Attention is Not a Currency
"Paying attention," a common metaphor, is misleading because there are different sorts of attention, and the relationship among them isn't reducible to numbers.
Before we get to today's main topic, some miscellaneous goodies…
The Ted Lasso Season 3 premiere on AppleTV+ was a yawn, a throat-clearing exercise reminding us of who the characters have been and not telling us a lot about where they're headed. Everybody is getting along so nicely that I worry who the villains will be. I'll give it another shot, but I'm skeptical.
Don't miss my friend Lou Paskalis' piece about how advertisers are complicit when they buy media on disinformation sources.
This Technology Review piece about how the changes in generative-AI-driven search go beyond Google and Bing is worthwhile. This adjacent piece on the new Microsoft Copilot is also interesting.
Here's a really nifty new ad format on LinkedIn (Turn your volume on.) It'll get annoying with overuse, though.
My thanks to my friend Claudia Batten for a just-at-the-right-time conversation that helped me think through this week's topic.
Happy Birthday to Helena Berens, my brilliant, kind, funny, and beautiful daughter.
You don't need to have read any other issue of The Dispatch to understand the piece that follows, but if you read last week's story about how culture is what's left over after quantification, then you'll see that this issue explores a similar theme.
Please follow me on Post and LinkedIn for between-issue insights and updates.
On to our top story...
Attention is Not a Currency
If you're in the Attention Business—and whether you're selling movies, cars, toothpaste, whoopee cushions, sex toys, health insurance, a ride hailing service, or a new ointment for that embarrassing rash, every business is in the attention business—then understanding that there's more than one sort or attention is a good first step towards getting the right sorts of attention for your business.
Attention creates experience. In Principles of Psychology (1890), William James wrote, "my experience is what I agree to attend to."
Context amplifies and changes experience so that two people with different contexts can experience the same moment in distinct ways. Here are two examples from entertainment, but the phenomenon holds true across the experience of many products.
Shakespeare's Romeo and Juliet: As 13-year-old Juliet is about to stab herself to death, she says something that changes the playgoing experience for the people in any audience familiar with Latin. "O happy dagger. This is thy sheath. There rust, and let me die."
The Latin word for "sheath" is "vagina."*
Playgoers who don't have a little Latin experience this tragic moment in an agonizing but uncomplicated way. For playgoers who do have a little Latin, that knowledge can dislocate them from the agony with troubling questions: Is Juliet making a pun as she is stabbing herself? Is Shakespeare making the pun through Juliet? (If Freud had Latin, then he must have enjoyed a heady, cocaine-fueled evening working through this one.)
The difference between how the playgoers with and without Latin encounter this moment comes from their different Experience Stacks.
An Experience Stack is the customer-facing counterpart to a company’s Tech Stack. A Tech Stack is the combo-platter of software and hardware that a company uses to create, manage, and track its products. An Experience Stack is the combo-platter of all the activities customers do over time with and around the things companies make. Customers improvisationally shift from context to context during any given experience, which is one of the key differences between analog (human) thinking and algorithmic (AI) thinking. It is easiest to see Experience Stacks with narrative products like movies and television, but customers generate Experience Stacks with all products.
Castle, Season 5, Episode 6: In "The Final Frontier" (2012), Beckett and Castle investigate a murder at a science fiction convention involving the cast of a long-cancelled TV show that resembles Star Trek. Director Jonathan Frakes, who played "Number One" (first officer William Riker) in Star Trek: The Next Generation, makes an early cameo and says, "I'm your number one fan" to Castle. Armin Shimerman, who played Quark in Star Trek: Deep Space Nine, plays a key role. The episode is chock full of references to other science fiction shows, including Firefly, a show that starred Nathan Fillion, who plays Castle.
If a viewer's Experience Stack includes any of these other shows, then in multiple moments awareness of those other shows erupts into consciousness, only to fade away as the plot gains momentum—like what happens to still water a while after you've dropped a stone into it.
Unlike in-jokes and Easter Eggs, neither Juliet's pun nor the abundant science fiction references in Castle depend on prior knowledge of Shakespeare or Castle: they depend on external knowledge of Latin and other TV shows.
Another way of putting this is that these sorts of references elicit a particular type of attention that doesn't fit into our common metaphor of attention as a kind of currency.
The two different kinds of attention.
We all use the phrase "pay attention!" when we're asking another person to focus on a task at hand, and the phrase suggests that we can measure all attention quantitatively, and that different sorts of attention are interchangeable like money.
They're not.
Different schools of thought talk about two general modes of attention: Automatic and Deliberate. We need both to make it through our waking hours without either despairing because our lives are racing past us or bursting because we have so many things to think about.
Automatic: this mode is fast but superficial. Behavioral Economist Daniel Kahneman calls this System 1, which is useful but lacks poetry. Some social scientists call this "schema-driven" processing. The early twentieth century Russian literary critic Viktor Shklovsky called this recognition. This is lantern attention: general, omnidirectional, casting its light indiscriminately.
Deliberate: this mode is deep but slow. This is Kahneman's System 2 and the "data-driven" processing of social scientists. It is "seeing" for Shklovsky, who thought the purpose of art was to make things we're accustomed to recognizing "strange" so that we can see them afresh. Instead of a lantern, this is spotlight attention, focused, unidirectional, zeroing in.
Experience Stacks rely on Deliberate attention.
There's an inverse relationship between depth and speed:
The more focused attention you give to something, the slower time goes. This explains why a two-day vacation can feel like a week, but a week can disappear and feel like a day. On vacation, you don't know where anything is or how to get anyplace, so you do more Deliberate thinking. The more Deliberate your cognition the slower your perception of time, the more Automatic the faster.
As we saw with the Shakespeare and Castle examples, we don't use the two modes separately: we improvisationally switch back-and-forth between them moment-to-moment.
It's hard to describe the relationship between Automatic and Deliberate thinking: there's a VENN diagram in there somewhere, but the overlap wiggles around like a boneless three year old trying to slide out of his mother's arms. Automatic thinking emerges from Deliberate thinking, but the relationship is more than quantitative.
The least bad metaphor I've been able to come up with so far is the relationship between—on one hand—water, flour, yeast, salt, and time, and—on the other hand—a loaf of bread. If attention were a currency, then you should be able to break a loaf of bread back down into its component parts, but you can't. The difference between the ingredients and the loaf comes from time in the oven.
The reason I'm unhappy with this metaphor is that it privileges the bread, the Automatic thinking, at the expense of the Deliberate thinking. Who cares about a bunch of ingredients? I just want something I can use to make a sandwich.
What's important is the baking.
Thanks for reading. See you next Sunday.
* The earliest recorded use of "vagina" to refer to female anatomy was 1612, but even if it wasn't a common term to the first audiences of the play, the association between sheaths and female anatomy dates back thousands of years. It's certainly available to Latin-knowing playgoers in the centuries since 1612.
You're right; attention is not a currency. It is the product, and as you point out there are different qualities of product. There's that superficial ("automatic") kind, the kind that might only last a handful of seconds (I'm remembering the famous Microsoft datapoint from some years ago that said we only pay attention for 8 seconds, which inspired a flurry of short-form TV ad experiments that went nowhere), and the deep and durable ("deliberate"), the kind that keeps us turning 652 pages of Harry Potter and the Half-Blood Prince to get to the last page.
In the media ecosystem, attention is a kind of agricultural product, fed and watered and sunned with certain kinds of content. The better the fertilizer, hydration, and sunlight the better the growth (to a point). Water it with Gatorade, you get Idiocracy...
I have some thoughts about this essay. Let me start with noting that Kahneman is a favorite of mine. I got to wondering about your speed/depth graph. You drew it as linear, but perhaps it’s not. Let’s look at some alternative lines and what they indicate about using the two modes.
You drew it as linear. That indicates that the user has a certain amount of effort (I’ll call it) and can divide it any way between the two modes. The line could also be concave, convex, or not a line.
You described it as inverse. In that case the line would be concave, using a picky mathematical definition. A sample graph would be a smooth curve through the points, (1,4), (2,2) and (4,1). If that is the representation, that indicates that there is less effort available when mixing modes. Is that so?
An alternative graph would be convex. That would indicate some kind of synergy if you could combine modes 1 and 2. That would be quite appealing, but is it really possible?
It may be that it is not a line at all. The most useful graph may be just the two endpoints. You hint at this: As we saw with the Shakespeare and Castle examples, we don't use the two modes separately: we switch back-and-forth between them moment-to-moment. If that is the representation, that means that the modes are hard to combine.
One personal experience for believing that the model is involves switching came at a international conference on mathematical education. When discussing understanding and proving math concepts that combine geometry and algebra, several mathematicians mentioned visualizing the concepts geometrically and proving them algebraically. In my experience, that indicates a switching process.
While I can’t claim this to be a carful analysis, I lean towards thinking the two modes as not being connected in a way represented by a line. A good metaphor would be helpful here, but I don’t have one.