3 ms·
1 in 1 million FAR (false acceptance rate) vs 1 in 50,000 is pretty misleading (as is Apple tradition). Do you think someone trying to hack into your phone wou
by mtgx 9y ago
1 in 1 million FAR (false acceptance rate) vs 1 in 50,000 is pretty misleading (as is Apple tradition).
Do you think someone trying to hack into your phone would shoot 1 million random pictures/3D profiles made from Facebook pictures at your phone, or do you think it's far more likely they will already start with your profile made from online pictures?
That will likely make the success rate even higher than with fingerprints, as it's significantly easier to get someone's photos than it is to get their fingerprints.
> Laughs were had, jokes were made but the underlying message was that Face ID isn't foolproof. Just like Touch ID. And PINs.
No, not "just like". There is a huge difference between most fingerprint authentication mechanisms and most face unlock mechanisms (at least so far). Most of them could be tricked with a 2D picture - including Samsung's latest. It's very annoying to see such a statement from someone like Troy Hunt. Plus, I have a hunch he'll be eating many of the words he wrote in a few weeks when Face ID will prove much easier to hack than Apple made everyone believe it will be.
- MBCook 9y agoDo you have a better metric? The author suggested it wasn't the best way to measure it but I'm not sure what would be better. "Most face unlock systems" is the point of Face ID. It's now most systems. It's supposed to fix those weaknesses. (We'll find out for sure when it's in people's hands)
- admiralpumpkin 9y agoPer Apple Face ID doesn't work with photos; TBD in real world.
- macintux 9y agoIf they created photorealistic masks to defeat the 3D mapping and those didn't work, I feel pretty confident in saying that photos won't work.
- MBCook 9y agoI think it would be really cool if it turned out that it could tell the difference between some pairs of identical twins due to seemingly imperceptible differences.
- macintux 9y agoI'm sure it's not 100%, but I'd bet it's close. Certainly by young adulthood the identical twins I've been around have been relatively easy to distinguish.
- MBCook 9y agoThat's sort of my theory. But I wonder how much of people's ability to distinguish identical twins is based on physical differences vs more subtle things like how they carry themselves/interact with the world.
- kccqzy 9y agoSome twins I've known deliberately create physical differences like different hairstyle so other people can recognize them easily without interactions.
- BoorishBears 9y agoIf you look any of the information on Face ID you'll see it has 3D mapping via a "Kinect-like"/"RealSense-like" IR dot projector.
- willstrafach 9y agoOnline pictures do not contain depth or other needed information. Regarding the false acceptance rate, they would not specify and compare to the already-known Touch ID FAR just for no reason. Logic behind the number is not that someone would actually try one million times (IIRC - just 5 failures will wipe keying material and force passphrase entry), but rather, they're saying derived data for Touch ID had 50,000 possibilities, and there are 1,000,000 possibilities for the Face ID derived data.