly advised that one to not return to). Both stores stopped selling the guns after receiving death threats. Not only that, but the Republican Attorney General of New Jersey is refusing the certify that the gun was sold, and active the law.

Instead of “promoting freedom” to own the gun one chooses, the NRA is actively campaigning against the sell of a particular firearm, because the RFID locks would increase manufacturing costs. Why do they care about this? Because the NRA gets significant amounts of funding from gun manufacturers.