Forgot your password?

Comment: automatic brakes (Score 1) 235

Yup, all the while current cars that won't even qualify as "A.I." but simply as auto-brake / collision-avoidance functions already have the ability to slow down, sound an alarm, and in worst situation slam the brakes to avoid colliding with big object (i.e.: avoid killing people without even being able to recognize people or even have the concept of "people" in their code).

We haven't already started bringing automated vehicles out of google labs, and we already have technology to avoid killing people, by using much simpler technology.

These etchics/philosophy discussion indeed look a bit pointless.

Comment: Just hit the break. (Score 1) 235

For example, hitting an elderly person in order to avoid hitting a small child.

A not even that much intelligent car would have notice a long time ago that there two object on the street (no need to identify them. There are just 2 big masses on the road), and the if car is kept on the same trajectory it is set for a collision course.

the would already have started pre-braking, sounding some imminent collision alarm, blinking lights on the dashboard

By the time you reach the situation where a human would need to steer some way or another, a car with anti-collision system would have slowed down and stop at rest (unless the driver has overridden the system by voluntarily smashing down the accelerator against all car's alarms).

No need for complex recognition and identification of pedestrian. Just plain simple recognition that there are 2 masses of significat size.
No need for complex ethics engine evaluation, just being able to notice that said masses currently occupy a place that is intersected by the current trajectory of the car.
No need to aim for one while sparing the other, just slow down and brake well enough in advance (and cars electronics are much faster at noticing and reacting as human's slow reflexes and limited attention (or lack of) ).

I'm not speaking about some potential futuristic technology. I'm speaking about car that are street legal and currently circulating on a road near you. They're not even self driving, but they are already able to efficiently avoid collisions.

We haven't already started producing self-driving car beyond a few prototypes at google's lab, and we already have the necessary technology to avoid both deaths.

All these "ethics in robotic cars" are nice though experiments for a highschool's philosophy classes, but they are completely out of touch with technology. For any of these though experiment, the technology will reach a development level where casualities can be avoided a long time before a car's A.I begin to be able to have an ethics discussion with the philosophy teacher about the value of life.

Comment: PrimeCoin & RieCoin (Score 2) 265

by DrYak (#47693839) Attached to: Are Altcoins Undermining Bitcoin's Credibility?

PrimeCoin and RieCoin currently have no known FPGA nor ASIC implementation and the few GPU implementation don't seem to bring the huge leaps in performance that GPU brought to Bitcoin's SHA256.

It's not that they were designed with the purpose of being CPU-only (ASIC-resistant was never the main foal in creating them).
It's that they are based around actual scientific problems that ARE difficult. Designing silicon for them *would* be a scientific improvement.

Comment: ASIC will always exist... (Score 2) 265

by DrYak (#47691629) Attached to: Are Altcoins Undermining Bitcoin's Credibility?

...but not necessarily have huge advantage.

ASIC-proof algorithms attempt to leverage the fact that ASICs are specialized. They try to incorporate things found in CPUs and GPUs but not in ASICs. {...} Newer generations ASICs are now taking over alt-coins and CPU/GPU miners are once again being driven out.

You can build an ASIC for pretty much anything, and beyond some point, it might be commercially reasonnable to attempt it (if the market is big enough and there are enough people interested into buying hardware, it might be worth trying to design sellable hardware).

Now the question is: is there an actual gain in doing it? And that's where all the details lie in.
Depending on the complexity of the Proof-of-Work algorithm, it might range from
- PoW is dumb easy. Each jump in technology (CPU, GPU, FPGA, ASIC) result in massive increase in performance. Hashing power jumps forward several orders of magnitude. Each new technology simply obliterates the relevance of the previous generation (that's the case with bitcoins' SHA256 algorithme).
all the way to the opposite:
- PoW is awfully complex. The algorithme has so much requirement, that your ASIC basically ends up being a slightly custom CPU. The only real benefit compared to GPU, is that the ASIC consumes a tiny bit less power when compared to a GPU.

Also, keep in mind that some PoW actually solve real-life scientific problems. If somebody managed to create ASIC hardware for PrimeCoin or RieCoin, there's some publication-worth fame to be made.

Comment: Which of the backdoors (Score 1) 93

by DrYak (#47684117) Attached to: Not Just For ThinkPads Anymore: Lenovo Gets OK To Buy IBM Server Line

Which of the back door are you speaking about?

The one mandated by NSA that they put in hardware of any American owned company ?
Or the backdoors that the Chinese put into any parts that they ship from their plants to US to build computer ?
Or the backdoors that the Russian somehow still managed to cram in even if they weren't in theory involved in the production of that precise piece of hardware ?

Comment: Certificate Patrol (Score 2) 166

by DrYak (#47682721) Attached to: Watch a Cat Video, Get Hacked: the Death of Clear-Text

I believe there is a plug-in for Firefox that alerts you when certs change too.

Certificate Patrol is an example of such extension.

It does detect strange changes in certificate authority (for exemple when a Man-In-Middle attacker is using a bogus certificate signed by rogue CA or by stolen keys from some CA).
It also detect un-called-for changes in certificate (for exemple, the actual authority has been coerced by the government to sign their spy-server keys, and thus you get a new legit-looking certificate, even if the old hasn't been revoked and and is still well within its validity range)

Comment: License (Score 1) 226

by DrYak (#47679191) Attached to: Posting Soccer Goals On Vine Is Illegal, Say England's Premier League

In pratice, that might be the case. The company is affraid of losing eyeballs (and thus ads revenue) from all the people who woould have wanted to only have a quick recap about the play and would only be interrested into seeing the goals.


The company has clearly bought an exclusive deal over THE WHOLE play. Absolutely whole 2 half-time, and any extra overtime and play-offs. They have paid an excluse right for perhaps *2 hours* worth of Soccer/Football.
The vines of the goals are only a few seconds-long GIFs (thats the whole purpose of Vine - short snips).

It would be very difficult to argue anything but the fact that the vines are only a tiny fraction of the whole copyrighted transmission. About a thousands of it (couple of secs vs. 7000 seconds or more). The published part is really minuscule, and clearly should be admissible under the fair use doctrine as its only a small part, juste like a quote of a book.

claiming copyright infrigement, would be eactly the same as send DMCA take-downs aagainst a Youtube video like "mash-up of the 20 greatest One-Liner of cinema history" on the grounds that you have copyright over 1 movie from which 2 one-liners are mentionned. (It doens't matter if your movie suck, and the actors' quips are the only thing worth whatching in it).

or trying to out-law a twitter account which has cited the two most funny puns in a book. (even if the rest of the book is worth less than the paper it's printed on).

In the end, it doesn't matter if you got a bad deal and hold copyright on an other wise really bad work and only a few bits of it are popular. You got a bad deal that's your own problem. All the example I gave, and the goal vines are in the same situtation: they are tiny fraction of a work. Showing them is definitely not the same as revealing the whole work (which is several thousands of times longer) and should be covered by fair use exception on lenght grounds only. If your local jurisdiction doesn't allow it, your law is broken and needs fixing.

(or you should sue for ccopyright infringement anyone who blurts out spoilers of book and movie plots)

Comment: It's the law. (Score 5, Insightful) 338

by DrYak (#47675283) Attached to: Berlin Bans Car Service Uber

It's just taxi lobbyists pushing back...

It has nothing to do with lobbies, or taxis reacting to Uber.

It's simply the law, and this law is much older than Uber itself.
It has always been there and still needs to be applied, even after Uber appears.
(Some other countries like Switzerland have similair laws).

If you transport people professionally (no mater if you're some minister's chauffeur, a taxi driver, working as a bus driver in the public transportation service, or simply driving a minivan with more than 7 passengers) the law requires that you have a special driving license and insurance companies require that you subscribe a different type of insurance policy (insurance is mandatory in EU).
Uber is note immune to the law. People get money to carry people around with Uber, they must therefore follow the state law.

This is not taxi drivers protesting against Uber because it's competing with them, it's simply the city ruling that Uber needs to play by the same rule like everybody else.

Comment: Bad example (Score 1) 105

by DrYak (#47633163) Attached to: IBM Creates Custom-Made Brain-Like Chip

Let me know when a computer can "see" with a pair of cameras. Identify an object heading toward the cpu(not just the cameras) and adjust its motors to dodge the incoming.

That actually do already exist.
It's a car's collision avoidance system.
It's already standard option from some manufacturer (e.g.: Volvo) (and should become mandatory in EU somewhere soonish).

Some like Mobileye rely entirely on camera, while other are integrating other sensors in the mix, like radar, infra red lasers, etc.

But yeah I see your point: complex task require complex network, way much more than this chip.

Comment: We need Mailvelope as an HTML standard. (Score 1) 175

by DrYak (#47633015) Attached to: Yahoo To Add PGP Encryption For Email

The Mailvelope Plugin - - already does that: encrypt webmails a la Gmail, Yahoo, Hotmail or your own Roundcube etc.. It does so in-browser, obviously.

The best would be it for such thing to be an actually HTML5 extension.

Gmail, Yahoo, Hotmail, etc. just flag which "TEXTAREA" tag contains the message body (or a greasemonkey script does it for them it they don't support it yet) and then the in-browser functionnality handles the encryption/decryption, completely outside of the reach of the webpage and its javascripts.

Comment: Hand it to the browser (Score 1) 175

by DrYak (#47632979) Attached to: Yahoo To Add PGP Encryption For Email

Except if only the browser it-self is exclusively in charge of the decryption/encryption.

The browser does the job, and all the webpage and associated javascripts ever see in the TEXTAREA is exclusively an encrypted stream.

That should be done in a plug-in, or even better: in a complete standard way - add it as an extension to HTML5.

Comment: Metadata protection (Score 1) 175

by DrYak (#47632927) Attached to: Yahoo To Add PGP Encryption For Email

Your e-mail metadata headers every bit as private as the address and the return address you write on a letter you send via the USPS.

...which in real life should only be used by the postman handling the delivery of the mail and shouldn't be mined by some 3rd party.

That is more or less doable (either server-2-server encrytion for the simplest form, or messaging over a tor-like network for the best protection) but has nothing to do with PGP.

PGP is about protecting the content (i.e.: it has nothing to do with the address written *on* the parcel handled by the postman. It's more like the content of the parcel being a safe box which can only be opened by someone having the key corresponding to the padlock on the safe box).

Comment: Metadata is irrelevant to TFA (Score 1) 175

by DrYak (#47632879) Attached to: Yahoo To Add PGP Encryption For Email

web-of-trust encryption (like PGP, and like the GnuPG implementation) is about encrypting the *BODY OF THE MESSAGE*.

I.e.: everything that comes after the subject is encrypted in a way that only the 2 end-points (author and recipients) are able to decrypt.
Without encryption, the content of an e-mail is as secure as a post-card.

Everything that comes before the subject, i.e.: all the headers that form this juicy "metadata" that the government wants, needs to be also readable to all the middle-men standing between the 2 end-point and who are in charge of distributing the mail. (e.g.: with paper mail, the postman needs to also see the address, otherwise he can't deliver it) But only to those in charge with the actual delivery (e.g.: only the postman sees what's written on the outside of the envelope. You don't want the gardener to keep a list of whom you're writting to).

That is encrypted by a completely different layer: it's the server-to-server encryption (things like the SSL and STARTTLS addition to IMAP/STMP/POP) which are in charge of keeping the metadata from beeing scoped.
But then you need to trust every server that your mail goes through (i.e.: you need to trust that none of all the various postmen who'll handle you mail is actually an undercover NSA spy posing as a postman) and you need to trust their security implementation (i.e.: that the postman delivering your mail pictures isn't clumsy and won't accidentally break your envelope and spill your nude picture on the ground, just right at the moment when a spy is around) (saddly, my comparison sucks: real world postmen aren't so clumsy, but real world cryptography is complex, and it's dead simple to bork something somewhere and leak secret information).

So yeah, metadata are important to protect too, but that's completely ireelevant. That remains instead for future discussion.

Note: perfectly safe messaging including secure metadata would require completely different infrastructure. Something like messaging over a tor network, instead of using a network of mail relay servers.

Brain fried -- Core dumped