WHAT PRICE IGNORANCE?

Contemplate, dear reader, the wry description of a mathematician as an organism that converts coffee into theorems.

Does that half-liter of water produce more theorems in the mathematician's coffee pot or somewhere off in the cloud?

For today, dear reader, we will not be too fussy about what data a "data center" processes.  A center might provide the servers that store your photographs, music videos, social media posts, and cribbed term papers.  As the Unofficial Newspaper of Record notes, "New Data Center Built To Handle Influx Of Online Rants Against Data Centers."  These days, mass storage using solid state components apparently does file allocation among servers so seamlessly that disk defragmentation is no longer necessary.

National Review's Charles C. W. Cooke's "Your Data Center Life" suggests the civilization-advancing capabilities of a variety of information technology complexes lumped together as data centers.

If, during the next Super Bowl, even 10 percent of the nation’s data centers were to be summarily switched off, the people of the United States would be on the verge of revolution.

The term we have chosen to describe the technology I have discussed here is “the cloud.” But, of course, there is no such thing. There are just data centers and the computers that run inside them. Conceptually, “the cloud” is a useful abstraction. But if taken too literally, it blinds us to an important reality: Each one of the 5,427 data centers in the United States is a real place, full of rows of real machines, that run on real electricity, and that are maintained by real people. Your data — by which I mean your emails, your bank account, your business files, your photographs of your kids — are sitting somewhere on a real hard disk. If I drove to where that hard disk is and smashed it up with a pickaxe, it would break, and the only reason that you probably would not permanently lose the data that it once held is that the data also exist in some other real location, on another real disk, in another real person. Bluntly put: The integrity of your data — and of your ability to access it whenever you want — is the result of your having cheap access to lots of different data centers, in lots of different places, so that if something bad happens to one of them, your stuff is backed up elsewhere.

Thus, a data center might be a switching center to keep the wireless internet and cell phone conversations working, something it does so well that the eternally ditzy former vice president could wave her hands about that data "cloud" being all around us, sort of like Yoda's Force but lacking the capability to lift your spacecraft, er, your campaign, out of the swamps of Dagoba.  Or it might include actual computers, although Tyler Cowen recently noted that computing capacity is still a constraint on artificial intelligence doing anything intelligent.  That doesn't surprise me: businesses and universities might have phased out their legacy mainframes, but true supercomputers still exist and they're still useful.  That Chat GPT prompt the above message complains about is going through that water and power to search through the information at its disposal, way more rapidly than the most assiduous of file clerks with the best associative memory.  Those data centers might be coming up with another "most strikeouts against switch hitters by any major league pitcher since 1908," or more quickly comparing millions of x-rays or gene sequences, and perhaps the computing capacity is finding a proof of Fermat's Theorem that does fit the margin of a Renaissance textbook.

Now contemplate replacing that file clerk with stepper switches and relay stacks.

That's a simple relay stack connecting two old-school dial telephones in a museum of telephony on the Iowa State Fair grounds.  A docent there told me that younger kids interpret "dial" as putting their finger on the number, as they've never had to physically dial in a number.  Now, you could build a computer network out of such relays and stepper switches, although it would be seriously impractical.  For that matter, the first mainframe computers replaced the relay and stepper switch accumulators with vacuum tubes to handle the logic, and the electrical, cooling, and maintenance requirements (one engineer replacing bum tubes, full time) extracted a much larger price on that "most critical resource."

The mind boggles at the resource allocation problems that would arise in the absence of solid state components and miniature circuits.

If the internet is an electronic delivery system for images of cats, the latest generated meme, and calls to civil disobedience, perhaps there's no reason to give streaming video services preferential handling.

On the other hand, in a world of high-frequency traders paying extra for co-location with exchange servers, there exists a market for fast lanes.  Will we be returning to the days of value of service pricing and uneconomic bypass?

That puts Nicholas Rabb's complaint about that water use, "By starting at the local and focusing on politics of material impact, the anti-data center movement has generated solidarity and success on a national scale unseen in recent movements for social justice," in a different light.  But those trade-tested improvements on relay stacks and stepper switches have the usual suspects going to the usual corners.  "We should have known that eventually the progressive [c.q.] wing of the Democratic Party would wake up and galvanize opposition; now an increasingly loud swath of Americans, mostly on the left, seem to hate data centers the way they used to hate your SUV, your Big Mac, and, well, you."

The category error the self-styled progressives persist in committing is calling attention to the downsides of those trade-tested betterments: you couldn't circulate a manifesto to millions of followers using a teletype machine and stepper switches; you couldn't run multiple errands on the way home from school or work without a motor vehicle; and somehow Super Sized humans are a different sort of public health problem than starving humans.  Trade-tested betterments emerge, which is to say, the people see improvements and act upon them.  Never mind the super-optimistic claims some boosters of digital technology make, such as Reason's Christian Britschgi's  "Advances in artificial intelligence and robotics could liberate humanity from boring, backbreaking labor. The early profits of data center development are a leading indicator of the increasingly productive economy that awaits us in the years to come."  Well, no, past performance is not an indicator of future results, and those "early profits" were true of early twentieth century interurban railroads, or of your pet bubble investment.  On the other hand, those stepper switches and relay stacks could not enable me to carry on at length about creative destruction and kids who never had to dial 9 for outside, or Charles Cooke to run some errands with Apple Play on his Handy.

That noted, the Babylon Bee, whether being satirical or not, correctly identified the political economy of information technologies.

Computer scientists estimated that roughly one-third of all data center processing is currently tasked with creating, circulating, and storing content about the dangers of data centers. "It's really the 'raison d'être' for data centers," explained computer engineer Varsheesh Patel. "Ironically, we probably wouldn't need a single new data center if people stopped lashing out at them. Ah, well. It's great job security for me."

At publishing time, Governor Shapiro's office had announced that the computing power provided by the new facility will allow the state government to collate citizens' concerns about data centers.

Nicholas Freudenberg raises similar questions for his Common Dreams followers.

Who should decide whether communities allow data centers? Who pays the long-term costs and reaps the benefits? What role should voters—rather than tech industry executives—have in directing the future of AI?

By offering clear, practical answers and using the experiences of past activist campaigns, the growing data center-AI movement can put these questions on the nation’s electoral and political agenda. In doing so, they can become a powerful force for healthier, safer, and more just communities and a world where AI is used to benefit humanity.

That's long been the political economy of commercial policy, whether it's a steel works or a feedlot or a data center.  There's a University of Michigan discussion paper, "What Happens when Data Centers Come to Town," that despite taking a position on behalf of Michigan's Environmental Justice Coalition, will reward careful study.  Those dimensions of the story will keep for the present.