Comments from Frankston, Reed, and Friends
Saturday, May 25, 2002
DanB at 11:49 AM [url]:
More from Kevin Marks
Reacting to a variety of blogs pointing to him (including my post below), Kevin Marks has lots more to say about connectivity and how networks should be architected. He points to Stuart Cheshire's "Laws of Network Dynamics" which make interesting reading. Stuart gives many examples of when "optimizing" a network for a particular purpose you get a worse network.
Friday, May 24, 2002
DanB at 5:37 PM [url]:
Making money on commodities
Kevin Marks writes about making money in a commodity business. After all, he says, Walmart is in one... (Thanks to David Weinberger for pointing to this!) It's the content business, Kevin argues, that is high risk with a dependence on "the fickle tastes of the public".
Yes! "Content" is a "fashion" business! The fluctuations in a fashion business can be a real mess for a large corporation with high fixed overhead. When you have high fixed overhead (like in the bit-moving business with fiber in the ground, racks of routing equipment, and service people on call in trucks) you want a more steady, low-risk business. Real fashion businesses farm out their manufacturing and other high fixed cost components. Many successful ones only commit to spending money when they have firm orders in hand based on inexpensive prototypes shown at trade shows. The idea of spending billions of dollars for an infrastructure that is dependent on tieing it to people wanting cartoons on their cell phones everyday for 10 years is as silly as making enough of this year's skirts to last 10 years or making all 10 years of a TV program before it's ever shown. Of course, as Kevin Marks points out, it's not easy to do a good job of providing a commodity -- you need good design and execution. But that's true in any business -- you need to do a good job and not just depend on an idea of a business model for success.
Together with David Reed's application of the concept of options to End-to-End and User Financing we are getting an economic basis for separating content from connectivity and why the Internet works and how it must be furthered.
Note that when David Reed says "options" he means it in a general sense, not the narrow version of those traded on Wall Street. An option is a "right" to spend money or effort or whatever at a future time, but not the "obligation" to do it. After paying for an option (such as the option to develop on a particular plot of land), you only have upside, since you can abandon what you thought you might want to do. If you commit to something -- i.e., not an option -- you still have to do it even if something better comes along. David uses the term to include the ability to do things not planned with a technological platform. You can calculate a value for options with formulas like the Black-Scholes options formula. People like options, that's why they buy things with more capabilities than they need right now "just in case".
Thursday, May 23, 2002
DanB at 11:50 PM [url]:
Connectivity 2002 notes
I was able to attend a few of the sessions at Pulver's Connectivity 2002. Bob and David were two of the speakers. (Actually, that's not correct: Bob ended up being called upon to speak several times when speakers didn't show, etc...) Here are some of my notes. (For more detailed coverage, read David Weinberger's reports (start with Tuesday morning and then work forward, especially his "Story for the lay person") and Halley Suitt's reports (starting with Tuesday).)
Frankston's "More's" Law: "You know that Moore's Law can be applied to something when if you want more, you can buy more." (If people want "more" of something like speed or capacity and can use the improvement, then some will pay extra for more leading to a market for improvements which leads to learning curves which lowers prices which increases demands but some people still want more...and you have a virtuous cycle.)
David Reed: It's important to admit when "We don't know" (i.e., uncertainty). That's where value is. "Stupid Network" (Isenberg) means it doesn't know what end user is doing, but is smart about how it implements the "simple" things it does (i.e., moving bits). The End-to-End Argument: "Implement functions at the end points or edges, if at all possible; add function in the network only if it's the only possible way to do it." Why? "An option is the right but not the obligation to invest money, effort... on a future opportunity; the value of an option increases with uncertainty; putting functions at the edges creates value in the form of options." Who finances new paradigms? Look at fax, PCs, cell phones, Internet (LAN, routers), wireless data, telephones. The user does, and they pay for options: Automobile vs. bus, 500 channel cable vs. broadcast big 3 or so, WWW vs. Britannica, PC vs. dedicated word processor. "The Henry Ford Solution: Buildout of 'Broadband' likely to cost $100B-$300B, what application could justify that cost? Making a car and providing it to everyone will cost more than the entire railroad system, what application could justify that cost?" The user can afford it: Home run fiber to the home from CO is $3500-$2500 and dropping. Painting a suburban home (lasts just a few years) costs $2000, automobile is $20,000+. Why would a bank invest in a Model T Ford in its first model year? Invest in the owner, not the car. Confusing the transport with its content could be a barrier to user financing.
John Levine (abuse.net): People can't believe nobody's in charge of the Internet. People find it hard to fathom that the Internet is two-way (talking about privacy, tracking, etc.).
There was a CLEC/ISP that talked about how they give their customers high speed Internet through phone wires at further than 30,000 feet from the CO -- they don't sell "DSL" (which "officially" only goes 18,000 feet or something) but "Internet" and use whatever new equipment they can find that works. (It's getting better all the time.) An example of Frankston's "More's" law.
Bob Frankston again: "About companies: It's not a crime to kill one." (Meaning, it's OK for a company to be forced to change what it does or go out of business -- e.g., buggy whip or slide rule manufacturers.) People (employees) won't starve -- we still need people to splice wires, take support calls, etc., even if a "phone company" no longer can charge a lot for "calls" or "caller ID". (They charge $6/mo to unblock the Caller ID signal that's there anyway.) Somebody noted how phone companies have not added new jobs in years, while new technologies (routers -- e.g., Cisco -- and cellular, Internet, etc.) have gone from zero to many, many thousands.
About content needing to subsidize connectivity: Dan Berninger of Pulver.com had the image of the sock industry needing to subsidize the shoe industry. Bob Frankston on the fixation on entertainment: "You'd think the purpose of a roof is to keep rain off the television."
Sunday, May 19, 2002
DPR at 8:57 PM [url]:
O'Reilly conference and wireless networking
This past week I was at the O'Reilly Emerging Technologies Conference. The set of people who show up there are really interesting. 802.11 was everywhere, and so were community networkers putting 802.11 everywhere they could.
One of the lucky circumstances was having lunch with Dan Gillmor who wrote a nice column about OpenSpectrum. I spoke on a panel with Larry Lessig, and explained why scalable wireless networks change the rules from scarcity to cooperative creation of abundance. It feels like there's a movement beginning, and this could be as big as the Internet all over again, this time enabling mobility and reconfigurability.