by Paul Chafe
A dolphin with her own ship? Tskombe suppressed his surprise. He had assumed the dolphin worked for Khalsa, a humanocentric mistake. “And who does Curvy work for?”
“Curvy is Senior Strategist for the Secretary of War. In point of actual fact, the ex-Secretary of War.”
“Ex-Secretary?”
“Muro Ravalla forced the no-confidence vote this morning, well ahead of the schedule we predicted. Secretary General Desjardins is out, and the entire cabinet is being replaced. Which means that you, I, and this ship are now in legal limbo. None of us are going back to Earth for a long time.”
“What happens now?”
“That depends on you. Would you like to talk to Curvy?”
“How's Trina?”
“The girl? She's still getting over the mercy needles. She took a lot, and she's half your mass.” Khalsa looked annoyed. “You risked a lot to get her on board here.”
“It was worth it.”
“What is she?”
“She's a street kid, unregistered.” Tskombe let Khalsa fill in the implications behind that. “She doesn't have much of a life on Earth.”
“A noble gesture to save her from her life.” Khalsa smirked sardonically. “What do you plan to do with her here?”
“We'll drop her with the Bureau of Displaced Persons on Wunderland. Get her some care and an education. She'll have a chance at least.”
Khalsa pursed his lips. “She's here now, it won't cost us anything. Not that you didn't nearly ruin this operation for the rest of us. You couldn't have been more conspicuous if you tried.”
Tskombe shrugged. “I couldn't leave her.”
“You still haven't answered my first question.”
Tskombe nodded. “I'll talk to Curvy.”
He followed Khalsa down the short ladder from the navigation blister. He was surprised the dolphin was on the ship, both because he had thought she was one of the ones who'd rescued himself and Trina from the sinking gravcar, and because there was simply no room for a dolphin tank on a ship the size of Valiant. He quickly learned that Valiant was the exception. The courier's entire cargo area had been converted to dolphin quarters. Tskombe found it somewhat surprising that the UNSN intelligence held a dolphin in enough esteem to give her unrestricted use of a ship. His next surprise was that Curvy didn't spend all her time in the tank, which only occupied half of the cargo compartment. The other half had the gravity switched off, and she received her guests there. She was wearing a flexprene body suit with grav polarizers built in and tanks and filters designed to keep her skin moist at all times, as well as a set of dolphin hands. In front of her was a command console, its holocube displaying an array of chessmen. The manipulator tentacles on the hands clicked over the databoard and the pieces moved in response.
Tskombe was about to speak, but Khalsa held up a hand for silence. The game was moving almost too fast to follow; a move per second seemed to be the rule. They didn't have long to wait before the black king flashed in defeat and the board reset itself. Curvy was playing white.
She turned to face them, popping and whistling. Her translator spoke, flatly inflected. “Welcome back, Colonel Tskombe.”
“Curvy. I assume I have you to thank for the rescue?”
“Indirectly only. The dive crew who intervened are my family pod. They were prepared mostly to assist me in event of emergency. As it was they assisted you.”
Tskombe bowed. “I thank you then, for them.”
“As I said, dolphins have long cooperated with humans. I have questions for you, Colonel.”
“Ask.”
“Why did you return for the female?”
“It had to be done.”
The dolphin-hand tentacles tapped the console and the chessboard vanished, replaced by numbers that began flowing through the display cube. “You are more impulsive than we had estimated. The strategic matrix needs updating.”
“Does this upset your plans?”
Curvy contemplated the numbers for a long minute, tapped more with her manipulators, considered a three-dimensional graph evolving in time. “The outcome space is shifted to a more extreme domain.”
“Meaning?”
“Your correlation coefficient remains essentially the same for both positive and negative outcomes. However, the scale of those outcomes is increased in both directions.”
“So I'll win big or lose big.”
“The strategic matrix does not apply to your personal outcomes but to our organizational goal outcomes. Making predictions for individuals is difficult and error prone.” Curvy's manipulators tapped on her databoard. “Does Captain Cherenkova remain your primary sexual interest?”
“Trina is just a girl.”
“I will assume that means yes.”
“Yes.” Not that Trina won't be a very desirable woman soon. But soon was not now, and the least of what drew him to Ayla was her physical beauty.
“Understood. How much additional risk are you willing to undertake to accomplish our mission in parallel with yours?”
“How do you quantify risk? I'll work to prevent a war, if possible. I don't believe that anything I can do will influence the Patriarch significantly.”
“Your willingness is the important factor. Remember that your actions do not exist in a vacuum.” Curvy's manipulators tapped more. “Thank you, Lieutenant Commander, I believe I have the necessary updates.”
It might have been the end of the interview, but Tskombe had questions. “So tell me about strategic matrix theory.”
Curvy pivoted in her grav suit. “What do you want to know?”
“I want to know how a dolphin can ask me three questions and make predictions about a future war between the Patriarchy and humanity.”
“You must have studied game theory in staff college.”
“Some.”
“Game theory is to strategic matrices as quantum theory is to statistical mechanics. In game theory each player has a number of strategies to apply to a given situation. Choice is assumed to be constrained to a small number of alternatives. A strategic matrix combines many, many games, each of whose players are constrained to some range of choice. The key is, these choices can then influence the choice set of subsequent iterations of the matrix. In raw form this is a highly exponential problem, making large-scale predictions impossible over any distance in time. Intuitively we know this can't be correct; a society as a whole is far more predictable than any particular individual in that society. In a strategic matrix we use statistical techniques to constrain the global choice set and thereby determine a range of possible futures. As it turns out, and as a few seconds' thought will tell you, most choices made by most individuals have almost no impact on the course of the global future, which inevitably means that a few choices have tremendous impact. With some probability of success we can back-trace outcomes to choices and the individuals who make them. Once we've identified such an individual we can then try to influence their choices to bring about a desirable outcome in the future.”
“That's as clear as mud. Do you think you can really predict the course of events on a large scale?”
“It isn't even that hard. All we are doing is quantifying the process.” Curvy pivoted in her wet suit to tap keys. “'Ceterum censeo Carthaginem esse delendam.' Predicting that Rome would go to war with Carthage is straightforward. With two major powers in the Mediterranean any outcome other than war was impossible. History shows us this with exquisite clarity, and in fact the wars continued until one side was destroyed, as such wars do. Not until conversion weapons were a serious factor in war did great powers resolve their differences other than through combats where the goal, if not always the outcome, was the annihilation of the enemy. This change was also predictable, and was predicted even before strategic matrices by those with the vision to see it, due to the effect that conversion weapons had on the outcome sets of those empowered to make decisions that might lead to war.”
“Meaning what?”
“In
the era of nation-states human leaders proved willing to exterminate millions or tens of millions of the enemy, and to allow a similar fate to befall their own populations in the name of victory, because their personal outcome set was highly positively biased through the war process, becoming negative only in the relatively rare case of total defeat. Conversion weapons brought the same highly negative outcome set to the leaders that the populace had always faced, and they are lethal enough that the leaders' best efforts could not mitigate that change. The growth of United Nations power and its eventual transformation into de facto world government is traceable directly to this, again an unsurprising result given the initial conditions. This is the predictability of large-scale social systems.”
“And you can do this with precision now.”
“Within limits. The wars between Carthage and Rome were inevitable, and predictable. However, predicting that Carthage would be destroyed in the war launched by Cato the Elder is much more difficult, yet clearly the events are intimately linked. Still we can assess that Cato the Elder was far more likely to influence this event than Caecelius, merchant of Pompeii. A predictive strategic matrix allows the appropriate links to be made, in probabilistic terms, a priori. We can then assess unfolding paths in global choice-space.”
Tskombe shook his head. “I find it hard to believe you can make predictions like that with enough accuracy to be useful.”
Curvy squeaked and squealed, something that might have been laughter. “Do you invest in the markets, Colonel?”
“Not directly. My money sits in a floating investment fund. The bank invests it for me.” Or it did. Reflexively he patted his beltcomp, where his total net worth sat in encrypted authentication codes.
“And you pay a fee for this service, yes?”
“One percent of all profit over the market average in a year.”
“Why pay the fee when you could invest yourself?”
“I lack the experience and expertise to do it myself. And more importantly, the time.”
“So how much better than the market average do you think your fund manager does?”
“I don't know, I've never thought about it.”
“I can tell you that without knowing where you bank.”
“You mean you don't have those details at your fingertips?” Mild sarcasm.
Curvy chirped something and the translator said “Untranslateable,” then “It works like this. Every day there are some stocks on the exchange that move many percentage points. Up or down does not matter. If you could predict those stocks reliably, you could make a fortune. We assume in market analysis that a priori evidence exists that can be used to make such predictions.”
“That's blindingly obvious.”
“Of course you can't win every bet. So how much better do you think your good bets have to be than your bad bets to start making money?”
Tskombe shrugged wordlessly.
“One percent. If you can consistently come out one percent ahead on every trading day, consistently reinvesting what you make, you'll triple your money in four months. If you could manage to be just five percent better your wealth would mount so fast than in a year you would own the world.”
“That seems impossible.”
“Do the math. Compound interest is powerful.”
“Literally own the world?” Tskombe's voice was heavy with disbelief.
“No, because you would start to own too large a fraction of the market. As your wealth becomes a significant proportion of the market you lose the ability to concentrate your fund base in those high-yield investments. You cannot beat the system when you are the system. However, on a smaller scale the principle applies.”
“So what does all that mean?”
“It means that market analysis doesn't work. Your fund manager's best efforts are no better than random choice. If he were doing even a tiny fraction of a percent better his results would be phenomenal, and you would be wealthy beyond the wildest dreams of avarice. Every day he has the opportunity to make not fractions of a percent but many percentage points, and yet he fails to do this, despite all the sophisticated predictive tools he employs.”
“That can't be true. He only gets paid if he makes money for me.”
“True, but it doesn't cost him anything if he loses money for you. The situation is asymmetric, to his advantage. He bets with your money, shares the gains, and leaves the losses to you. Suppose he splits every investment both ways. Half his customers will lose and half will win, and he'll collect his pay on the winners. The losers curse his name, but as long as he can replace them with new clients he doesn't need to care.”
Tskombe blinked. I'm talking about the stock market to a dolphin. “So if I understand you, you're telling me strategic matrices don't work.”
Curvy whistled. “Exactly correct in the case of markets. Everyone is trying to make predictions based on the same information, but once you make a trade based on a prediction you change the market. Since everyone is processing the same information at the same time any advantage you might have is immediately erased. You cannot predict those large market moves that are reactions to events which are by definition unpredictable. If they were predictable, they would have been widely predicted and so any advantage you might have extracted is immediately erased. It is theoretically possible that you could develop a better method of extracting information from the market, and hence retain an advantage, but this is very unlikely, and you would retain your advantage only as long as you could keep your methodology secret, which in practice is impossible, since there exists a large segment of traders who analyze the investment patterns of the rest of the market in order to determine exactly these strategies. Because of the feedback loops formed in this way the market is inherently unstable. In that it's just a mirror of the real world, which is in fact exactly what it's supposed to be.”
“So strategic matrices are your predictive tool. Why do you spend so much time and effort developing the damn things?”
“There is a fundamental difference. We apply them to the socio-political arena, which does not include feedback loops as large or rapid as those at work in the market. We can in fact make reasonable predictions and expect to be correct, within error bars. The future may evolve along widely different lines, but the lines, at least, can be assessed in advance, the critical path junctions identified, and sometimes we can act to induce events to unfold along one path instead of another, by influencing individuals who in turn can influence events at these critical points.” The dolphin paused and seemed to be reflecting on something. “It is of course an error-prone process, and we lack the ability to repeat our experiments.”
“You don't inspire much confidence.”
“Were you looking for confidence?” If Curvy was human her voice might have been sardonic. “I thought you were looking for Captain Cherenkova.” Curvy turned to her console. “If you will excuse me, Colonel, I have much work to do.” She looked at her console, manipulator tentacles tapping. “I would enjoy talking with you again, if you like.”
Later Tskombe sat in the courier's tiny wardroom with Khalsa. The Navy commander produced coffee in bulbs, pretended not to be offended when Tskombe added synthetic crème de cacao from the food processor.
“I don't understand what you're doing here. You've taken tremendous risks to help me, on Curvy's recommendation, but Curvy herself has no confidence in her strategic matrices. It doesn't make sense.”
“Astute observation, but there is another factor.”
“Which is?”
Khalsa leaned back and sipped his coffee. “Consider the Second Punic War. Hannibal is outnumbered, undersupplied, outmaneuvered, still he wins battle after battle. He doesn't win by force of numbers, he wins by outthinking his opponents. This is not a statistical fluke; he manages to beat the odds. In the markets, you've heard of Markland Stage?”
“The financier?”
“He started from nothing and made his two percent per day, until he owned too much
of the system to keep beating it like that. There are other examples. Henry V of England, Erwin Rommel, Gael Sistorny. There are those who do beat the odds, people who are so far out on the bell curve that the math simply doesn't work for them anymore. Consider: the emergence of Hannibal's military genius could have been predicted, in probabilistic terms at least, given sufficient data. Predicting the specific tactics Scipio Africanus might use in any given battle is much more difficult, but this was exactly what Hannibal excelled at. There are commanders, like Hannibal, like Alexander, like Genghis Khan, who not only win consistently despite the material and numerical superiority of their enemies but who defy statistics as well, a much more difficult feat. How they do it we don't know, maybe they're just extremely lucky. The important thing is, they do it.”
“And you think I'm that person?”
“No. Curvy is that person. Strategic matrices are an old tool, but it takes more than tools to make a craftsman. Curvy knows how to use her tools. If she tells me that sending you to Kzinhome will make a difference, I'll send you to Kzinhome.”
“Curvy is not a commander.”
“Dolphins don't have commanders. They don't form large-scale societies, and they don't fight wars, certainly not as we know them. Nevertheless, Curvy's predictions are accurate. She knows what has to go into a matrix, she knows what has to stay out. She beats the odds, Colonel. If she's betting on you, I'm betting on you.”
Tskombe reflected on that for awhile. “Curvy had to update the matrix. She didn't have all the information. How do you know she has it all now?”
“I don't. Updates are an ongoing process. The amount of information entropy, basically how much we don't know, is computed in the matrix.”
“Ever have it happen that new information turns your calculations upside down?”
“Sometimes.”
“What do you do then?”
“What else can we do? We recalculate and keep trying to influence events toward positive outcomes.”
“Positive in your own definition.”
“Of course.”
