Placeholder Image

字幕列表 影片播放

  • This is a footnote to the main video about mathematically protecting the privacy of individuals

  • when youre publishing statistics from a private dataset, like a medical study, or

  • a census, or whatever.

  • There are in fact two kinds of privacy violation that can happen from a survey, and theyre

  • qualitatively very different from each other.

  • The first kind is the direct breach of the privacy of an individual by somehow revealing

  • private information specific to them (like their birthday, or blood type, or Harry Potter

  • house), and this is the kind of privacy violation the main video focused on.

  • The second kind is an indirect violation of privacy via association with a group (like

  • how men are more likely to be overpaid, or how Slytherins are more likely to be evil,

  • or how overpaid men are more likely to be Slytherins…).

  • Of course, revealing trend-based information about a group is precisely the purpose of

  • doing surveys; we, as a society, want to know the expected lifespan of smokers vs non-smokers,

  • or the typical month in which professional hockey players are born.

  • But if a survey reveals that hockey players are more likely to have January birthdays,

  • then knowing somebody plays in the NHL gives you insight into a supposedly private piece

  • of information, and does so regardless of whether or not that player themselves participated

  • in the survey!

  • If we wanted to protect the privacy of individuals 100%, pretty much the only option would be

  • to outright prohibit all studies and surveys that use any individual information, whatsoever.

  • But then we couldn’t have representative democracies, or study diseases, or keep an

  • eye out for dark wizards coming out of Slytherin, or lots of other useful things.

  • So if you are going to do a study, the best you can do is to not violate any participant’s

  • privacy more than their privacy would have been violated if they hadn’t participated

  • in the study.

  • That is, the current wisdom is that it’s ok to reveal that NHL players are more likely

  • to have January birthdays , but it’s not ok to reveal the birthday of a specific player.

  • And of course, you’d have to reveal the January birthdays fact using a mathematically

  • guaranteed privacy protectionbut that’s what the main video is about.

  • And if you want to ensure you don’t have online information specific to you stolen

  • or published, I highly recommend using Dashlane, this video’s sponsor and a service/tool

  • that can greatly simplify and secure your online life (as it has mine) - with Dashlane,

  • every single site or online service you use gets a strong, unique password, and Dashlane

  • securely remembers them so you don’t have to.

  • Dashlane also (with your permission) auto-fills online address forms, credit card info, saving

  • you time and hassle.

  • It’s really nice.

  • You can get a free 30 day trial of dashlane premium (which also includes Dashlane’s

  • VPN) by going to

  • Again, that’s, and use coupon code minutephysics for 10%

  • off at checkout.

  • And I’d like to thank Dashlane for simplifying and securing my life.

This is a footnote to the main video about mathematically protecting the privacy of individuals


影片操作 你可以在這邊進行「影片」的調整,以及「字幕」的顯示

B1 中級

什麼時候可以侵犯隱私 (When It's OK to Violate Privacy)

  • 32 1
    林宜悉 發佈於 2021 年 01 月 14 日