Adaptive Segmentationmicro-segmentation September 2, 2016

Cybercrime in Russia, algorithmic bias and patent trolling

Nathaniel Gleicher,

Four things I'm reading this week:

4 Things In Cyber

  1. Algorithms Are No Less Biased Than the Humans That Design Them: The last week has been full of stories about algorithmic bias. The marquee topic was Facebook’s troubled effort to improve their news feed by turning more of the process over to algorithms. But there were also several other very thoughtful analyses of how algorithms impact bias. I’m reading:

  2. Tracking Down a Cybercriminal Group: Fascinating story from Kaspersky of tracking down an active Russian cybercrime organization. This story bears many of the hallmarks of other organized cybercrime groups I have seen over the years – specialization, corporate culture, discipline, and sophistication. My favorite part: head-hunter-organized job interviews where the applicant was asked whether they “had strong moral principles. Those who didn’t got the job.” Imagine the management headaches from an entire workforce that answered “no” to that question. I’m reading: “Lurk: Seek and destroy
  3. Friendly Neighborhood Robots Delivering the Mail?: Swiss Post is testing six-wheeled robots that will deliver mail packages in three cities. I can already foresee the traffic jams on the sidewalks, with the tiny robots beeping apologies as they thread their way through the crowds. I’m reading: “Switzerland enlists robots to help deliver mail

  4. Cybersecurity Researchers and Short-Seller Learn a Troubling Lesson from Patent Trolls: Earlier this week, a cybersecurity firm partnered with a short-seller to publicly allege security vulnerabilities at a hospital. The short-seller bet that the announcement would hurt the hospital’s stock value, and the cybersecurity firm actually got a share of the profits when it did. The hospital has since disputed the research in question, and we’re now left with a fight over money instead of security. This approach is particularly troubling because the researcher gets rewarded for “outing” the company – not for improving security. They have no incentive to resolve the vulnerability before disclosure (best outcome for security) because that could reduce their payout from the stock drop. Scary echoes of patent trolling. I’m reading: “A new hacker money-making strategy: Betting against insecure companies on Wall Street


Adaptive Segmentationmicro-segmentation
Share this post: