Uncle, You Don’t Make Sense

I have this uncle who’s in his early fifties. We had a family dinner a few weeks ago and it was the first time I saw him since the pandemic. The dinner was at a fancy restaurant and they checked that we were all vaccinated. During the meal, I found out from my cousin (his daughter) that he forged his vaccine status to get in. We were both peeved and couldn’t comprehend why anyone would do something like that. I gave him a cold look for the rest of the night and was stewing about it for the next couple of days. I found out a week later that he got Covid and infected the rest of the household.

I think I finally understand why he did it. Actually, I feel a little sorry for him now.

Natural-born storyteller

I’ve seen pictures of my uncle when he was 3 years old at my grandma’s house. I imagine the world around him those days was very straightforward. It all made sense. He didn’t have to deal with abstract concepts like geopolitics or interest rates. My uncle used what he saw around him to build a coherent and intuitive picture of the world. He needed this associative coherence since it was the stable interface to the dizzyingly rich sensory world he found himself in.

As a kid, the few pesky details in reality that he couldn’t see didn’t really matter. He could just make up a story about the missing bits and navigate life based on what he did see. One time his sister got Chickenpox and he thought the red spots were bite marks from a spirit that was angry with her. His mom played along and told him to stay far away from any kid who also got bitten. He listened to her though, and so never got Chickenpox as a child.

It’s worth noting that none of this was done consciously. My uncle wasn’t conflicted about making up stories to fill in the parts of reality he couldn’t see. He understood his world perfectly and walked around with confidence. What he understood to be reality applied seamlessly across what was real and what was made up. He knew almost nothing, but he didn’t realize it.

What he sees is all there is

As a child, my uncle could be excused for assuming that what he sees is all there is. Back then, his mind made use of his eyes in a straightforward way. It’s like he was born with a camera that gave him a personal, context-rich feed of the world. Sure, his camera was a little wonky and might have been fabricating some of the details. But those bits didn’t really have any bearing on him anyway. He trusted his camera enough to get him through the day. Trying to second-guess what his camera was telling him all the time was exhausting and hardly ever worth the effort.

The way my uncle operates his camera is radically different these days. The camera is no longer tethered to what he experiences in person and must function in a much bigger world. In this expanded reality, there are many more essential details that he needs to know, but he’ll never get a chance to experience in person. Getting the most important bits wrong can lead to a life of unrealized potential or worse.

To get things right, he must depend on other camera operators now. His camera screen is no longer just a simple viewfinder of what’s in front of him. It’s connected to the Internet and downloads feeds of what other operators are seeing on their screens.

When he’s working off his own experience of the world, he sees his personal activity feed. His screen shows him a rich detailed stream from his own aperture. A lot of the time though, he has no personal experience that he can rely on. He imports feeds from others and tries to re-create their experience using these lower-resolution feeds.

Here’s where my uncle gets in trouble. He’s still using the same janky old camera. He instinctively treats this new Internet-enabled screen like the one he used as a kid. He unwittingly makes up stories about the bits he can’t see and takes for granted that it’s all true.

Transmissions are lossy

I was tempted that night to lecture my uncle about how vaccines clearly save lives. I had looked at the research and I was convinced. To send him something from my own camera, I would use words to create a transmission. This note you’re reading is a transmission of what I see on my personal feed, encoded in words. But keep in mind that words are leaky abstractions. What I write and say can only be an approximation of all the detail I’m trying to convey. There will always be gaps between what I mean and what can be captured in the words I use. Any set of words I choose will chop away some different bit of reality.

This gets tricky when my uncle inadvertently fills in the gaps to autocomplete the words. When he receives my stream of words, he’s going to autocomplete the missing context to come away inferring something. It can feel like this something was absolutely what I was trying to say, even if it wasn’t. There’s no error message that pops up and tells him to stop trying to form an opinion since too much context was dropped in the degraded transmission.

If the feed coming to him is about a topic that he has personal experience with, he might interpolate the gaps correctly enough for the transmission to be useful. But if this is something new for him, the stories he’s making up about the details he doesn’t see might meaningfully distort what I was trying to say.

The primacy of coherence

In fact, the less he knows about the topic and the more degraded the transmission, the more confident he’s going to be that he’s understanding it correctly.

The confidence he feels isn’t determined by the accuracy of what he knows, but by the coherence and ease of the story that’s constructed from what’s available. It’s unsurprising that the confidence he feels measures coherence rather than accuracy or “truth” about a subject. There’s no way for any of us to be sure that there is a “truth”. The closest we can get is civilization-wide coherence about what we experience. The “truth” is a useful explanation that we all agree on.

A coherent explanation is indeed a decent proxy for truth in my uncle’s personal feed. If his camera comes up with a story that cogently explains all the rich detail that he encounters in person, he can rightly feel confident about this “made up” explanation. An explanation’s coherence also gets stronger the more familiar it gets as it repeatedly survives in-person testing. A very familiar explanation from his personal feed is going to be a very useful one. A strongly coherent explanation on his camera is intuitive and believable, gives him the confidence to act decisively and puts him in a good mood. It also takes less mental energy to process and has a clear evolutionary advantage.

When feeds get imported from other operators though, he can get in trouble. The lossiness of external feeds gives his camera many degrees of freedom to craft different persuasive stories. The more degraded the feed, the greater the number of possible stories.

His camera first latches onto an initial story that’s sufficiently cogent and easy for him to make sense of. Once it finds a good candidate, it works to expand its scope, hiding details and invalidating perspectives that would disrupt his coherence. For the gaps that are left, the camera projects his own abilities and experience of the world into them. There’s now a complete picture on the screen that makes sense. Remember that his camera isn’t trying to show him what is “true”. It’s just trying to show him a picture that’s maximally coherent. If necessary, it will sacrifice the “truth” to build this picture for him. Anything that challenges coherence will feel misguided and suspicious, and his mind will defer to what the screen builds for him. All of this happens without him realizing it.

So having fewer details available in the imported feed makes it easier to fit the details he can see into a coherent, compelling story that breeds confidence. It gets even more compelling if all his buddies around him are convinced of the story as well. In a perverse way, having less context makes it easier for him to get fooled into thinking he really understands what’s going on.

Through all this he’s exposed to a perfect storm. In our hyper-connected world, events that happen far away can have a big impact on his life, so he needs to understand them well. He learns about these faraway matters from remote camera operators who send him lossy transmissions. His wonky camera fills in the missing details by making up a story that’s coherent with the few bits that can be seen. The story is convincing, so he’s very sure he understands what’s going on. His confidence prompts him to share his insight with others and ignore those who think he’s wrong. He acts decisively on these beliefs only to find that something has gone awry and his instincts have failed him. My uncle doesn’t stand a chance.

Useful explanations disrupt coherence

If there isn’t truth per se, but only explanations, what makes one better than another? The Rabbit and Duck tribes in the earlier illustration each have a coherent explanation. What would I say if my uncle asked which tribe to join?

Deep inside, he wants the best health outcomes for himself. We all do. A useful explanation is like a how-to video that shows him what to do in the real world to get these outcomes. These are found in the blue and green quadrants in the image below.

Not all explanations that show up on his camera screen are useful though. Cameras look for coherence, and coherent stories feel good even if they aren’t useful. This is why a good movie can be so captivating, even when the audience knows it’s made up. It’s easier to tell a good story when you can hide the messy details.

Enjoying a well-crafted movie is all fine until the movie starts pretending to be useful. An explanation from the yellow quadrant is more like an infomercial. It’s compelling and believable, but isn’t going to get my uncle the outcomes he’s looking for. That night before dinner, he took some pills he bought from someone online. According to my cousin, this online “doctor” guaranteed he wouldn’t catch Covid with the pills. When he did get infected the week after, he apparently got pretty shaken up and sent the doc an angry email. It took him some time to recover from the dislocation he felt when his expectations got pulled out from under him. This is the dashed line pointing down.

Unlike an infomercial, a useful explanation needs to account for all the inconvenient but critical details since this is where the value accrues. Including all this nuance will, at least initially, make the explanation less coherent, i.e. the curve pointing up. My uncle’s camera will ironically be less confident and enthusiastic about a more useful explanation that comes from say, a medical journal. Exposure to all the messy details of how viruses and vaccines actually work will cause more confusion and uneasiness than following the simple truth from the next confident medical “expert” who comes along. Cameras prefer yellow to blue.

He’ll be even harder to persuade if the new explanation challenges the coherence of an existing image that’s dear to him.

My uncle never graduated college. A few years ago he got a degree in Ayurvedic medicine from a school offering part-time courses. Ever since then, he’s convinced that he makes better medical decisions than the rest of us. Telling him outright that his vaccination theories are absurd will threaten to invalidate who he thinks he is and what he thinks he’s good at. Having his self-image forcefully re-calibrated can cause tremendous strain and inner turmoil. His camera will naturally hide dissenting voices, and reinforce his current view of the world as more proven and reasonable. To get better outcomes, he’ll need to override his camera and seek out less coherent, but more useful explanations. Overriding the camera all the time is hard work though.

Camera on autopilot

Soon after he got his part-time degree, my uncle found a group on social media discussing the latest medical treatments. He told my cousin that he joined the group to chance upon new remedies and keep up to date after his course ended. He couldn’t tell if anyone in the group actually worked in the medical industry, but it didn’t really matter. This was a convenient, serendipitous way for him to stay current. It was something he was just doing on the side anyway.

An algorithm decides what content from the group pops up on my uncle’s camera screen. This algorithm is designed to amplify and spread coherence. It’s what cameras are hungry for after all. Since useful explanations are prone to disrupt coherence, it is much harder to come up with a coherent explanation that’s also useful. Most of the content that’s created out there is going to be yellow rather than green. As the algorithm is yellow-green colorblind, the group feed that it assembles is going to skew yellow.

My uncle could override his camera and check that each post was actually useful and not just coherent, but this would be exhausting. Focused attention is a limited resource and he’s just scrolling casually after a long day at work. Leaving his camera on autopilot most of the time is both reasonable and unavoidable. He needs to save deep focus for more critical tasks. A camera on autopilot can lead to a troubling path dependency though.

What shows up on my uncle’s camera matters. Like screen burn-in, what he sees on his camera screen biases whatever appears after it. Even if he disagrees with the post he’s scrolling past, or plans to ignore the subject, he’s going to get mentally anchored to those frames. After a tiring day at work, he’s not going to have enough mental energy to resist forming an opinion about something he just read. If he’s constantly thinking about a topic only because it happened to pop up in his group feed, his camera has already been hijacked by the algorithm. All of this can cascade and compound in unpredictable ways.

This is a contaminated version of serendipity. When consuming recommendations from his social media feed, serendipity draws from a pool that’s stained yellow. Without curating the sources of feeds being imported, his camera on autopilot can be led down a path that ends in poor outcomes.

Choosing the right operators

To get exposed to more useful explanations, my uncle will need to be picky about the camera operators he’s getting feeds from.

Most operators mean well. A coherent explanation they produce from their own in-person experience is most likely going to be useful. Importing content from their personal feeds will be fine as long a my uncle corrects for the lossiness of the medium.

It gets much trickier when an operator is relaying an already distorted feed received from someone else. If the operator in the middle doesn’t understand how cameras and lossy transmissions can misrepresent what was meant, they’re going to confidently pass along explanations that are a lot less useful than they think. Reverse-engineering what went wrong at which point of the chain isn’t easy. Choosing high-quality operators will matter.

My uncle comes across countless operators online in the many lists and groups he subscribed to. He isn’t going to be able vet every one of them in person so he’ll need to gauge operator competence remotely. This can be challenging. Like the online vendor he bought the pills from, a malicious operator can craft coherent narratives that keep getting picked up by the algorithm. Repeated exposure makes them seem more coherent and trustworthy. The medical “experts” who get pushed to my uncle’s screen are typically the operators who want to be heard, not the ones he should be listening to.

One option is to evaluate an operator by the epistemic quality of their feed, but this isn’t easy. It’s going to require a lot more time and effort, and my uncle will need to take his camera off autopilot. It’s an uphill battle for someone just trying to get by and do his best in a media environment that preys on cameras.

I wasn’t sure if knowing any of this would change my uncle’s vaccine skepticism, but I still wanted to let him know how his camera was being exploited. And then it hit me. I have a camera that’s easily exploited as well, we all do. All operators have cameras that suffer from yellow-green colorblindness. But I also see myself as a rational person who makes good decisions with the best information out there. My camera was focused on my uncle and took itself out of the frame to protect the coherence of my own self-image.

In the same way, telling my uncle a story about his camera will have limited value. His camera will gloss over the story and move the focus away from him to safeguard coherence. So instead, I’m telling his story to you.

Thanks to Cedric & Visa for reading drafts of this note.