« NOT Metcalfe’s Paradox – A Puzzle | Main | Answer To “Not Metcalfe’s Paradox” Puzzle »

July 14, 2008

The Smart Grid Should Be Stupid

It is becoming conventional wisdom (uh-oh) that a large part of the strategy for reducing dependence on imported oil is to use electricity from sources like wind turbines, solar arrays and concentrators, and nuclear plants as a substitute for that oil. There is general recognition (uh-oh again) that getting that electricity from where it is produced to where it is used will require rebuilding our sadly obsolete electrical grid. Moreover, that new grid should be "smart" so that demand and supply can better be matched across the network. The important question is want do we mean by smart?

David Isenberg famously explained that the success of the Internet as a platform for innovation is due to the fact that it is a "stupid network", blissfully unaware of and unoptimized for the applications which send packets over it. David's thesis was heresy at a time when telcos were spending billions of dollars building "smart networks" – each had its own brandname and advertising campaign. The smart networks were and remained innovation-free for the simple reasons that you couldn't build an app for them which didn't fit the preconceptions of the engineers who built the smarts into the network and you couldn't deploy an app onto those networks without the permission of the carriers that owned them.

On the Internet, the devices ("endpoints" to us geeks) are smart but the network knows about nothing but packet routing; it is (or should be) blissfully unaware of what these packets contain, what applications the endpoints are running, and all of the many complex interactions flowing through its routers. The designers of the Internet didn't even plan on its being used for email let alone web browsing, voice over IP, Googling, or video streaming. It could be made a better network for any one of these applications by optimizing it for that application but only the expense of other applications, particularly those yet to be invented.

So what's the lesson for the new electrical grid we want to build?

Make it stupid! Make it open! Enable innovation! And, again learning from the Internet, make it robust by making it distributed without a central command point or central points whose failure would bring down the whole network.

Of course we want smart metering. But smart metering's an application on this network. The information flow for smart metering should be over the existing Internet – which today shares poles and conduits with the power grid and may be extended in some places over the current-carrying wires.

Of course we want to be able to deploy new sources on the power grid and the grid has to be built to where the sources are.  These new sources are like server farms on the Internet; their output is accessible anywhere on the grid. The grid should be "smart", in a distributed sense, about having power flow from regions with a surplus to regions with excess demand. The power network, or the Internet as a parallel information network, should provide information used for real time pricing but the actual pricing should happen at the edges of the network and accommodate many different pricing and settlement schemes.

There are differences between the Internet we know and love and the power grid. One is that transport is a more significant component of the cost of power than it is of the cost of information (put another way, only  a tiny fraction of a kilowatt hour is need to move a packet from anywhere on the Internet to anywhere else). So transport and distance MAY be a more explicit part of energy pricing than they are of information pricing. On the other hand, electricity is fungible and information is not. I want the response to my query from Google, not someone else's fun video; but, if I buy a kilowatt hour, I don't really care(economically and physically) if it comes from the wood-fired plant in Burlington, Vermont Yankee, the dams of Canada, or a windmill in Nebraska, I couldn't tell the difference even if I did care.

The new grid will have to accommodate today's model in which most of us buy power from a local company which buys regionally when it has to and usually belongs to a pool which can buy nationally. But the new grid should NOT be built assuming or, worse, assuring that this is the only model that will work. The point is that we don't know what our mix of electric power sources will be in the future: how much local to save on transportation and for reliability? How much where the wind blows or the sun shines? We don't know much about how we'll use electricity: heating? transportation? lighting of what? recharging of what? we don't know whether or where we'll store electrical energy.

Despite all we don't know, we still need to rebuild our electrical grid. Because of what we don't know, we want to build a stupid grid which, like the Internet, will allow almost limitless innovation at the edge.

| Comments (View)

Recent Posts

Grapes of Wrath

Who Outed Jeff Bezos?

The Noes Have It

FireTVStick Thrashes at&t’s DIRECTV

An Invaluable Lesson in Colonial Williamsburg


TrackBack URL for this entry:

Listed below are links to weblogs that reference The Smart Grid Should Be Stupid:


blog comments powered by Disqus
Blog powered by TypePad
Member since 01/2005