3 ms·

All you need is one bit: silence -> neutral!, 0 -> buy!, 1 -> sell! Or two: 00 -> neutral!, 01 -> sell!, 02 -> buy!, 11 -> wait! Plus error correction... Whe

All you need is one bit: silence -> neutral!, 0 -> buy!, 1 -> sell! Or two: 00 -> neutral!, 01 -> sell!, 02 -> buy!, 11 -> wait! Plus error correction... When you add error correction and consider the number of missed bits and the need to retransmit / keep transmitting, it's probably not possible to realize a latency win here.

close04 6y ago
Think about it like this: the absolute vast majority of neutrinos pass through the whole planet without interacting with a single particle. One descriptive saying is that a neutrino can pass through 100 light-years of steel without interacting. This puts in context how hard it is to detect them (have them interact with your detector). So your communication turns into some random string of detections where you never know if the absence of a detection means there was no neutrino, or it was just missed.
mhh__ 6y ago
> latency win Perhaps, but if you consider the probability of missing an individual neutrino your error correcting scheme will have to be either very long or very clever because your odds of collecting the right neutrino at the right time will go down very quickly as the packet gets longer. On top of that, if you used some kind of pulse scheme (i.e. morse code with neutrinos) it has to be slow enough to be detectable, but fast enough to beat the latency of a cable (let's say 100ms - speed of light + processing and errors) and also fast enough not to use enough power as to be unprofitable.
- cryptonector 6y ago
  > Perhaps, but ... Isn't that what I wrote: > > it's probably not possible to realize a latency win here ? > On top of that, if you used some kind of pulse scheme (i.e. morse code with neutrinos) it has to be slow "slow pulse" == long pulse. It will be "fast" in that it will go faster than the speed of light in fiberoptics and the path will be shorter, but it will be slower because error correction will demand a great deal of redundancy which, among other things, means long pulses. I think we're in agreement.
  - mhh__ 6y ago
    I have the attention span of a hyperactive toddler so when I write HN comments I can end up responding to everything I've read in the comment tree, sorry.
    - cryptonector 6y ago
      Lol. Me too sometimes :)
whatshisface 6y ago
You don't need error correction, you can win on the stock market with something that's very poorly correlated.

close04 6y ago

Think about it like this: the absolute vast majority of neutrinos pass through the whole planet without interacting with a single particle. One descriptive saying is that a neutrino can pass through 100 light-years of steel without interacting. This puts in context how hard it is to detect them (have them interact with your detector). So your communication turns into some random string of detections where you never know if the absence of a detection means there was no neutrino, or it was just missed.

mhh__ 6y ago

> latency win Perhaps, but if you consider the probability of missing an individual neutrino your error correcting scheme will have to be either very long or very clever because your odds of collecting the right neutrino at the right time will go down very quickly as the packet gets longer. On top of that, if you used some kind of pulse scheme (i.e. morse code with neutrinos) it has to be slow enough to be detectable, but fast enough to beat the latency of a cable (let's say 100ms - speed of light + processing and errors) and also fast enough not to use enough power as to be unprofitable.

cryptonector 6y ago
> Perhaps, but ... Isn't that what I wrote: > > it's probably not possible to realize a latency win here ? > On top of that, if you used some kind of pulse scheme (i.e. morse code with neutrinos) it has to be slow "slow pulse" == long pulse. It will be "fast" in that it will go faster than the speed of light in fiberoptics and the path will be shorter, but it will be slower because error correction will demand a great deal of redundancy which, among other things, means long pulses. I think we're in agreement.
- mhh__ 6y ago
  I have the attention span of a hyperactive toddler so when I write HN comments I can end up responding to everything I've read in the comment tree, sorry.
  - cryptonector 6y ago
    Lol. Me too sometimes :)

whatshisface 6y ago

You don't need error correction, you can win on the stock market with something that's very poorly correlated.