The Millisecond Is a Human Invention, Not a Natural One
The second is ancient — it comes from the sexagesimal counting of the Babylonians, who divided the day into 24 hours of 60 minutes of 60 seconds. The millisecond is a recent parasite on that system, born only when we needed to measure things faster than a heartbeat. A mechanical clock could not resolve a millisecond until the 1600s; an electronic one could not until the 20th century. Today, the millisecond is the currency of digital life, and almost nobody knows its name even as they spend thousands of them every second.
When you scroll a social feed, your finger's movement is sampled by the touchscreen at roughly 120 Hz — every 8.3 ms. When you watch a 60 fps video, a new frame arrives every 16.7 ms. When you speak on a call, your voice is encoded into packets roughly every 20 ms. None of these numbers is an accident; each is a compromise between feeling responsive and not melting the battery or saturating the network. The millisecond is where engineering meets perception, and the border is sharper than most people realize.
Reaction Time: The Millisecond as a Biological Limit
The fastest conscious human reactions — a sprinter reacting to the starting pistol, a goalkeeper reading a penalty — bottom out around 120–150 ms, and that floor is set by nerve conduction velocity, not willpower. Signals travel along sensory neurons at 30–120 m/s; the path from eye to brain to spinal cord to muscle is simply a fixed length of wet wire. You cannot train below ~100 ms because your own biology won't permit it. This is why "reaction-time" arcade games plateau: the machine measures in single-digit milliseconds, but the player is stuck at 150.
| Event | Typical Duration | In Seconds | Why It Matters |
|---|---|---|---|
| CPU clock cycle (5 GHz) | 0.0002 ms | 2×10⁻⁷ s | Millions of instructions per millisecond. |
| 60 Hz display frame | 16.7 ms | 0.0167 s | Below this, motion looks smooth. |
| Touchscreen sample interval | 8.3 ms | 0.0083 s | How often your swipe is read. |
| Human audible reaction (sound) | ~150 ms | 0.15 s | Fastest conscious response. |
| Human visual reaction | 200–250 ms | 0.20–0.25 s | Why you can't dodge a fast ball. |
| Geostationary satellite round trip | 500–700 ms | 0.5–0.7 s | Why satellite calls echo and lag. |
| Blink of an eye | 100–400 ms | 0.1–0.4 s | You miss ~10% of a movie to blinking. |
s = ms ÷ 1,000 ms = s × 1,000
Worked Examples
250 ms reaction time → 0.25 s
The textbook human visual reaction. In a car traveling at 100 km/h (27.8 m/s), 0.25 s is 7 meters of travel before your foot even reaches the brake. Add another 0.3 s of braking buildup and you've covered ~14 m before deceleration really bites. This is the physical argument for automatic emergency braking, which reacts in under 0.1 s and routinely avoids collisions a human would not. The millisecond is literally the difference between a fender-bender and a funeral.
16.7 ms video frame → 0.0167 s
At 60 frames per second, each frame persists for 16.7 ms. Drop to 30 fps and each frame lasts 33.3 ms — and the eye starts to see judder in fast motion. This is why gamers pay for 144 Hz and 240 Hz monitors: 6.9 ms and 4.2 ms per frame respectively. Past ~240 Hz (4.2 ms) the benefit flattens for most humans, but competitive players still perceive it. The "smoothness" you feel is just shorter milliseconds.
1 ms gaming monitor → 0.001 s
A "1 ms response time" LCD claim refers to pixel transition, not input latency, and is usually marketing — real pixel transitions are 4–10 ms. But the framing tells you everything: the industry sells the millisecond. A true 1 ms gain in end-to-end input latency (from a 144 Hz panel, a wired mouse, and a close server) is the edge that wins a tournament. Professionals care about milliseconds the way investors care about basis points.
20 ms voice packet → 0.02 s
Most VoIP codecs (Opus, G.722) packetize speech every 20 ms. Too small and you drown in protocol overhead; too large and conversation lags. 20 ms is the sweet spot arrived at by decades of trial. Multiply by the number of hops between caller and callee and you get the end-to-end delay. Cross a congested mobile network and that 20 ms balloon into 200 ms — suddenly you're the person who keeps saying "sorry, you go ahead."
Speed of Light, in Milliseconds
Light moves 299,792,458 m/s, so in one millisecond it covers 299.8 km. This single fact explains why geography is destiny for networks. The theoretical minimum round-trip time between New York and London (5,585 km of fiber, but fiber slows light to ~200,000 km/s) is about 28 ms — and you can't beat it without bending physics. High-frequency trading firms spend billions laying straighter cable and microwave towers purely to shave a few milliseconds off this floor. In 2010, a firm called Spread Networks laid a straight fiber route from Chicago to New York specifically to cut latency by ~3 ms; rivals then built microwave relays to beat even that. The millisecond is worth more than gold per unit mass.
Engineering Context
Milliseconds-to-seconds is the most common time-scale conversion in software and signal processing. Sampling audio at 44.1 kHz means a sample every 22.7 µs (0.0227 ms); converting to seconds is how you compute the period of a tone. Debouncing a mechanical button means ignoring contacts that bounce for ~5–20 ms. A real-time operating system's task scheduler might wake a thread every 1 ms (a "tick"). Once you internalize ms = s ÷ 1,000, you can read any datasheet's timing diagram, size a buffer for a network stream, or estimate how long a database query feels to a user (anything under ~100 ms feels instant; 100–300 ms feels sluggish; over 1 s needs a spinner). Related: seconds to milliseconds (the inverse), and the broader time ladder in seconds to minutes, minutes to hours, and hours to days. For the full story, see the Time Conversion Guide.
More: seconds to ms · seconds to minutes · minutes to seconds · hours to seconds · Guide
Related Unit Converters
Frequently Asked Questions
Is 5G actually faster in milliseconds, or just marketing?
Both, but the latency claim is real. 4G LTE has a radio round-trip latency of ~30–50 ms; 5G New Radio targets ~1–10 ms under ideal conditions. The catch: that's radio latency only. The internet beyond the tower still adds another 10–100+ ms depending on distance and congestion. So a 5G phone on a server across the country still sees ~30–50 ms total, not 1 ms. The 1 ms figure applies mainly to ultra-reliable low-latency (URLLC) use cases like factory robots and remote surgery, on private 5G networks with local servers. For your Instagram scroll, 5G shaves maybe 20 ms off 4G's ~45 — noticeable to gamers, invisible to everyone else.
Why does a "ping" of 20 ms feel fine but 200 ms feels broken?
Because of the 100 ms perceptual threshold for conversation. Below ~100 ms one-way, a two-person voice exchange feels like being in the same room — you can interrupt naturally. Between 100–300 ms, the gap is long enough that people start talking over each other, then pause awkwardly. Above 300–400 ms, you consciously wait before speaking. Video calls are most tolerable when total latency stays under ~150 ms one-way; this is why a New York–Sydney call (fiber distance forces ~120–160 ms one-way minimum) always feels slightly off no matter how good the codec. The millisecond is the unit of politeness.
How many milliseconds is a "frame" in filmmaking?
Cinema runs at 24 fps, so each frame lasts 41.7 ms. Television and YouTube standardized on 30 fps (33.3 ms) and 60 fps (16.7 ms). High-frame-rate cinema (e.g., 48 fps in The Hobbit) halves that to 20.8 ms, which some viewers found "too real" — the soap-opera effect. The choice of frame rate is a milliseconds decision made by committee: 24 fps was the slowest speed that avoided flicker on celluloid projectors in the 1920s, and it stuck for a century because it looks like "film." Every aesthetic you associate with movies traces back to a 41.7 ms shutter.
Can humans perceive a 1 ms difference?
Not directly as a discrete event — you can't "see" one millisecond. But you can perceive accumulated or comparative differences at that scale. Two audio tracks offset by 1 ms cause detectable phasing (comb-filtering) in headphones. A 1 ms input-lag difference between two monitors is felt by esports players as "this one feels snappier" even if they can't name the number. Audio latency above ~5–10 ms between a musician's action and the monitored sound causes measurable performance degradation; under 3 ms feels like direct monitoring. So while 1 ms is sub-perceptual in isolation, it is perceptible in aggregate — the sum of a hundred milliseconds is a felt experience.