[1122334455667788]

Okay, let's try this. Maybe this pic will explain it better.
Attachment 417466
I hope that picture attached correctly...

Again, I should re-iterate that my opinion is while this is a weakness that can, and should, be easily fixed, it still seems more secure than other common methods. Numerical unlock code (iOS, 4 or 6? digits now) or pattern unlock, if I can see you doing it as would be required by this Picture Password hack, I can guess your unlock 1st try! Unlike the Picture Password which would still likely require multiple tries, depending on your luck.

Nice slide!
I'll just add that while you can technically pick any number to line up while "hacking", using the same number as you observed in the spot will make it much easier to align the grid perfectly.

[bobshine]

Hey, aren't you the guy who won this contest already a few years ago?

Posted via CB10 on Z10

OMG! You have great memory!

Yeah I won't the contest... but it was an educated guess. I knew that he wasn't putting his thumb in the number and I just watch which number were the most probable. Then I guessed

[werkregen]

It depends on the attack vector. If someone is going to see/record your "password", no matter what it's form is, enough times, they're going to break your algorithm. The vulnerability lies in human nature most of the times anyway.

I assume these authentication methods are made to withstand 3-4 attacks at most. It's difficult to make something easy to do and hard to break at the same time.

[spARTacus]

OMG! You have great memory! .....

Haha indeed but credit goes to google, not my memory. While looking around for some deeper level description details about picture password I came across the old contest and saw your handle as the winner, then noticed it again in the thread. Oh the irony. I couldn't resist but to somehow mention it.

Posted via CB App for Android on Tab4 (interim Playbook replacement)

[1122334455667788]

t's difficult to make something easy to do and hard to break at the same time.

And yet simply randomizing the distance between numbers would make this sufficiently hard as to be near impossible in 5 tries, without really being any more difficult.

[Richard Buckley]

And yet simply randomizing the distance between numbers would make this sufficiently hard as to be near impossible in 5 tries, without really being any more difficult.

Only if they also did not allow cancellation to discard a pitch setting. Even if the separation was truly random there are going to be limits on the upper and lower boundaries of the pitch. If an attacker can still cancel out of an attempt with a pitch not close enough to the one observed the situation is the same. The pitch the attacker uses doesn't have to be identical to the one observed, just close enough so that the real number is within the error radius of the real location.

LeapSTR100-2/10.3.3.2205

[werkregen]

And yet simply randomizing the distance between numbers would make this sufficiently hard as to be near impossible in 5 tries, without really being any more difficult.

Maybe I can't wrap my mind around this. Wouldn't assymetric spacing make it easier to guess

Posted via CB10

[1122334455667788]

Only if they also did not allow cancellation to discard a pitch setting. Even if the separation was truly random there are going to be limits on the upper and lower boundaries of the pitch. If an attacker can still cancel out of an attempt with a pitch not close enough to the one observed the situation is the same. The pitch the attacker uses doesn't have to be identical to the one observed, just close enough so that the real number is within the error radius of the real location.

LeapSTR100-2/10.3.3.2205

Okay good point. It would still make it a lot more time consuming to find a similar grid.
They really should be able remove the ability to request a new grid. Even just having an option to have every grid generation count towards the unlock attempt total would be nice.

[Polt]

I kinda get what ur saying op, ur saying that the number grid is in cycle, and by pinpointing one specific number you might get lucky enough when the system generates another grid with the number you choose and the number that the owner choose in the same position, but I'd say that's still a very slim chance, as the distance between the numbers also change.

Posted via CB10

[1122334455667788]

I kinda get what ur saying op, ur saying that the number grid is in cycle, and by pinpointing one specific number you might get lucky enough when the system generates another grid with the number you choose and the number that the owner choose in the same position, but I'd say that's still a very slim chance, as the distance between the numbers also change.

Posted via CB10

No I don't think you get it.
The probabilities have already been discussed, and they are well above slim chance territory.

[Polt]

No I don't think you get it.
The probabilities have already been discussed, and they are well above slim chance territory.

Believe me i do, I'm just saying that the distance between the numbers also change, so you'd have to cycle enough times to make sure the grid is right, then between all that cycling you'd have to at most cycle ten times the get the right number. Which I think multiplies the chances of trying.
And also the guy who is trying to crack the phone must know how bb's picture password works, and he'd have to think hard enough to get the theory, so I'd say my phone is pretty safe from. At least 99.9% of the people around me.

Posted via CB10

[Richard Buckley]

Okay good point. It would still make it a lot more time consuming to find a similar grid.

I would have to see the math on this before I could agree. I suspect that the 8 column grid is as fine as one can go before you start getting two numbers within the error radius, and 5 is as course as it can go before the selected number is too frequently not visible on screen.

They really should be able remove the ability to request a new grid. Even just having an option to have every grid generation count towards the unlock attempt total would be nice.

I have thought of this too, but again would have to see the math, which would be much more complex than what I've posted so far. Just like changing from digits to letters sounds like it would have a large impact until you do the math, making this change might not give enough of an improvement to justify the additional problems users would have. Especially since this is supposed to be a simple system.



LeapSTR100-2/10.3.3.2205

[EFats]

Believe me i do, I'm just saying that the distance between the numbers also change, so you'd have to cycle enough times to make sure the grid is right, then between all that cycling you'd have to at most cycle ten times the get the right number.

Nope, maybe too many messages to follow, but the one trick that helps is that you can cancel your attempt (power button or Cancel) to generate a new grid without using up your tries.

I think disabling cancel option is problematic due to accidental swipes and pocket unlocks.


Posted via CB10

[Q10Bold]

I cant reproduce it.

#soon , later, later this year, fiscal year, commitment,...blablabla ~J.BlaBla Chen

[markus2107]

I tried to reproduce it, too. But I couldn't.

In my attempts I also got confused about what space I memorized exactly. Already this isn't too easy, because on my Z10 it just takes about 1–2 mm to place the number wrong.

Then I also didn't see any weakness in the random number pattern. I don't think there are just some patterns getting cycled, too. Can't say this for sure, but it would be incredibly foolish, to integrate PicPas this way.

[sir mictol]

Am sorry, have tired this countless times, its absolutely impossible because, when you move your hand to set your password to the location, the "grid " changes too. The grid that comes on display is going to be different from the one that appears on the location when your no and your location is correct.

To repeat that particular grid is next to impossible in 5 tries. I understand what the op is saying but it wouldn't be a problem.

Like me, my location for the correct pattern is upper left and I start moving the grid from lower right of the screen, someone looking at the screen when I do that will most definitely be confuse. That's my take on this.

Posted via CB10

[1122334455667788]

For everyone claiming it's impossible, please reread this thread. It has been explained multiple times now.
I can't be bothered to explain it again.

If you want to take the stance that it doesn't matter, or that "vulnerability" is to strong a word, that's fine. But please don't keep claiming the OP was completely wrong.

[spARTacus]

I agree.

Posted via CB10 on Z10

[1122334455667788]

I think disabling cancel option is problematic due to accidental swipes and pocket unlocks.


Posted via CB10

Remember in the unlikely event that you end up using all 5 tries you still get to use the regular password/pin.

Still, I think it would be preferable if you could cancel but it would save the grid layout (ideally even the current selection of random numbers) and restore it next time. I'm not sure if this is technically possible or not.

[Richard Buckley]

Am sorry, have tired this countless times, its absolutely impossible because, when you move your hand to set your password to the location, the "grid " changes too. The grid that comes on display is going to be different from the one that appears on the location when your no and your location is correct.

To repeat that particular grid is next to impossible in 5 tries. I understand what the op is saying but it wouldn't be a problem.

Like me, my location for the correct pattern is upper left and I start moving the grid from lower right of the screen, someone looking at the screen when I do that will most definitely be confuse. That's my take on this.

Posted via CB10

If the grid size is wrong, press cancel and try again until you get the correct grid size.

LeapSTR100-2/10.3.3.2205

[Thud Hardsmack]

Remember in the unlikely event that you end up using all 5 tries you still get to use the regular password/pin.

Still, I think it would be preferable if you could cancel but it would save the grid layout (ideally even the current selection of random numbers) and restore it next time. I'm not sure if this is technically possible or not.

If the grid size is wrong, press cancel and try again until you get the correct grid size.

LeapSTR100-2/10.3.3.2205

On my Z10 I get the same grid if I press Cancel and try again, and it counts against my unlock attempts. The grid changes after a successful unlock or reboot (on mine).

[spARTacus]

On my z10, if the grid comes up and if I do not try it but instead hit the top button or cancel, then I'll almost always get a different grid the next time. On my z10, it never counts against my attempts if I do not try it but instead hit the top button or cancel.

Posted via CB App for Android on Tab4 (interim Playbook replacement)

[Richard Buckley]

On my Z10 I get the same grid if I press Cancel and try again, and it counts against my unlock attempts. The grid changes after a successful unlock or reboot (on mine).

On my Leap and Z10 the grid is selected at random each time even when I cancel. Of course sometimes that means you get the same grid several times in a row. And cancelling doesn't count against the 5 tries.

LeapSTR100-2/10.3.3.2205

[Thud Hardsmack]

On my Leap and Z10 the grid is selected at random each time even when I cancel. Of course sometimes that means you get the same grid several times in a row. And cancelling doesn't count against the 5 tries.

LeapSTR100-2/10.3.3.2205

I was on 10.3.2 until just a few minutes ago, now that I'm on 10.3.3.1463 it works like yours. So it would appear we stumbled across another hidden update.

[spARTacus]

I am still on 10.3.2 on my z10. All my earlier comments and observations were framed from my perspective of 10.3.2 on my z10.

Posted via CB App for Android on Tab4 (interim Playbook replacement)