Why the iPhone will beat the ChatGPT slab

Because talking is the fastest way to express ideas for a person, but watching video is the fastest way to consume ideas.

Here’s a concise table benchmarking human information intake and output across different modalities, roughly quantifying bandwidth in terms of bits per second (bps) or words/ideas per minute (wpm/ipm) where applicable. These are estimates grounded in cognitive psychology and neuroscience research.

Modality

Type

Bandwidth Estimate

Notes

——

Listening

Input

~60 bps / ~150 wpm

Conversational speech; limited by working memory.

Talking

Output

~60 bps / ~120–150 wpm

Slower for complex concepts; real-time feedback modulates clarity.

Reading (text)

Input

~300 bps / ~250–300 wpm

Faster than listening; denser information per unit time.

Writing (text)

Output

~20–40 bps / ~20–40 wpm

Very low throughput, but high precision and permanence.

Images (viewing)

Input

~10,000 bps (estimated)

High parallelism; extremely rich but hard to quantify precisely in wpm.

Video (watching)

Input

~20,000–40,000 bps

Combines audio and visual streams; good for narrative and emotion.

Gestures/Body Language

Input/Output

~10–50 bps (wildly context-dependent)

Subtle, low-precision, high-contextual load; rapid emotional signaling.

Touch

Input

~1,000 bps (Braille: ~30–50 wpm)

High fidelity for texture, pain, pressure; rarely used for abstract info.

Smell

Input

~1 bps (symbolically negligible)

Powerful memory triggers, poor for abstract communication.

Taste

Input

<1 bps

Like smell, limited symbolic bandwidth.

Facial Expressions

Output

~10–100 bps

Rapid and universal emotional cues; not good for detailed data.

Leave a Reply