SFM fragments your mind—but what constitutes “short”? Certainly, watching TV (on a TV) with its stable, regimented ad breaks can be a non-fragmenting length of content. Broadcast TV ad-break spacing is ~6–8 minutes. There’s also data that MOOCs with ~6-minute segments are about the max the average viewer tolerates. YouTube mid-rolls appear every ~7–10 minutes for longer videos (as short as ~6). There’s signal here about what our minds tolerate as “continuous” content--long enough that an interruption (ad) isn’t jarring and we can keep our mind chugging with chunks this size. Less time than this is where we naturally intuit that our mental machinery is getting jerked around.
How rapid are context changes on Reels, TikTok, X, or FB? An order of magnitude faster. On TikTok, ~46% of raw impressions last <8s; across all impressions, average watch time is ~23s. On Reels, ~60% leave <3s; among those who stay >3s, average watch is ~11s. Average tweet “gaze time” is ~3s; FB feed dwell time ~1.7s per item. These are wild numbers when you consider context switching and re-orienting to an orthogonal information tidbit every 2–3 seconds. It appears our minds need a floor/minimum of ~60s consecutively in a given narrative context to stabilize the event and feed memory. This is the “floor” for coherence. Short of this, visual/auditory cortices and mid-level pSTS associations stitch adjacent sensory events (<4s; mid-level 10–30s) but there’s no situational sense (DMN activity needs ~60–90s of continuity). If there’s local-but-not-global coherence <60s, no memory is formed. Nothing is understood.
Coherence builds hierarchically with at least 1 minute needed for a stable situation model. This doesn’t mean we only need 60 seconds; we keep accumulating >1 minute. Under that, we’re constantly short-circuiting coherence circuitry.
What does this mean for SFM and mental fragmentation? Tweets don’t even activate local coherence given time spent--it’s pure sensory. Unless you pause and process (which ~2.9s doesn’t allow), you’re too disjointed to get anything meaningful. Reels/FB/IG/TikTok: it’s all trapped in local coherence. Few are watching >60-second videos or thinking long enough to establish strong memory. For ~60s of reading, that’s ~200 words or ~1,200 characters (≈5 coherent near-max tweets). The medium itself is coded for fragmentation. Tweets are 280 characters. TikTok videos are encouraged to be <45s.
People retain more from five 6-minute clips than from a single 30-minute lecture. People stop watching MOOCs after ~6 minutes. We’ve naturally found 6–8 minutes of a TV mini-arc tolerable; it doesn’t seem to disrupt coherence or attention. We experience it as a whole because a globally coherent narrative is built around the events.
Too short is DEFINITELY <1 min but probably <5 minutes if repeated. We can stitch a few disconnected ~1-minute events, but a continuous stream leaves weak memories and fragmented thought.
But too long--MOOCs >6 minutes—and working memory saturates and needs a reset.