They are "not a bank" in the same way Uber is "not a taxi service". It's all down to avoiding regulations.

Should'a said "doesn't run third party apps natively".

Current iteration as far as I am aware does not run apps natively, it is just a front end for your phone. In a generation or two of the API (hardware does not need to be updated) this will be addressed and we'll see.

That's it really...

Yup, that's a face. I'd recognize one of those things anywhere.

Really... when was the last time any of us thought Cisco was the best choice for a project?

Don't accept things on face value.

My understanding is that at least one of the victims has come out and said it was not rape, and that the facts that were presented were inaccurate. She has made claims that she was coerced into her statements.

Last I saw, Wikileaks was still alive and well.

I really doubt this is going to blow over.

Computer Fraud and Abuse Act. This is what they charge you with if you used a computer.

Then Espionage and Conspiracy. The CFAA only goes up to 10 years, I assume someone would want to make an example out of him.

Finally.

It's not like this is unprecedented. I don't know what's so special about Assange that they could not have done this a long time ago.

My guess on what's about to happen:

- Sweden interviews him and drops the charges.

- Assange steps out of the embassy and is immediately arrested.

- Assange is charged in the US and extradited within a few days.

End-user billing information should be accessible to the called party. If someone wants to front for someone else, they can assume the liability too.

They do it with the SWIFT system for banks, surely phones are simpler.

I should also note that I'm not doing an OCR based method -- I'm using a "fingerprinting" perceptual hashing method that instead looks at the entire picture of the card (similar to how Google's "Find Similar Images" function works)

Thanks!

Yeah, it could possibly be sped up a bit, but right now I'm doing a linear search for the nearest Hamming distance in a data set of about 25k cards (all of the MtG cards printed) -- if I were to optimize the Hamming search with a tree of sorts (similar to the algorithms used for spell-checker algorithms) I could possibly speed it up, but no need to prematurely optimize things at this point.

Maybe 0.1 seconds, on average? Not quite fast enough for 30 fps, but close enough that the lag isn't really noticeable to the user.