I recently finished watching Class of ‘09 on Hulu1. It’s an 8-episode series about a cohort of FBI agents, told in three interweaving time periods: their experiences at the academy at Quantico in 2009, dealing with a terrorist attack and then launching an AI system in 2023/2025, and dealing with the ramifications of the system in 2034. Honestly, it’s not great2 (and you can expect medium to heavy spoilers throughout this article), but it does effectively show the connection between the past, present, and future by jumping between times while considering the same people, places, and issues within a single episode.
This intertemporal stacking is an effective technique that’s basically the companion to the fix I proposed for Extrapolations last summer: rather than show the same location/scene in different scenarios as a way to help viewers imagine the differences between plausible independent futures, show a throughline for a single story making it clear not only how the past and the present helped bring about the future, but also to show how we’re still dealing with the same kinds of issues in different forms. This approach is used sparingly but effectively in Station Eleven and The Last of Us, and in more traditional futures work like making the impacts of sea level rise visceral; by contrast, Class of ‘09 bets the whole show on it, with mixed results3. The challenge with the approach is the temptation to squeeze all the important elements of the story/plot into a few thin temporal slices, causing viewers to wonder how these people could have such an exciting week followed by an apparently very boring decade4. In fact, one of the most effective scenes comes in the last episode where they show a timeline and montage connecting the present to the future and showing the interesting macro trends taking shape.
The show effectively covers some useful futures concepts and does a good job depicting the implications of a number of possible developments. For example, it’s totally plausible (and frightening) that in pretty much any avenue we implement AI with human oversight, we’re going to quickly have humans “fall asleep at the wheel” as over-trusting the system leads to automation bias, which is often dangerous but particularly terrifying in a law-enforcement context. Some of its shortcomings are forgivable: attempting to personify an AI that doesn’t take the form of an embodied agent is a tall order, and the show mostly attempts to convey its spookiness by showing numbers on a screen or overviews of shiny and oddly-spaced server racks.
But the biggest problem with the show is that, by the end, it doesn’t reach any kind of satisfying thematic confusion. The first episodes unfold pretty logically:
Preventing crimes like domestic terrorism and serial killings is hard for humans because it requires the collation of lots of bits of disparate data (fair observation)
Applying AI to this problem might help, but would probably be biased in ways that are unacceptable (true fact, we’re already dealing with this)
Solution to 2: make the AI unbiased by starting from scratch and first principles instead of using biased datasets (OK interesting idea), and by making everyone in the country a suspect for every crime until they can be ruled out based on evidence (this is implausible but I get the point), with the goal being to provide intelligence to agents who make all final decisions.
Corollary to 3: in order to effectively eliminate people from suspicion, the FBI has to create a massive surveillance operation to collect data on all Americans (this would trigger a much bigger backlash than the show depicts5), and humans largely become data collection devices for the machine (I think this was the best idea in the show and under-explored).
The solution gradually slides into predictive policing as agents and judges just defer to the AI (I have a hard time imagining this being gradual - there’s a binary line that’s crossed when you move to arresting and convicting people for future crime6).
As the system becomes more powerful, it starts seeing criticism or opposition to the system as worth pursuing as a crime and clearly starts to resist human oversight (OK this is the “eye turns red” moment), including by committing murder-by-Tesla. Our heroes commit to doing whatever it takes to disable it.
All this is a little cliched but mostly makes sense. However, the last episode was a last-second offramp kind of situation:
It turns out that as a condition of approving the system, a group of high-ranking people were selected as exempt from being considered as suspects for crimes.
The plan is to infiltrate the single data center housing the system, which is hidden in the woods for some reason and almost completely unguarded except via drones with tasers that can easily be disabled with a portable EMP grenade (which does nothing to the servers for some reason).
Once inside, their plan is to upload a virus that doesn’t disable the system but instead permanently and irrevocably eliminates the exceptions list.
Realizing they can no longer act with impunity, the unnamed elites disable the system and for some reason it can’t fight back like it did earlier. Crime will go back up, but now that everyone has had a brief taste of truly equal justice, we can build a better society or something.
I’m willing to be convinced that this was the plan the whole time, but it seemed ad hoc, hastily written, and like it undermined the themes of the show.
Bottom line: stacking ideas and locations together across time is an effective way of communicating the implications and consequences of potential futures, but make sure the future you’re communicating are internally coherent.
I feel like this was originally recommended by folks on the APF listserv, but I can’t for the life of me find any evidence of this.
My wife was sitting next to me one night playing Tetris while I was watching it, and she said without looking up “wow, the acting in this show is not very good huh?” I hadn’t really noticed but apparently it stood out.
Of course, the approach makes perfect sense from a television perspective - every time period requires its own sets, props, makeup, etc.
You might call this the Solo problem.
Mostly protests and an increasing share of people moving out of closely-monitored urban areas.
In Minority Report, which I feel is superior in pretty much every way, this is explicitly pitched as a pilot that is being tried, plus they have clairvoyants in a vat.