Thursday, February 26, 2004
More on Noam: Ain't no Oversupply of Connectivity at My House!
Eli Noam's recent FT column applies a "high fixed costs, low marginal costs, incentives to oversupply" formula equally to infrastructure and application. But wait a minute! Nobody got incented to oversupply my house with connectivity.
High fixed costs -- check! Low marginal costs -- check! So where's my oversupply?
There's no oversupply because the market for applications is driven differently from the market for connections.
The apps market, despite Noam's litany of troubles, is robust, competitive and dynamically stable. Microsoft has successfully cartelized a big piece of it, but its monopoly is far from secure. Other players hold chunky positions in app space (Google, Oracle, Yahoo, Symbian, etc.). There's a wide diversity of smart platforms -- from cell phones to servers -- for all kinds of apps, new and old. As technology advances, its newest benefits flow unimpeded to end-users. Sure there's oversupply. I've got four computers sitting there wasting megaflops, a 160 gig disk that's 1% used, six browsers, nine email clients, dozens of programs I've never touched and a stack of writable CDs in the closet. I love it.
The connectivity market marches to a slower, more lugubrious cadence. What's wrong? The technology isn't lagging. There are optical interfaces on US$49.00 DVD players. Gigabit Ethernet comes standard on today's laptops.
Then why does mass-market-priced connectivity run at a mere 0.3% of a gigabit in the U.S.? In advanced nations, e.g., Korea, why does mass-market connectivity run at 2% of a gigabit? In three or four years, 10-gigabit hardware will be "popularly priced." But will carriers deliver this progress affordably to their customers? No way.
Why not? It's not market demand. The adoption curve for today's lame "high" speed connectivity is one of the steepest in history. Customers want it.
So if it isn't affordable technology, and it isn't customer demand, what is it? Let's return to Eli Noam's, "high fixed costs, low marginal costs" formulation and ask what makes Internet connectivity different.
I think it's the low marginal costs. They're not just low; they're extremely low, and infinitely lowerable for all practical purposes. Carriers can modulate faster, modulate better, add another channel, or add another wavelength. Because of this, if a carrier sells one fiber, or even one "dry" (or sharable) copper pair, they might never sell another. Connectivity providers have to restrict connection speeds if they're to make a profit. But they can't stop there. They must also keep competition at bay, because if there's real competition some competitor surely will offer better technology.
This infinite lowerability works with a second property -- graceful degradation. Like the village green of old, where all the townspeople turned their milk cows to graze, it is always possible to add another user. And like the village green, it is difficult to tell when adding one more user causes the total utility of the connection to decrease. I attended a conference recently where hundreds of packet-grabbing geeks crowded onto a single T-1, which occasionally became so congested that it was useless to everybody. The good news was that unlike a village green (where the grass, once overgrazed, might take years to recover) the T-1 service degraded so gracefully and came back so seamlessly that it was hard to know which state was operative at any given moment.
Graceful degradation theoretically lets me "share" my cable connection with my neighbors. To the cableco, I'm "stealing" customers. A cableco that added more capacity would make it even easier for me to "share". It wouldn't want to do that, because its goal is to maximize profits.
What kinds of network owner would be driven by a different kind of goal: to maximize overall utility to users of its network?
It's no accident that fiber to the home (FTTH) is so slow to catch on. Fiber is the most infinitely and cheaply expandable of all connectivity media. It is not surprising that municipalities operate 32% of the FTTH in the United States. Munis are interested in overall utility, not in profit. Munis want their city to be a better place to live.
In summary, the telco (and the cableco) are victims, not beneficiaries, of the Communications Revolution. Nor are they giving end-user customers what we want. This, then is the market failure!
Fortunately there are alternatives. There are munis. There are (still) CLECs. (Maybe some day some CLEC will get that, "It's operational efficiency, stupid.") There are other utilities. There's condominium network ownership. There are networks owned by single customers. And there are "volunteerist" networks without tragic instabilities.
I'm with you, Professor Noam; national economies are at stake, global volatility is upon us. But let's not paint with too broad a brush, and let's not declare "incentive to oversupply" until I have price competition (or a public spirited provider) for 10 gigabit-per-second connectivity at my house.
High fixed costs -- check! Low marginal costs -- check! So where's my oversupply?
There's no oversupply because the market for applications is driven differently from the market for connections.
The apps market, despite Noam's litany of troubles, is robust, competitive and dynamically stable. Microsoft has successfully cartelized a big piece of it, but its monopoly is far from secure. Other players hold chunky positions in app space (Google, Oracle, Yahoo, Symbian, etc.). There's a wide diversity of smart platforms -- from cell phones to servers -- for all kinds of apps, new and old. As technology advances, its newest benefits flow unimpeded to end-users. Sure there's oversupply. I've got four computers sitting there wasting megaflops, a 160 gig disk that's 1% used, six browsers, nine email clients, dozens of programs I've never touched and a stack of writable CDs in the closet. I love it.
The connectivity market marches to a slower, more lugubrious cadence. What's wrong? The technology isn't lagging. There are optical interfaces on US$49.00 DVD players. Gigabit Ethernet comes standard on today's laptops.
Then why does mass-market-priced connectivity run at a mere 0.3% of a gigabit in the U.S.? In advanced nations, e.g., Korea, why does mass-market connectivity run at 2% of a gigabit? In three or four years, 10-gigabit hardware will be "popularly priced." But will carriers deliver this progress affordably to their customers? No way.
Why not? It's not market demand. The adoption curve for today's lame "high" speed connectivity is one of the steepest in history. Customers want it.
So if it isn't affordable technology, and it isn't customer demand, what is it? Let's return to Eli Noam's, "high fixed costs, low marginal costs" formulation and ask what makes Internet connectivity different.
I think it's the low marginal costs. They're not just low; they're extremely low, and infinitely lowerable for all practical purposes. Carriers can modulate faster, modulate better, add another channel, or add another wavelength. Because of this, if a carrier sells one fiber, or even one "dry" (or sharable) copper pair, they might never sell another. Connectivity providers have to restrict connection speeds if they're to make a profit. But they can't stop there. They must also keep competition at bay, because if there's real competition some competitor surely will offer better technology.
This infinite lowerability works with a second property -- graceful degradation. Like the village green of old, where all the townspeople turned their milk cows to graze, it is always possible to add another user. And like the village green, it is difficult to tell when adding one more user causes the total utility of the connection to decrease. I attended a conference recently where hundreds of packet-grabbing geeks crowded onto a single T-1, which occasionally became so congested that it was useless to everybody. The good news was that unlike a village green (where the grass, once overgrazed, might take years to recover) the T-1 service degraded so gracefully and came back so seamlessly that it was hard to know which state was operative at any given moment.
Graceful degradation theoretically lets me "share" my cable connection with my neighbors. To the cableco, I'm "stealing" customers. A cableco that added more capacity would make it even easier for me to "share". It wouldn't want to do that, because its goal is to maximize profits.
What kinds of network owner would be driven by a different kind of goal: to maximize overall utility to users of its network?
It's no accident that fiber to the home (FTTH) is so slow to catch on. Fiber is the most infinitely and cheaply expandable of all connectivity media. It is not surprising that municipalities operate 32% of the FTTH in the United States. Munis are interested in overall utility, not in profit. Munis want their city to be a better place to live.
In summary, the telco (and the cableco) are victims, not beneficiaries, of the Communications Revolution. Nor are they giving end-user customers what we want. This, then is the market failure!
Fortunately there are alternatives. There are munis. There are (still) CLECs. (Maybe some day some CLEC will get that, "It's operational efficiency, stupid.") There are other utilities. There's condominium network ownership. There are networks owned by single customers. And there are "volunteerist" networks without tragic instabilities.
I'm with you, Professor Noam; national economies are at stake, global volatility is upon us. But let's not paint with too broad a brush, and let's not declare "incentive to oversupply" until I have price competition (or a public spirited provider) for 10 gigabit-per-second connectivity at my house.
Comments:
Post a Comment
