15 ms·

Also, I tried a wireless keyboard once and there was a perceptible delay between me pressing the key and the keypress showing up on the screen, which is really

by tines 11y ago

Also, I tried a wireless keyboard once and there was a perceptible delay between me pressing the key and the keypress showing up on the screen, which is really unacceptable.

epistasis 11y ago
That's certainly not a bluetooth problem, but some other hardware/software along the chain. I've been using a mix of bluetooth and wired peripherals for years and never been able to tell the difference.
- VLM 11y ago
  Probably depends on sport vs data entry. Its very hard to find data but is possible. One huge problem is most of the folk explanations or broscience level discussions insist on irrelevancies like long term bandwidth, refuse to debate transactional latency, and insist on not discussing varying deviation of latency effects. Stereotypically the discussion devolves into a debate over the exact WPM its possible to achieve over a five minute averaged typing test being 10 to 100 times faster than the fastest human typist, even on gaming forums specializing in hand/eye coordination intensive games, its like a meme strange attractor. Here's an ancient study of overall systemic latency from hitting the switch to processing code for PS/2 vs USB. http://link.springer.com/article/10.3758/BF03195452 http://link.springer.com/article/10.3758/BF03195452 In summary the PS/2 latency was only single digit ms and critically the deviation of the data was very low, but USB was around 20 to 30 and varied more. I know bluetooth is far worse and far more variable although I don't have a convenient handy link. It seems fairly intuitive that direct hardware I/O will be better than embedding it in a packetized lan, which will be better than running a packetized lan over a radio link. The problem will never appear in bulk data entry or causal text editing, but its an immense problem for e-sports or pretty much anything relying on hand-eye coordination (maybe CAD?). As a simplistic example of the scale of the problem, 10 ms doesn't sound like much, but a fast pitch baseball goes about half a meter in 10 ms... Training yourself via hand eye coordination to hit the space bar to swing a bat using one protocol and then switching to a different protocol would almost certainly result in 100% failure rate, and even worse, competitors who cross protocols would almost certainly have an unfair advantage on the lower latency, more predictable PS/2. Using the modern definition where the only true form of "gaming" is FPS sequels, I would imagine it would be nearly impossible to train on moving or leaping targets using one keyboard protocol and then switch to another protocol. Also I'd anticipate that some game strategies that rely on reliably consistent latency would be possible only on PS/2 and not on bluetooth due to inherent latency and occasional interference issues. I'd go so far as to theorize that possibly one reason console gamers are crushed when competing against PC gamers on otherwise level playing fields might have something to do with high and variable latency wireless console controllers vs very low latency predictable PC controllers. The effect is probably not huge, but over a very large number of transactions it can eventually add up to be a real problem. Its similar to the question of if you would accept a cryptographic random source that was 1% biased? Its possible to come up with numerous scenarios where it won't matter "on very long term average" and so on, but ...

by tines 11y ago

Also, I tried a wireless keyboard once and there was a perceptible delay between me pressing the key and the keypress showing up on the screen, which is really unacceptable.

epistasis 11y ago

That's certainly not a bluetooth problem, but some other hardware/software along the chain. I've been using a mix of bluetooth and wired peripherals for years and never been able to tell the difference.

VLM 11y ago
Probably depends on sport vs data entry. Its very hard to find data but is possible. One huge problem is most of the folk explanations or broscience level discussions insist on irrelevancies like long term bandwidth, refuse to debate transactional latency, and insist on not discussing varying deviation of latency effects. Stereotypically the discussion devolves into a debate over the exact WPM its possible to achieve over a five minute averaged typing test being 10 to 100 times faster than the fastest human typist, even on gaming forums specializing in hand/eye coordination intensive games, its like a meme strange attractor. Here's an ancient study of overall systemic latency from hitting the switch to processing code for PS/2 vs USB. http://link.springer.com/article/10.3758/BF03195452 http://link.springer.com/article/10.3758/BF03195452 In summary the PS/2 latency was only single digit ms and critically the deviation of the data was very low, but USB was around 20 to 30 and varied more. I know bluetooth is far worse and far more variable although I don't have a convenient handy link. It seems fairly intuitive that direct hardware I/O will be better than embedding it in a packetized lan, which will be better than running a packetized lan over a radio link. The problem will never appear in bulk data entry or causal text editing, but its an immense problem for e-sports or pretty much anything relying on hand-eye coordination (maybe CAD?). As a simplistic example of the scale of the problem, 10 ms doesn't sound like much, but a fast pitch baseball goes about half a meter in 10 ms... Training yourself via hand eye coordination to hit the space bar to swing a bat using one protocol and then switching to a different protocol would almost certainly result in 100% failure rate, and even worse, competitors who cross protocols would almost certainly have an unfair advantage on the lower latency, more predictable PS/2. Using the modern definition where the only true form of "gaming" is FPS sequels, I would imagine it would be nearly impossible to train on moving or leaping targets using one keyboard protocol and then switch to another protocol. Also I'd anticipate that some game strategies that rely on reliably consistent latency would be possible only on PS/2 and not on bluetooth due to inherent latency and occasional interference issues. I'd go so far as to theorize that possibly one reason console gamers are crushed when competing against PC gamers on otherwise level playing fields might have something to do with high and variable latency wireless console controllers vs very low latency predictable PC controllers. The effect is probably not huge, but over a very large number of transactions it can eventually add up to be a real problem. Its similar to the question of if you would accept a cryptographic random source that was 1% biased? Its possible to come up with numerous scenarios where it won't matter "on very long term average" and so on, but ...