✶Explainer01:00
How attention spans fell from 2.5 minutes to 47 seconds in 20 years
Dr. Gloria Mark has measured on-screen attention since 2004, first with stopwatches and later with computer logging. Average time on any screen before switching fell from about 2.5 minutes in 2004 to 75 seconds around 2011 to roughly 47 seconds today. The median is even starker: half of all measurements show attention under 40 seconds.
- 2004: about 2.5 minutes per screen, measured with stopwatches
- 2011: about 75 seconds as computer logging arrived
- Last 5-6 years: a steady state near 47 seconds
- Median measurement is under 40 seconds
- Findings replicated by other researchers within a few seconds
“in the last 5 6 years, we find it's about it's reached a steady state of about 47 seconds on any screen before switching.”
“half of all of our measurements show people's attention to be less than 40 seconds.”
#attention#research#screen-time#distraction
✶Explainer26:30
The real cost of task-switching: more errors, switch costs, higher blood pressure
Switching attention carries measurable costs. Studies of physicians, nurses and pilots show more errors when they switch, including prescribing errors. Every switch incurs a switch cost, the time to reorient. Worst of all, multitasking physiologically raises stress: blood pressure and heart rate climb with faster switching, and people also report higher perceived stress.
- More errors under multitasking, including physician prescribing errors
- Switch cost: reorientation time on every task change
- Blood pressure and heart rate rise with faster switching
- Perceived stress reports rise too, so all measures agree
“The faster the attention switching, the higher is the stress measured.”
#multitasking#stress#productivity#errors
✶Explainer40:30
Algorithms have two ways to win: predict you better, or make you easier to predict
Drawing on Stuart Russell's Human Compatible, Williamson explains that recommendation algorithms are prediction engines maximizing clicks. They can either predict your preferences more accurately, or reshape your preferences to be easier to predict, for example by nudging people toward more extreme, low-nuance views. The result is a two-way street between platform and user.
- Algorithms optimize for click-through and watch time
- Path one: predict existing preferences more accurately
- Path two: change preferences to be easier to predict
- More partisan, extreme users are easier to predict
- Even developers can't fully see inside these black boxes
“The other way is to change your preferences to be easier to predict. So, there is a two-way street going on between AI algorithms and…”
#algorithms#ai#social-media#polarization
✶Explainer42:00
Audience capture: how creators get reprogrammed by their own followers
Williamson adds a third party to the algorithm-user loop: the creator. Through audience capture, a creator keeps feeding the audience what it responds to rather than what feels creatively right, and the audience's clicks tell them what to make next. Over time this produces an ever more caricatured, one-dimensional version of the creator.
- The creator sits between the algorithm and the audience
- Creators drift toward feeding audiences red meat
- Clicks reshape what the creator makes next
- Result: an increasingly extreme, caricatured persona
“a creator starts to continue to feed red meat to his audience... as opposed to what to them feels creatively right.”
#audience-capture#creators#social-media#incentives
✶Explainer43:30
Why our social nature, status and identity keep pulling our attention online
Mark explains that much distraction is rooted in social nature. We chase social capital and want to appear in good standing, so we respond to messages. We invest in impression management, crafting an online identity we can control more than our real one. Power relations matter too: lower-status people watch harder for messages from those above them.
- Social capital drives us to respond to others
- Impression management: we curate identities we want others to see
- Online identity is easier to control than the real-world one
- Lower-status people monitor higher-ups more closely
“we have more control over our online identity than we have over our identities in real life.”
#social-media#status#identity#psychology
✶Explainer46:30
Neuroticism and conscientiousness shape your attention (and self-regulation can be built)
On the Big Five, people high in neuroticism, prone to stress and replaying events, tend to have the shortest attention spans. People high in conscientiousness are disciplined but check email more because they fear missing something. Self-regulation is partly innate luck, but Mark stresses it is also a skill anyone can develop.
- High neuroticism correlates with the shortest attention spans
- High conscientiousness means more frequent email checking
- Conscientiousness predicts life outcomes but carries a checking cost
- Self-regulation can be developed even if you weren't born with it
“Neurotics tend to have the shortest attention spans.”
#personality#big-five#self-regulation#attention