The Justice Department gets out its black highlighter; the Pulitzer board wants Trump's tax returns and medical records; Sam Bankman-Fried is not a good lawyer.
as an alum of one of the *other* Claremont Colleges, please continue to mention Claremont McKenna College in the ongoing capacity whenever the opportunity presents itself, please and thank you. (and happy holidays!)
In regards Gait Analysis. I was involved in a DARPA program to look at ways to authenticate people by various means, often without needing their active participation. This is useful for things like access control or a second layer after something like a password. Gait analysis was analyzed by one performer, as were things like “how you move the mouse” and “how you select things on your phone screen” - all of these have some amount of uniqueness per person. In general, they are sort of 80% accurate when you ask the question “Does this data match Person X” (the authentication question, versus, “who, among the millions of people, matches this data”). About 10% false positive (says it’s a match, but isn’t) and 10% false negative (says it isn’t a match, but actually is) - that’s adjusting the detection threshold for what they call “equal error rate”.
I’ll note that for gait, it’s not too tough to learn to match someone else’s gait (especially if you have the recording, and you’re wearing the gait analyzer), and similarly, it’s not too tough to make your gait distinctly different (a rock in your shoe helps). There’s two big families of gait analysis by the way. One relies on external sensors - video or radar; the other relies on measurements made by your cellphone (the same way they know how many steps you took, and how many flights of stairs you went up and down).
Mouse (or swipe/touch) movements is how sports and porn sites do “anti sharing of password” - all you need is a bit more “friction” in the transaction to seriously reduce loaning the pw to a friend, versus, say, sending a code in a text message that you have to enter, which is definitely friction. And it’s trivially easy to implement in either server or client side web code. (Captcha works similarly)
I think it’s worth pointing out that obstruction is a much more common charge and there is a lot more case law regarding it than concealing an alien. So it’s not surprising to me that the obstruction charge instructions were more lenient about what the judge specifically needed to know than the concealing an aliens charge, even though obstruction is the felony. Anything the United States doesn’t like is obstruction.
But if ICE exclusively goes around in unmarked vehicles, how are we supposed to KNOW we're towing a government vehicle?
as an alum of one of the *other* Claremont Colleges, please continue to mention Claremont McKenna College in the ongoing capacity whenever the opportunity presents itself, please and thank you. (and happy holidays!)
In regards Gait Analysis. I was involved in a DARPA program to look at ways to authenticate people by various means, often without needing their active participation. This is useful for things like access control or a second layer after something like a password. Gait analysis was analyzed by one performer, as were things like “how you move the mouse” and “how you select things on your phone screen” - all of these have some amount of uniqueness per person. In general, they are sort of 80% accurate when you ask the question “Does this data match Person X” (the authentication question, versus, “who, among the millions of people, matches this data”). About 10% false positive (says it’s a match, but isn’t) and 10% false negative (says it isn’t a match, but actually is) - that’s adjusting the detection threshold for what they call “equal error rate”.
I’ll note that for gait, it’s not too tough to learn to match someone else’s gait (especially if you have the recording, and you’re wearing the gait analyzer), and similarly, it’s not too tough to make your gait distinctly different (a rock in your shoe helps). There’s two big families of gait analysis by the way. One relies on external sensors - video or radar; the other relies on measurements made by your cellphone (the same way they know how many steps you took, and how many flights of stairs you went up and down).
Mouse (or swipe/touch) movements is how sports and porn sites do “anti sharing of password” - all you need is a bit more “friction” in the transaction to seriously reduce loaning the pw to a friend, versus, say, sending a code in a text message that you have to enter, which is definitely friction. And it’s trivially easy to implement in either server or client side web code. (Captcha works similarly)
I think it’s worth pointing out that obstruction is a much more common charge and there is a lot more case law regarding it than concealing an alien. So it’s not surprising to me that the obstruction charge instructions were more lenient about what the judge specifically needed to know than the concealing an aliens charge, even though obstruction is the felony. Anything the United States doesn’t like is obstruction.
Only two days into 2026 and Drake is being sued civilly for RICO. What a time to be alive.
Walk without rhythm and it won't attract the worm.
Which Heinlein story is it (the stone in the shoe)? I think I've read pretty much everything he wrote, but can't place that.
Any advice on how to pronounce "⬛⬛⬛⬛⬛⬛⬛⬛⬛⬛"?