Hi Jim.
In my previous post in this thread I took some off the top of my head liberties by SWAG-ing numbers out of the blue and now realize I overstated the delays of an optimal link earlier, since
[1] I probably doubled the time for DSP processing times, and
[2] the latency demands on the codecs used in wireless-mic applications are generally under 5 ms, while some devices are capable of under 2 ms.
As for the physics you asked about, see Table 6 .1: Single Hop Delay Calculation at this Cisco White Paper: bit.ly and subtract the receiv buffering and dejittering processes I mentioned earlier. I stand corrected on the 'send-side' buffer I mentioned earlier. As you can see in Table 6.1 referenced above, it was the packetization delay that I should have cited, instead.
As for my mentioning automobile insurance and mobile telephony applications as a couple of examples of emerging, dynamic, algorithmic-based transaction processing applications (over in the HFT thread), it was merely an attempt on my part to steer the discussion back to its technology-oriented aspects, and in no way was it an attempt at minimizing the risks associated with HFT. You may recall, in fact, that I posted similar concerns earlier on, although not to the same depth that you've apparently examined the subject over the years.
As for your posting a longer treatment on the topic of risks presented by the current HFT paradigm, I suggest that you devote a little time to creating a treatise-like writing on the subject, if it is your intention to be comprehensive about it, and post it in one or only a few longer messages for expository and reference purposes. So yes, I'd be interested in seeing the breadth of the threats as you perceive them. Also, I'd gladly entertain using the old FCTF thread for a longer discussion on the topic, or even engaging you, or following your views on the topic, if you were to post them to a new thread named after the subject, if such a thread on SI doesn't already exist, but that is your call to make.
FAC
------ |
