re: ". I just read a quote from NT's Roth boasting that processing was more expensive than bandwidth. This will be true."
Oh I don't know. Sometimes. It depends on the specifics. If you have to constantly blast an uncompressed stream with 10-20x the bandwidth of a compressed version, versus putting a $25 chip down to decompress the compressed version, that chip pays for itself in no time flat by way of reduced bandwidth charges. If you do that decompression "soft," in the host processor that you have anyway for all your other computing purposes, the payback is even sooner. It would be looney to believe that we will go from an environment of $20 per month for 40kbps or so, to an environment of infinite (for all practical purposes) bandwidth to everyone, everywhere, at any time, for the same price (or some fixed, low price anyway). There's at least 10 generations of incremental changes that will happen before that's even on the horizon. Too many spigot controllers have a vested interest in keeping the model the way it is.
And aside from that, regardless of the model, you'll need hugely increased processing power in the endpoints to be able to process the data from/to these infinite, free links these guys think are just around the corner. Having the Colorado river run through my living room doesn't do me much good unless I also have the Hoover dam there too (harnessing big streams takes big hardware!). Bandwidth and processing power will be in a long tug of war, back and forth, price-wise. Both sides better hope that's the case because they can't live without eachother. |
