Americans need 100mbps broadband access? Another super-sized joke.

Its been nearly three years since I updated this blog. Since I wrote my first . three . posts about human input bandwidth, I never felt I needed to continue but one story this morning got me so mad that I felt compelled to write something.

A report states that all Americans should have access to 100MBps broadband by 2012. [Access the PDF summary here.] and in my opinion it's terribly wrong. Not just in America but anywhere.

Did anyone writing this report stop to think exactly what broadband access is? Its an advertised maximum access rate. It's not a true measurement of throughput or latency or ability to access content. Increasing your access bandwidth rarely increases your ability to pull data from servers and it has no relationship to improving the ability of children to learn or an economy to grow. Its akin to a super-sized burger and encourages marketing to those that enjoy consuming the most. It also puts ISPs under deeper financial pressure resulting in more of those back-scratching 'partnerhips', sneaky marketing and a whole bunch of technical tricks that reduce QOS to the end user.

You can't just dial in new core backbone fibres to prop up access bandwidth. The core routers need re-sizing and most importantly and most expensively, the edges will have to be re-architechted to be able to handle the increasing nu,ber of access products that need to be supported. How do you stop 10x100mbps users from killing the backhaul link for the 1mbps user that wants to watch his grandsons latest YouTube video? It takes engineering, technology and time. Its a cost that gets passed on to everyone. Re-architechting backbones for this orgy of bits is something that no ISP can afford in the current competitive market and so you know what will happen don't you... tricks. Oh sorry, they call it data management. Grooming. I know, I used to architect ISP networks and even four years ago we were forced into thinking about traffic grooming because of the unexpected rise of the P2P consumer. Traffic compression, proxying, deep packet inspection, and protocol redirection is something that I'd rather not see happening. Do you want your images transparently re-compressed? (This already happens in some mobile internet networks BTW. Ask Vodafone and T-Mobile.) Your audio streams could be recompressed. Even your video streams could be transparently re-compressed on the fly thus reducing QOS for everyone and when third parties start offering to do it on third party netoworks your data will be shipped to an unknown location, unpacked, inspected, databased, modifed and repackaged in the cheapest way possible.

What exactly are these people going to need 100Mbps for anyway? My previous estimates show that 35mbps is a top-whack requirement for a home and that includes multiple high-definition video streams. Normal people will actually be very happy with a real 10mbps. How on earth could anyone require a faster speed than that? The report tackles this question. Or rather misses the point completely.

How much bandwidth capacity do we really need? In Part IV, we ask and answer the question “How much bandwidth capacity do we realistically need to enable our businesses, institutions, and residents to survive and thrive in the emerging global economy?” In addressing this question, we took into account (1) the vast array of benefits that high-capacity broadband networks can provide, (2) the potentially explosive growth in demand for broadband capacity between now and 2012, particularly as bandwidth-guzzling video applications gain popularity, (3) the unrelenting drive of the leading Asian and European nations to develop networks with bandwidth capacities of 100 Mbps to 1 Gbps or more; (4) the arguments of EDUCAUSE, the Fiber-to-the-Home Council, and other entities in the United States that have addressed this question; and (5) the many other factors discussed in Part IV.

As you can see, the report didn't actually stop to consider how much bandwidth a person could or should be consuming. They appear to be treating it as some sort of race to catch up with the small Island of Japan and quoting 'benefits', 'growth', unrelenting drive from other nations (did anyone stop to consider why they have high-bandwidth access networks?) and some education requirements. Maybe schoolkids need three concurrent HDTV streams to learn these days? If that's the case then i'm sorry but the last time I went into my daughters classroom they had stacks of dog-eared books, a crappy old TV a video player, a couple of old desktop PCs, a bunch of wobbly-sounding cassettes and a teacher.

You just can't physically take 100mbs of data in through your senses. OK, you can, but you're up there in audio/videophile land if you really need it and you're part of a tiny percentage of users. Even if you think you need it, how do you think content providers are going to deliver this extremely high quality content unless they start introducing new charges for accessing high-bitrate content? Remember that they have to pay for access too. I guess if you're happy to pay for it then go ahead but there's no way you should be subsidised by 'low' bitrate users. Pay per content is the only payment model these fetishists users should have access too.

100mbps might make a nice marketing figure but it will cause chaos on the backbones and peering points if the marketing people start forcing these products through just because some short-sighted people think that more bandwidth will enable their comrades and children thrive. Too much of anything was always bad. Energy, food, TV, Cars. To bring prices down, shortcuts are always made. Bad quality is wrapped up in expensive marketing. No-one wins except those people that made the report for the marketing companies in the first place.

Here's a better plan. Lets build honest, quality backbones that allow ISPs to sell good multi-location access products. Give people the freedom to access the internet wherever they are by providing reliable mobile services, not just a huge pipe that ends up feeding unused content to an over-sized computer in an empty house.

Will it happen? Of course not. Super-sized internet and cheap content is coming your way because it's a corporate game. I feel better having got my thoughts off my chest but I realise, reluctantly, that my opinion doesn't fit very well with large corporates. They need to serve their shareholders in quality ways, not their customers.

Via ArsTechnica.

P.S. See you in another three years!

Time-Shifting and background input

I talked about 13Mbps in the last post. Having reviewed that, i'm happy with it.

The things that need to be added to it are the recording of input for later use, multiple concurrent inputs and monitoring.

Time-shifting, unfortunately, requires a doubling of the input if you are to retain quality. Although it might be possiblt to obtain the input over a longer period for a lower bandwidth (it doesnt need to be recorded real-time) the trend is actually to record things at a faster rate that real time (riping a CD, downloading a movie for example.) So lets double the bandwidth requirement at this point

We're at 26Mbps now.

Multiple concurrent inputs will also require multiplying the bandwith. Although we could be talking about picture-in-picture and background music (which arguably could be at lower bandwidths) we could also be talking about providing occasional second inputs for other people (although in theory, we'd be borrowing bandwidth from the second person for this activity.) I dont feel we need to double the requirement for a second concurrent input so i'm going to set it at 50% of a normal input. 50% of 13mbps. (i'm not including time-shifting in this equation). This gives us an extra 7.5Mbps.

We're at 33.5Mbps now.

And finally, lets add monitoring. Its a low-bandwidth background activity. Even the video aspect would only require 2Mbps. A few remote cameras feeding into your monitoring equipment wouldnt require much more than that.

So that brings us to 34.5Mbps. Lets call it 35Mbps.

This is a top-whack figure. In practice i'd argue that through advances in technology and a reduction in the need to time shift (as most broadcast data is going to be available on-demand from suppliers in the future) we are only going to need 25Mbps per person.

Thats it. Assuming that it would be at a cost where it wouldnt affect the choice, 25Mbps per person is all thats required.

Eyes and ears.

The satisfication of ears wont take much. For the last 20 years, we've been happy with a stereo image of around 20khz bandwidth. Sure, you can add a few more channels to that. DVD-Audio is up there at the top of the audio definition tree offering 6 channels at 96khz with 24bit resolution. That takes 9Mbps compared to CD at 1.4mbps. Lets assume we can compress DVD-Audio resolutions by the same amount as CD resolutions (by about 1:10 ratio) and we have around 1mbps for all the audio you'll ever need. In fact, i'd go as far as to say that with some of the newer psycoacoutic algorithms, you'll be happy with 512kbps audio rate.

Eyes, however are a different matter. Can you imagine the bandwidth needed to cover all the data flowing into your eyes? Gigabits per second I guess. However, when did 180 degree cinema ever get popular and can you imagine the costs of making films at eye-resoution. Impossible that it would ever happen. So, we're left with something thats around the HD resolution mark (1920x1028) I guess. Maybe a bit more for the future so lets add 25%. Using something like mpeg2, we can deliver broadcast quality HD in around, 25mbps. But thats not current technology, so lets chop it in half and say we can deliver HD in mpeg4 format in about 12Mbps. Seems reasonable.

So what do we have so far? 0.5mbps for ears and 12Mbps for eyes. Anything else required? No, I dont think so. Lets round it up - 13Mbps will be fine thank you.


Next thoughts - what about a bit of internet thrown in, some picture-in-picture, recording while watching and a second person - This changes the requirement somewhat. I'll think about this in my next post.

First thoughts

How much bandwidth can one take?

The human body only has a certain number of inputs right? And only a few of those can easily be fed from computer outputs. How much bandwidth would it take to satisfy all possible inputs?

Note I said 'easily be fed from computer outputs' - I mean a screen and speakers basically. I'm not talking about smell generators, movement generators or other 'stimulations' that can be achieved. I also said 'satisfy all possible inputs.' Theres a difference between reahing the limit of inputs (I believe its not really possible to reach the limit of inputs on the human body!) and satisfying the inputs. Granted, different people will need more input but lets just take Joe Average shall we.

First guess is around 20Mbps of data and 2005 compression rates. Lets just make that statement now and see how I feel about it in ten years time when I re-visit this first thought.

Until the next post, i'll be wondering exactly what can be carried in 20Mbps!