Date: Wed, 31 Oct 2001 10:59:04 -0500 To: "Coppernoll, Julie" <julie coppernoll of intel.com>
From: "David P. Reed" <>
Julie - as you may know I was both one of the original people who designed the Internet Protocols and also was VP R&D and Chief Scientist of Lotus Development from 1985 to 1992.
Since you asked, there is an alternative that can bring a richer and more competitive "broadband" deployment.
The key ideas to be combined are:
- packet transparency
- medium independence
- decoupling content from connectivity
You focus on the $100-300B cost. The industry acts as if this has to be planned out and financed in one "grand buildout". But we all know that highly innovative technologies are built incrementally, by creating incentives for evolutionary innovation in the marketplace. Why ILECs and other utilities won't do this job is that they cannot finance such a large risk with their traditional tools (10-30 year bonds) - because the risk comes both from technological obsolescence (Moore's Law will continue to let competitor beat your "sunk costs"), and from market risk (is the system you build going to match the applications beyond the 2 year horizon). The solution the ILECs propose is that they should be allowed to control the risk by being granted monopoly power to block new entrants and new applications - thereby guaranteeing themselves a revenue stream. And the framework to do this exists, not only in the FCC, but in the PUCs and state/local laws, and right-of-way laws.
The point I'd like to make is that today the technologies exist (especially software and protocol standards in the Internet world) and are cheap enough to be user-financed. For *much less than the cost of an automobile* new fiber can be deployed from a "central office" to one's house. This leads to the idea of what I call the "Henry Ford" solution:
The genius of the Model T Ford was that it was within reach of much of the population, and they could either buy it outright or "finance" it based on their own personal credit. Now personal credit has rarely been used to finance telecom - so little that the grand pundits of telecom cannot even think about it. But the great thing about personal credit is that the *investor is the customer*. That means that one does not have to justify to the investor how the cost is going to be recovered.
If you think about it, the "Henry Ford" solution is what made the PC revolution take off. Even though the primary customers of PCs turned out to be business customers, the fact that a departmental decision could justify a PC led to a market where the end user became the customer, not the "central IT department". This meant that a vibrant, competitive market got created where innovations in functionality could be directly evaluated by users. Only in such a market would "user interface" become a selling point - so much so that GUIs (such as the original Mac) were bought, despite strong objections from the "central IT folks" who viewed such things as unproductive folderol. But users knew that GUIs, WYSIWYG, etc. made their jobs and lives better. So demand was directly connected to supply, in a competitive market.
The key thing here is that both fiber and broadband wireless can be installed without committing for the long term to a particular modulation technique or coding standard over the whole system. Each individual fiber can be upgraded, merely by changing the termination equipment. Since Moore's Law is constantly improving termination and switching, that means that we can separate the fiber and wireless "ether" (which have long lifetimes, and whose value increases with the technology attached to them) from the terminations.
User-financing lets users (or groups of users such as neighborhoods, towns, etc. in the case of condominium fiber - Bill St. Arnaud's term) upgrade to new capabilities as they need them and can afford them. In contrast, in the old Bell model, the phone company owned your telephone and leased it back to you. You could not "upgrade" without their permission, and that permission was tied to them selling services.
Packet transparency is essential to the future. What I mean by this is that packets that go into the network come out at whatever destination they select unchanged. If the network performs translation, filtering, or otherwise "adds value" in a way that is not optional, then it becomes difficult to innovate on either the client end, or the service end. Fortunately, the current Internet is "packet transparent". But there are those who want to change that, and to the extent that they have control of the access network, they will be incented to do so, in order to maximize profits. This is bad for innovation, and bad for companies like Intel that benefit by innovation, because it creates a perverse incentive to use non-transparency to block innovation in hopes of extracting larger rents from innovators who have no other game to play. By reducing the rewards for innovators, innovation is discouraged and will stop.
Medium independence is a term I use to point out that different physical media can deliver the same application service. Much political debate goes on about whether DSL or cable or fixed wireless is best, and which one should be built out. In a true competitive market, this would be settled by letting them compete and find customers that match the different characteristics of those media. DSL, for example, seems to be a short-term step that might work in some places where the copper is good enough. Broadband wireless packet is in its infancy as a technology, and probably can and will get better over time (the obvious failure of point-to-point fixed wireless so far just points out that the business problem of bootstrapping such a business is quite difficult, whereas the amazing success and growth of 802.11 technologies shows how "user financing" might be a much better plan). And putting dark fiber in the ground (when done by newstarts who manage to avoid the regulatory traps that force them into the cost structure of ILECs) has been shown to work in places like Stockholm (Stockab) where a rich competitive market full of innovators has resulted - without having to have a "plan" to grant monopolies so that costs can be recovered.
Finally, at all costs, we need to break the link that ties content to connectivity. The unholy alliance between copyright holders and bit-transporters serves mainly to destroy competition. Intel succeeded because there was a rich set of applications and content for PCs - not because it decided to "pick winners" among applications. I know. At Lotus we would have loved to have Intel lock in our profits by choosing which spreadsheet would be allowed on Intel processors, and creating an infrastructure that would allow us to block competitors from creating competitive spreadsheets. I suspect we would have paid Intel a lot of money to "own that franchise". But Intel was smart - they did not sell franchises to Lotus, to Microsoft, or to any other software vendor. They knew in the long run that Intel (lacking the kind of gov't regulations that they could use to block competition) would benefit by creating a platform for competition, and making sure that that competition was focused on innovating for user needs, not on trying to buy franchises from Intel.
I have tried to be brief. There's a lot more to talk about. I'd be happy to expand on any of the points in this note, should Andy or anyone else at Intel be interested.