Are there any HN'ers (patio?) who live in Japan and could shed some light on what you actually use all this bandwidth for, if you do? Is the current situation really as it's painted?
I'm in the UK and I current have a (fairly flakey) 24Mbps ADSL2+ connection and I can see running everything like TV, radio, music, voice, etc via this connection. I would do this more if it was more reliable - and I already download a lot of video and stream music via last.fm, use skype, etc... I can see myself using maybe 5-10x my current bandwidth quite happily but not really knowing what to do with anything on top of that.
Obviously while it was novel you'd just gleefully 'download the whole internet' because you could, but what then? Apart from bigger, better, faster, HD/HiFi versions of current applications, what do you (or would you) use a really fast connection for, in the long term?
How do people in Japan use their current 1 Gbps connections today? Are the ISPs making money or are they in some kind of government-subsidized fiber bubble?
1 Gbps is faster than a hard disk and 10 Gbps is faster than an SSD; this could change app design.
To be fair, network latency is both much higher and much more variable than hard disk latency, even if the bandwidth is higher. Design can change, but it's hard to get away from needing enough local smarts and storage to make an interface feel snappy.
Edit - To back up my otherwise unsupported assertion:
--- ycombinator.com ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 4105ms
rtt min/avg/max/mdev = 80.985/112.433/163.591/29.114 ms
So a little over 100ms with significant variance for the open internet, results fairly typical. (Edited to add: I can't reproduce these results; later attempts give closer to 450ms, so this is biased in the network's favor.) Hard drives (http://www.newegg.com/Product/ProductList.aspx?Submit=ENE..., for lack of a better source) have latency around 5 ms.
You can solve bandwidth problems by buying more lines, but latency, once introduced by a slow network hop, can never be recouped by any amount of hardware.
And it will. I signed up for the speedier FIOS (20/20) just so that future comes to my house a little quicker than others. This should give me a better shot at inventing the next bandwidth hog like YouTube.
Right now I'm thinking if it's worth it to buy a home server or just put everything into S3/JungleDisk. We're not quite there yet, but it's getting awfully close.
Michael's works are remarkable. Many things he writes about seem obvious in retrospect, yet I find myself thinking differently after reading his wording.
It's also awkward talking to people who haven't read him as I end up coming from different perspective and grasping for common references. Hence why I plug him everywhere I can. :-)
The size of the pipe going out of your house does not matter if the upstream routers do not have #users x max_bandwidth available as well. Seeing as how state of the art equipment does not have enough bandwidth on the backplane to support even a small neighborhood with connection speeds like this it seems that a better description of this endeavor would "waste of effort."
Please define an actual use-case for 10G burst to the home. The reality is that the most important parts of the network are not the last mile, but all of the interconnects upstream and at the peering points.
This sounds nice, but provides no real benefit to the end-user over a 100M connection that was actually backed by infrastructure that could maintain some reasonable fraction of this amount of bandwidth. If the amount of cash that the telco dumped into this 10G marketing ploy were actually used to upgrade the backhaul all of the users would benefit more.
This is a long way from nation wide adoption, but several places are rolling out 1GB connections in the next few years.
Today when you give users 50MB+ connections they start using a tiny fraction of their total bandwidth. The real advantage to 1+GB links is the ability to grow the backbone as needed without caring about the last mile for a long time. Going the other way limits the adoption of services that need such high speed connections. And it's just fiber, they don't need to actually put 10GB equipment at both ends of every home, just run a cable which could in theory support those speeds.
I can't think of any use case for 10 Gbps to the home, so I admit this is a theoretical discussion. I also agree that backhaul is needed, but higher burst speeds don't hurt anything as long as they don't cost more.
I think it has to do more with policy than it does with the telcos. Japan, Korea, and Sweden are outliers which have had massive funding (direct and indirect) from the governments to build up their broadband infrastructures. The hands-off, let the market decide, strategy of the previous Republican administration has led to the largest commercial fiber deployment in the world (FiOS), but it isn't as far-reaching or speedy as the government-assisted rollouts. When you look at penetration rates and bandwidth capacities of different nations, the US is in the middle of the pack.
Setting standards too early can be worse. I remember reading that some countries set an analog HDTV standard in the 90s and ended up changing over again to digital at great public expense.
I'm in the UK and I current have a (fairly flakey) 24Mbps ADSL2+ connection and I can see running everything like TV, radio, music, voice, etc via this connection. I would do this more if it was more reliable - and I already download a lot of video and stream music via last.fm, use skype, etc... I can see myself using maybe 5-10x my current bandwidth quite happily but not really knowing what to do with anything on top of that.
Obviously while it was novel you'd just gleefully 'download the whole internet' because you could, but what then? Apart from bigger, better, faster, HD/HiFi versions of current applications, what do you (or would you) use a really fast connection for, in the long term?