The funny thing about scale -- Kafka's numbers are wrong

In a telephony online article in early May, BellSouth's Chief Architect Henry Kafka was quoted as saying:

The average IPTV user will likely consume about 224 gigabytes per month, he added, at a monthly cost to carriers of $112, a giant leap from the less than $5 attributed to Internet use. If that content were high-definition video, the average user would be consuming more than 1 terabyte per month at a cost to carriers of $560 per month.

“Clearly that’s not what the average user is going to pay per month for their video service,” Kafka said. “That’s why we need help.”

Using the numbers cited in the article, I calculated what the dollar/gig was at each of the bandwidth levels and prices that he indicated, and saw what I expected to see, a cost projection based on a linear cost/meg.

Gigs	dollars	dollars/gig
2	1	$0.50
9	4.5	$0.50
224	112	$0.50
1120	560	$0.50

Kafka's numbers are wrong because the cost of bandwidth is not linear as volume increases. Scale creates economies that result in a lower cost per Mps (or Gigabyte downloaded). I have personally noticed in my own studies that the cost doubles in order to quadruple bandwidth, although this is not confirmed by my colleagues (or any other sources for that matter). If I build my own table using some simple math and a cost of $0.50 per Gigabyte at 2 Gigabytes and a doubling of cost of for every 4x in traffic growth (rather than a quadrupling of cost for a quadrupling in bandwidth), I get the following:

Gigs	dollars	dollars/gig
2	$1.00	$0.50
8	$2.00	$0.25
32	$4.00	$0.13
128	$8.00	$0.06
512	$16.00	$0.031
2048	$32.00	$0.016

As you can see, at a figure roughly double Kafka's doomsday HDTV family usage figure of 1+ Terabytes per month, the total cost can be projected at $32, not $560.

Don't get me wrong, the prospect of building a network to support HD IPTV is as daunting to me as the next guy. The issues of getting enough bandwidth into the home to support the 20Mbps+ required to stream more than 3 HDTV channels simultaneously makes it out of reach to most of the US Population given the distance limitations on current DSL technology. This is, however, the hardest part of the problem. An intelligent metro network design for distribution of IPTV and the use of Multicast can lessen the load quite a bit, leaving VoD to drive the rest of the bandwidth consumption.

I was glad that I caught this article, however. It confirms the gut feeling I've had in the pit of my stomach on this whole net neutrality debate, which is that the bells don't deserve our sympathy. Based on this rather simple analysis, their business case seems to work out just fine without having to charge the likes of Google for access to their customers.

Hat tip to Nels Thompson for the telephony online link, and another one to Mike at Techdirt.

Technorati Tags: Net Neutrality

Trackback URL for this post:

http://blogs.globalcrossing.com/trackback/158

dsiegel – Mon, 2006 – 06 – 19 19:15

broadband – Internet – IPTV – net neutrality – US

dsiegel's blog

As i am based in the UK it is

As i am based in the UK it is at times dificult to judges these net neutrality and IPTV related debates becuase everything is so different over here. From the UKs point of view the "bandwidth costs too much" argumaent against IPTV makes no sense. We have IPTV providersr already in the UK and there are more to come. The bandwidth costs are not that big a deal because IPTV is provided over the IPTV providers network and not over the wider internet. I.E. The DSL (or whatever) connection, does in fact provide an internet connection but all the IPTV servers are located on the providers network so no external bandwidth is required. If you have ever looked at the cost difference between a point to point leased line and a leased line internet connection, you can see that the cost is greatly reduced. When you introduce things like multicast the costs of providing TV over IP are similar to providing cable TV. Then again we have wholesale DSL and local loop unbundaling here in the UK which has introduced real competion into the market and allows providers to run their own networks in a cost efective manner, becuase they don't all have to lay cable. Perhaps that's what's needed in the US.

Ben Robinson (not verified) – Wed, 2006 – 07 – 05 04:30

Tariff Rebate Passthrough

I think I found a compromise that will satisfy the stated goals of both net neutrality sides. Even better, it should actually work, and work in the consumers' interests. The idea is Tariff Rebate Passthrough -- i.e., the ISP can charge by byte for QOS (but only by byte) and the information service provider (Google) can rebate the costs directly to the consumer (but only to the consumer). This works because it meets the need to pay for differentiated QOS, without letting the telecom companies' control over that payment become actual control over content. I.e., all the good parts of net neutrality are preserved, but there's no need to give something costly away for free.

The idea is spelled out at http://www.monashreport.com/2006/06/19/net-neutrality-tariff-rebate-passthrough/

Curt Monash (not verified) – Tue, 2006 – 06 – 20 06:32

Re: Tariff Rebate Passthrough

Thanks for the comment Curt.

I left a response to your idea on your site. Our readers can follow the link in Curt's post to read his idea as well as my response.

dsiegel – Thu, 2006 – 06 – 22 11:57

IP Convergence: Beyond VoIP, Beyond Cost Savings

Regional Blogs

Podcasts

Subscribe

Topics

Industry Bloggers

Recent blog posts

User login

Navigation

Terms

Technorati