Notes for season 3 episode 20, “Death Benefit“
3:19 Finch and Shaw leave $1 million (which we presume can’t be linked back to Finch) on the street, next to the drug dealers who were going to kill each other? Effective altruism.
6:10 The congressman is played by John Heard, the dad in Home Alone.
6:39 Does every article in the Washington Herald have exactly one comment? One of the one-comment articles appears at the start of the episode.
08:55 We briefly see Reese’s fake DHS PIV card (FIPS 201, known as Common Access Card in the US military) which is a card with and integrated chip which can be used to access secure computers, and a PDF417 matrix code on the front. One thing that came out after the OPM breach (the real life one, not the one in the last episode) was that the Federal Government was quite behind in deploying this 2005 standard for ID smart cards that could be used for multi-factor authentication. (Some UK banks issue a tiny device that lets you generate MFA codes for online banking using the card.) In 2016 it was pointed out that the “PIV” cards issued to Senate staffers had photos of IC chips, but no actual chip.
12:07 Garrison and Greer meet in front of The Garden of Earthly Delights, presumably on loan to a Washington gallery. The Prado Museum displays their Bosch triptychs with the panels at an angle, instead of flat against the wall as here.
19:29 The VIN number on the car service rental agreement has “G” as the 10th character, indicating a 2016-model car in 2014. (It’s not a valid VIN anyway, it’s 16 instead of 17 characters.)
29:04 “I was able to get into his phone through a key vulnerability in the SMS encryption.”
31:40 The twist: it turns out the Congressman was only opposed to government surveillance, and was secretly in favour of private-sector surveillance.
33:40 Finch determines that “it’s conceivable” that the Machine would instruct an operative to take a life in order to save lives. Well this is literally how the Machine was working with ISA for years prior to the previous episode.
But yeah, its quite the classic AU quandary. “An AI predicts that an individual, without malintent, will cause the deaths of many others – could you, without being able to interrogate the AI’s reasoning, kill that individual?”
In TV and movies the plot never suggests that killing people based on AI predictions is the right move…