I am not a trader, only an observer of technology. I do play with time series data processing, not networking, which is why I find HFT one of things I try to tap for tips and tricks. After all, a lot of very smart physicists and mathematicians being paid to work 40+ hours a week on it, does produce something worthy of study. A lot of optimizations that sometimes find use in other areas.
That was my comparison to the porn industry with online payment systems, and the military with DARPA projects and the early days of funding for what was the nascent Internet.
The crazy speculation that goes on in the minds of the 'quants' that work on these issues has no bounds, as brainstorming should before the realization stage. They are looking into quantum computing and using custom boards (FPGAs, ASICS and GPGPU) that find use eventually in things like bioinformatics, visualizations, big data processing and huge neural nets.
I program in J, and with Jd, which is similar to k in the APL family, you can process billion rows of data better than some of the systems built by the big money players in tech. k/qdb is mainly used in the financial sector, in which I have never worked.
My point was that it seemed kind of silly to put cable in a box to make a 'speed bump' to sort of negate something everyone already knows or has, and then still have to process their dynamic orders without the delay. I understand the intention, and I am not technical enough to suss it out, but it just seemed to go against common sense. That is all.
I think it's inaccurate to put hft in the future tech division with darpa or porn. Maybe it will lead to fpga's/c++/microwaves being widely used for ML, but it seems unlikely that the low latency and minimal cross country transmission has has much value. If you're building a system to process lots of data quickly transmitting a few bytes between the UK and Belgium slightly faster than a wire won't be a constraint. Maybe it will lead to real time instance based learning, but I'm not aware of anyone going in that direction
Algorithms are also worked on, not just transmission or switching. And speed is always needed by some field or another, so I would argue it has its fruits beyond what you may have posited.
