Keyboard Latency

By Xah Lee. Date: .

there's this article:

[keyboard latency By Dan Luu. At https://danluu.com/keyboard-latency/ , accessed on 2017-10-17 ]

the article gave this result:

keyboardlatency (ms)connectiongaming
apple magic (usb)15USB FS
MS natural 400020USB
das 325USB
logitech k12030USB
filco majestouch30USB
dell OEM30USB
powerspec OEM30USB
razer ornata chroma35USB FSYes
olkb planck rev 440USB FS
ergodox40USB FS
MS comfort 500040wireless
easterntimes i50050USB FSYes
kinesis advantage50USB FS
genius luxemate i20055USB
topre type heaven55USB FS
logitech k36060โ€œunifyingโ€

the whole study seems questionable. It tests latency of key press to the time USB receives the signal, not from the start of key activation. That means, thin keys, such as laptop keys, will be much faster than thick keys such Unicomp Keyboard. And also, Cherry MX Red switches will be faster than Brown.

[see Keyboard Key Switch Mechanisms]

[see Evolution of Laptop Keyboard]

also, how did he measure the timing of pressing the key? if, it's a human finger, then, the whole study is questionable. His result shows from 10ms to 50ms keyboard latency. But those timing, is easily within human reaction. Namely, it'll take you that much time just to press a key all the way till it activates, and different people of age and health will vary greatly.

i was quite excited reading his article, finding it very nerdy and well done, but was disappointed by the main subject of the article.

also, in the article, you see that he claims that latency of down to 2ms is detectable, by citing a Microsoft video study of latency on touch screen devices.

This video shows that a latency of 10ms for a touch screen, is still very noticeable.
Applied Sciences Group: High Performance Touch
Mar 6, 2012
Microsoft Research

What he failed to mention, is that, the detection of latency from the Microsoft study is by somewhat accumulated latency of a moving object. In short, if you have a jet plane flying at sound speed, even a 0.1ms latency will be noticeable like fire alarm.

let's look at his conclusion again. He is measuring keyboard latency, counting the human finger key press, not key activation.

Now, if you actually count the key press, then, on a thick key, it's going to be obviously noticeably slower than a thin key keyboard such as laptop, by a margin that is easily more than the result he gave. So, this means, the study is rendered almost useless, as anyone knows, that thin keys have shorter latency.

also, if you actually count the key press time, then, there's the question of human response time. How exactly, do you measure the time a human takes to press a key? Do you shout โ€œgoโ€ like a 100 meter sprint race? also, how old is the guy pressing the key? 18? 50? And, is the test done when the guy just woke up and had coffee? or just had a big lunch ready to nap? oh, and which keyboard is the first one being tested? and which is the last?

Great article, bad conclusion.

2017-10-17 thanks to Milosz Danczak for pointing out Dan Luu's article to me.