When brokers returned to their desks from lunch, what had been a fairly calm Friday was something else entirely. Every office had a news board that gave shorthand announcements of national and international events, because such things had effects on the market. The notification that the Fed had jacked up its benchmark rate by a full point shocked most trading rooms to a full fifteen or thirty seconds of silence, punctuated by not a few Holy shits.

Technical traders modeling on their computer terminals saw that the market was already reacting. A rise in the discount rate was a sure harbinger of a brief dip in the Dow, like dark clouds were of rain. This storm would not be a pleasant one.

The big houses, Merrill Lynch, Lehman Brothers, Prudential-Bache, and all the rest, were highly automated, and all were organized along similar lines. In almost every case there was a single large room with banks of computer terminals. The size of the room was invariably dictated by the configuration of the building, and the highly paid technicians were crowded in almost as densely as a Japanese corporate office, except that in the American business centers people weren't allowed to smoke. Few of the men wore their suit jackets, and most of the women wore sneakers.

They were all very bright, though their educational backgrounds might have surprised the casual visitor. Once peopled with products of the Harvard or Wharton business schools, the new crop of "rocket scientists" were just that—largely holders of science degrees, especially mathematics and physics. MIT was the current school of choice, along with a handful of others. The reason was that the trading houses all used computers, and the computers used highly complex mathematical models both to analyze and predict what the market was doing. The models were based on painstaking historical research that covered the NYSE all the way back to when it was a place under the shade of a buttonwood tree. Teams of historians and mathematicians had plotted every move in the market. These records had been analyzed, compared with all identifiable outside factors, and given their own mathematically drawn measure of reality, and the result was a series of very precise and inhumanly intricate models for how the market had worked, did work, and would work. All of this data, however, was dedicated to the idea that dice did have a memory, a concept beloved of casino owners, but false.

You needed to be a mathematical genius, everyone said (especially the mathematical geniuses), to understand how this thing operated. The older hands kept out of the way for the most part. People who had learned business in business schools, or even people who had started as clerks and made their way up the ladder through sheer effort and savvy, had made way for the new generation—not really regretting it. The half-life of a computer jockey was eight years or so. The pace on the floor was killing, and you had to be young and stupid, in addition to being young and brilliant, to survive out there. The older hands who had worked their way up the hard way let the youngsters do the computer-driving, since they themselves had only a passing familiarization with the equipment, and took on the role of supervising, marking trends, setting corporate policy, and generally being the kindly uncle to the youngsters, who regarded the supervisory personnel as old farts to whom you ran in time of trouble.

The result was that nobody was really in charge of anything—except, perhaps, the computer models, and everyone used the same model. They came in slightly different flavors, since the consultants who had generated them had been directed by each trading house to come up with something special, and the result was prosperity for the consultants, who did essentially the same work for each customer but billed each for what they claimed was a unique product.

The result, in military terms, was an operational doctrine both identical and inflexible across the industry. Moreover, it was an operational philosophy that everyone knew and understood only in part.

The Columbus Group, one of the largest mutual-fund fleets, had its own computer models. Controlling billions of dollars, its three main funds, Nina, Pinta, and Santa Maria, were able to purchase large blocks of equities at rock-bottom prices, and by those very transactions to affect the price of individual issues. That vast market power was in turn commanded by no more than three individuals, and that trio reported to a fourth man who made all of the really important decisions. The rest of the firm's rocket scientists were paid, graded, and promoted on their ability to make recommendations to the seniors. They had no real power per se. The word of the boss was law, and everybody accepted that as a matter of course. The boss was invariably a man with his own fortune in the group. Each of his dollars had the same value as the dollar of the smallest investor, of whom there were thousands. It ran the same risks, reaped the same benefits, and occasionally took the same losses as everyone else's dollar. That, really, was the only security built into the entire trading system. The ultimate sin in the brokerage business was to place your own interests before those of your investors. Merely by putting your interests alongside theirs, there came the guarantee that everyone was in it together, and the little guys who had not the barest understanding of how the market worked rested secure in the idea that the big boys who did know were looking after things. It was not unlike the American West in the late nineteenth century, where small cattle ranchers entrusted their diminutive herds to those of the large ranchers for the drive to the railheads.

It was 1:50P.M. when Columbus made its first move. Calling his top people together, Raizo Yamata's principal lieutenant briefly discussed the sudden run on the dollar. Heads nodded. It was serious. Pinta, the medium-risk fund of the fleet, had a goodly supply of Treasury notes, always a good parking place in which to put cash in anticipation of a better opportunity for later on. The value of these notes was falling. He announced that he was ordering their immediate transfer for Deutschmarks, again the most stable currency in Europe. The Pinta manager nodded, lifted his phone, and gave the order, and another huge transaction was made, the first by an American trader.

"I don't like the way this afternoon is going," the vice-chairman said next. "I want everybody close." Heads nodded again. The storm clouds were coming closer, and the herd was getting restless with the first shafts of lightning. "What bank stocks are vulnerable to a weak dollar?" he asked. He already knew the answer, but it was good form to ask.

"Citibank," the Nina manager replied. He was responsible for the blue-chip fund's management. "We have a ton of their stock."

"Start bailing out," the vice-chairman ordered, using the American idiom. "I don't like the way the banks are exposed."

"All of it?" The manager was surprised. Citibank had just turned in a pretty good quarterly statement.

A serious nod. "All of it."

"But—"

"All of it," the vice-chairman said quietly. "Immediately."

At the Depository Trust Company the accelerated trading activity was noted by the staffers whose job it was to note every transaction. Their purpose was to collate everything at the end of the trading day, to note which buyer had purchased which stock from which seller, and to post the money transfers from and to the appropriate accounts, in effect acting as the automated bookkeeper for the entire equities market. Their screens showed an accelerating pace of activity, but the computers were all running Chuck Searls' ElectraClerk 2.4.0 software, and the Stratus mainframes were keeping up. There were three outputs off each machine. One line went to the monitor screens.