Clifford's Spiral: Chapter 20

Chapters are serialized here for paid subscribers.About This Novel

In Clifford's Spiral, the stroke survivor’s past is blurry, and his memories are in pieces. He asks himself:

Who was Clifford Olmstead Klovis?

Stroke sufferer Clifford Klovis tries to piece together the colorful fragments of his memories. Can Jeremy prove it actually happened?

Chapter 20

Data drilling was Jeremy’s favorite thing. Within his own lifetime, it had become possible to query the collected wisdom of the human race in seconds — on any subject. What was Sandy Koufax’s batting average in 1952? What is the average annual rainfall in South Sudan? How do you add metal links to an expansion watchband? What makes the holes in Swiss cheese? Why did the ancient Mayan civilization disappear so suddenly?

During his day job, Jeremy developed and then fine-tuned algorithms for conducting searches of online databases. He’d built on a scheme known as binary trees. When a search engine finds a new piece of information, it stores the path to it as a sequence of directions through the networked information maze of cyberspace. Choice of direction at each junction, or node, is binary — left or right, one or zero. The path to the South Sudan rainfall result might involve a billion such choices: left-right-right-left-right-left-left-left, and so on. Before anyone ever searched for the information, finding the path to it took a while. But once the path was stored (as when a Web crawler hunted down a new keyword), traversing it again to receive the result typically took less than a billionth of a second. But in Jeremy’s world, even those billionths can add up and snarl the works. His persistent goal was to make searches shorter and therefore more efficient. Doing so would make it feasible to search ever-larger and more complex databases — and collections of databases.

Jeremy’s specialty as a programmer was the balanced binary tree. With all those left-right branchings, search trees tended over time to become lopsided, having many more nodes on one side than the other. But in a network of branches that is roughly symmetrical — having as many choices on one side as on the other — the time required to traverse any branch is a short as it can be. The tree at that point is said to be balanced, or tuned. Jeremy had found ingenious ways to rebalance binary trees in the background — performing his balancing act between searches, invisibly. His job was like a mechanic’s on an old Jaguar sports car — the kind with manual rather than hydraulic valve lifters. The newer hydraulic lifters were forgiving, but sloppy. The mechanical type were precise, but fussy. Due to wear, weather, and changing oil viscosity, the valve timings kept going out of adjustment, requiring an expensive visit to the shop every few weeks. And in the days before electronic diagnostic equipment, the veteran mechanic had to develop an ear — like a piano tuner’s — for the sweet spot when the engine was humming at its peak performance.

Jeremy thought of his job as a quest for the sweet spot. His goal was an elusive thing of beauty, all the rarer and more wonderful because few people — not even his colleagues, and certainly not their system users — would ever notice his invisible magic.

Read more