Eugene Leitl <eugene.leitl@lrz.uni-muenchen.de>:
> Last time I looked spectrum was still a valuable resource,
> protected by legal enforcement. There should be room for low-power
> spread-spectrum grass-routed cellular which looks indistinguishable
> from noise (remember CryptNet?), but such infrastructure needs
> a minimal coverage threshold to be interesting and its hardware be
> a commodity item to be affordable while being violently opposed by
> telco and information provider oligopolies. Also, we need something
> along the lines of ATM for a matrix, while distinctly suboptimal
> TCP/IP still reigns supreme.
Eugene, you're taking all the fun out of this! You seem to be approaching this whole issue from the perspective of a law obiding citizen living within the system. I'm trying to think of it from the perspective of a "lets start over and do it right in a big way" on a locality by locality basis.
Play it like a time travel game. 1998 rules, laws, and infrastructure in one locality, 21st century rules, laws, and infrastructure in a different locality. The 21st century localities will be so technologically advanced, that they can assist in bringing the other localities on line. *Everything* will be affected by broadband. Transportation, education, health care, entertainment, economy. Quite simply, the broadband community will be privately managed by open systems, and the societies will be cashless, and primarily electricity based rather than primarily gasoline based.
Let's say the legal system favors ecology over all other priorities, but rather than simply exterminating the people and bulldozing the infrastructure, it's more of a "go home, be entertained, communicate, enjoy self-paced learning... if you need a ride, we'll handle the transportation."
> That HDTV is very dead I certainly can agree with. As to streaming
> video, the global bandwidth is not yet adequate.
We can upgrade each locality to broadband, with all that entails. It can be priority one, above and beyond anything else. We don't need to make every phone company and cable tv company go bankrupt at the same time. We'll just make them completely useless and obsolete on a town-at-a-time basis.
Who needs 25 different video rental stores and 40 different department stores with CD shelves, when the few thousand titles that are redundantly available can be archived on high-capacity local servers and streamed to local consumers using a local infrastructure of mass produced broadband router/proxies feeding beeper sized distribution nodes.
This is a cultural shift. The pagan communities will upgrade before the orthodox religious communities, I'm sure, due to the relative technological leanings of their respective citizens.
> Unless arrested at
> protocol level (pay by packet), xDSL will push further bandwidth
> clogging, and thus will require a yet another networks upgrade.
> And after streaming video there will come a yet another resource
> hog, I am certain.
The biggest resource hog is distributed scientific computing. The computing requirements are endless. A good start, though, will be something that can be thrown together fairly quickly using off-the-shelf technologies.
You can't please everybody, all the time, (without direct brain stimulation, that is), so why not do this locality by locality.
Approach it like a game. Our community is in the year 2020, your community is in the year 1998, and their community is in the year 1893. We can't practically bring every community to 2020 at the same time (again, without direct brain stimulation/modification), and that's really not worth it.
> DSPs don't have virtual memory, which most Linux flavours require.
Who cares? It's something to do. Can't you simulate virtual memory? DSPs are cheap and fast. Couldn't there be some assembly programming beneath the C functions that compensate for architectural defects. We can use existing DSPs at first in the first wave of devices. The chip-level designs can be modified for the next generation of production. Imagine a flat memory model being used to keep RAM proxies of media files. These things could be thrown together fairly quickly, given the will. Linux has a cult following.
It's not really the *WHOLE* OS that's needed. Really, just enough of a familiar set of tools for file management. (Although, in a no-moving parts device, it will really be array management).
I say hide the complexity of chip architecture from the programmers, and let them focus on creating games and multimedia applications (spectragram libraries, continuous speech recognition, music synthesis). Just preserve enough familiarity so that you can capitalize on the available knowledge workers, including their emotional attachments to old structures. It's more of a social transition strategy at this point that a plan for perfect computation.
> With a high-speed wireless modem can do that with a wearable, today.
> A receive-only unit is not flexible enough by far. Moreover, watching
> movies while crossing street does not exactly appear to be a healthy
> occupation.
Actually, I was being facetious about the one-way cellular. Really, if I see another 20 pound casette playing ghetto blaster in my local "Good Guys" electronics store I think I'm going to vomit. You tear the things apart, and they're nothing but empty shells weighted down. Just being the front display is a little chip that handles all the radio. It's just consumerism garbage. Why not make the same thing out of a cell phone capable of streaming a 64K audio feed? If the consumers need to see big dials and FM style LCD tuners, just spoof the damn things. Let the dial be a placibo, doing nothing but sending an http query to an online streaming media provider. The radio stations in that locality can keep all their studio equipment. But instead of sending their signal out at 10,000 watts, they're really just another site on the web.