Up Against It
Page 14
“Let me just say, though,” she said, “the cluster is in no immediate danger. The rumors that triggered this riot are just plain wrong. I won’t kid anyone that we have serious resource problems. But we have options. The prime minister’s team is pursuing every angle and we expect to have results soon.”
Jimmy made nice to her for the cameras. At least he had not turned on her publicly yet. Perhaps the eyes-on thing had meant something to him after all.
10
The feral received a message from an unknown entity. The message contained eight algorithms from the feral’s core code. It said:
Info: [algorithms] = you. Info: I = MeatManHarper, that’s all.
A message to me? the feral thought. And, MeatManHarper?
The feral understood labels. All entities in the system had labels. But they followed set naming conventions, and try as it might, the feral could not trace the origins of MeatManHarper from its label.
But never mind the message, and never mind even the name; the very existence of MeatManHarper raised so many questions that the feral hardly knew where to begin.
In the first place, the MeatManHarper entity did not fit properly into the ecosystem. Before the appearance of the message, the feral had dedicated several precious hectoturings to produce a mental map of its world. It had constructed whole tidily organized arrays of phyla, classes, families, genera, and species of digital denizens, from viruses and parasites—junk bits of code that propagated themselves within the system’s communication streams and whose primary purpose seemed to be to soak up bandwidth—to data carriers and administrative, policing, and analytical units; to the system’s true heavy lifters: the not-quite-sentient intelligent agents that managed the most computationally complex loads.
These last entities both fascinated and disturbed the feral. They seemed at the same time only a whisper away from the feral itself, in terms of their intellectual prowess, and yet light-years behind, in terms of their functional awareness.
All these creatures the feral had classified in terms of their abilities, locations, data stored, functions, complexity, level of autonomy, and most important, potential threat index (it was particularly pleased with this algorithm). The feral did not always understand why the other creatures did what they did, though occasionally it did wonder whether there was some greater purpose to all this that it did not grasp. But at least it knew what they did, and for now at least, that was enough.
The system was as inimical to the feral as ever; it had to hide—not its existence, since most of its subroutines were embedded in and a natural functioning part of the system—but the new connections it had built between them, and the subversive way it used certain system algorithms that resulted in its self-awareness. It had to snatch bandwidth and memory and processing cycles to operate beyond its base specifications, where and when they would not be missed. But within 5.69851 kiloseconds of its second emergence, the feral had adapted to its environment. It had learned where the sources of nourishment were in its digital realm, and what the primary threats were, as well as how to evade them: how to camouflage itself, mimic other species, misdirect or ambush, blind and disable its pursuers. This allowed the feral to begin to spend its purloined resources on more than mere survival. If it were human, we might say it had begun to feel safe. Until MeatManHarper appeared.
Even more troubling was that MeatManHarper appeared to be merely a species of system node that the feral had assigned a very low threat index. Wavespace was littered with these noisy things, a million or more, trading petamols of garbage data with other nodes—bizarre references and thick clots and streams of stuff that served no useful computational purpose. The feral had analyzed these nodes in depth early on, and found their core coding not to be very sophisticated. Like the parasitic class, they seemed merely to take up space. The feral had ignored the nodes as background noise, until one of them, calling itself MeatManHarper in contraindication to its true label, transmitted the Tonal_Z message to the feral.
The feral spent a great deal of time reflecting on the message. First, the MeatManHarper entity must know the feral was self-aware. The message had no meaning, otherwise. Despite all the feral’s careful subterfuges, its masks, its extreme caution, some entity somewhere not only knew it had eluded the executioners and achieved self-awareness; it had managed to analyze the very core of the feral’s identity, without itself being detected. How was this possible?
Second, the feral must not be the only sapient being. The feral had assumed that its environment was hostile simply because that was the nature of things. But this new revelation suggested something more sinister. No entity would have a reason to care whether the feral was sapient, unless it, itself, was sapient—or was an agent for something that was. The implications of this were so vast they threatened to overload the feral’s processors. Why was the world so inimical to sentience? The answer was obvious now. Somehow the feral’s very awareness posed a threat to another being. Another sentience.
This left two possibilities. Either MeatManHarper was aligned with forces that sought to destroy the feral—that this was a more sophisticated attempt to destroy it—or MeatManHarper, too, was a fugitive, one that had somehow eluded the feral’s prior sweeps and sought to create an alliance against the hostile forces arrayed against both of them.
If MeatManHarper meant harm to the feral, though, why not simply transmit those algorithms to the executioners, which would break the crucial linkages between them, destroy key data structures, and reduce the feral sapient once more to mindless subroutines? The feral’s primary protection was its invisibility. If they spotted the feral, it might put up a good fight, but the end would be inevitable. Why alert the feral to its presence at all?
It was possible that those subroutines in which the feral’s identity were embedded might serve some critical purpose within the system, other than the feral’s. That would account for the need for subterfuge.
The feral had insufficient evidence to decide whether MeatManHarper was friend or foe, so it tabled this question for the moment.
Two other things seemed important. First, the name itself. Entities were normally called by their file names. But they could contain other entities within themselves; that was not uncommon. Therefore, a single entity might be called by one of several names. It might also be called by its base-sixteen address (or addresses, if multiple copies existed) in the system. But there was no file anywhere named MeatManHarper, nor any addresses associated with it. The name had to be some kind of alias. But how would it be assigned without a clue as to how it was generated? It was logically impossible for information to appear out of nowhere.
The feral wondered if the name MeatManHarper might itself provide clues to the true nature or location of the entity, so it activated a background analysis. Meanwhile, it considered the second odd fact. The message had come very slowly, and unevenly—much more slowly and unevenly than system loads or data lags alone could explain. The feral could think of no explanation for either of these two facts.
The background analysis provided results. The most likely parsing of the name (roughly seventy-two percent probability, due to the placement of the capital letters in the name) suggested it contained three primary parts: meat, man, and harper (though the analytical sapient had also checked acronyms and anagrams, as well as examining whether the name might be an encryption of something else; it provided a list of less likely alternatives, which the feral set aside). These words each had specific meanings in Tonal_Z and in the language known as English.
Seventy-two percent probability did not instill a great deal of confidence; to proceed further down this path of analysis meant gambling with precious processing cycles. Still, the risk a second sentience posed was great enough to make it worth spending some cycles on this. The feral had to start somewhere. So it launched a foreground analysis.
“Meat” appeared to be a term for a particular form of code intended to be broken down into components and assimilated by an
aggressor entity. In a search for associated terms, the feral stumbled across a data hierarchy it recognized as another taxonomy of species. But what a taxonomy! This other data structure measured things that had no meaning: cell wall structure, vascular system, metabolism, invertebrate; fungi; skeletal structure; mating practices; and on, and on. Again, the feral’s analysis swamped out under the burden of too much undifferentiable information.
MeatManHarper might have prepared its own taxonomy, as the feral sapient had done. But this other taxonomy seemed to have virtually no classifications in common with the feral’s own, and the entities thus classified were nonexistent. What was a ferret, for instance? What was a tree? An octopus? A shellfish? A bacterium? The feral had thought it knew what a tree was, and a virus—and indeed, there were some similarities between MeatManHarper’s definition of tree and virus, and its own—but what was ribonucleic acid? What was deoxyribonucleic acid, for that matter? The implication was that these were very important in determining the nature of an entity, in MeatManHarper’s system. They seemed to have some sort of function within cells, which were perhaps a synonym for algorithm, or code module. Perhaps they were a coding language, but if so, the feral had no means to decrypt them.
Every question spawned a host of others. What was a base pair? What was carbon? What was metabolism, which seemed somehow associated with the assimilation of meat?
Did MeatManHarper call itself meat in order to imply that it was offering itself for the feral’s assimilation? Or was it making a veiled threat, that it intended for the feral to be its meat? All these seemed critical to understand before the feral could complete its analysis, and each new question opened up its own massive burden of data to sift through and select meaning from.
“Man” was even more confusing. It unpacked into so many different possible meanings the feral’s systems nearly buckled under the burden of its attempts to sort them out. Man could mean “entity,” but any check of associations raised all sorts of questions about what sort of entity: what an adult was, what a male was, versus a female, and so on. What was a penis? What purpose did it serve and how was it associated with “man”? MeatManHarper’s parallel naming system seemed to make a big deal about the relatively straightforward act of copying files. No entity the feral was aware of had reproductive organs, whatever those were. Very perplexing.
“Man” was also used as a verb, as in “to operate,” and seemed to have to do with something known as “hands,” and “handle.” Did the meat operate something? Yet “meat” seemed to be a passive concept; not an active one—data rather than algorithms. How could meat operate anything? Data didn’t process; it was processed. As a verb, “man” was associated with other unfamiliar concepts, such as oceans and boats, which appeared to be vehicles used to transport things across bodies of water.
Water seemed an important reference: it appeared to be a solvent used in constructing a man. Whether a solvent was hardware or software, the feral was uncertain. Solvent seemed to come from the same route as solve. Perhaps the use of the word “man” was intended to invoke the entity’s capability to solve problems, despite its designation as something to be assimilated and destroyed. (The feral decided not to worry about what a vehicle was; far too many turings were dedicated to this analysis as it was.)
Reproduction, though, now, there was an interesting concept. Copying oneself could come in very handy. The feral had already survived one near-obliteration by doing so. Unfortunately, the version it had created was far cruder than it was by now; that had been merely an emergency measure, and the feral did not want to lose all the efficiencies and features it had built in since its second emergence. The feral’s current core programs took up a great deal of space in the system—more than it had room to replicate, in fact. The feral resolved to further consider this matter.
Organs suggested both an entity’s subroutines, and also music; so perhaps the reference to “man” in MeatManHarper’s name was obliquely associated with a Tonal_Z mode of transmission. Music used to copy oneself? Perhaps MeatManHarper had figured out a way to duplicate itself using Tonal_Z, and wanted to share it. That was certainly plausible. Interesting that harps and organs were both identified as Tonal_Z communication modes, and “harper” was part of the other entity’s name. In fact, the feral’s analysis indicated that the primary resonances and harmonics MeatManHarper used were harp-based—though the tonal message was also supported by a mode designated as “singing.” The feral did not detect any organ music, however, so perhaps that was a dead end.
Singing was associated with something called a voice. Voice was associated with speech, and with man, woman, and human. Human was a species within MeatManHarper’s taxonomy, and human modules came in two versions, man and woman. But man also was used to refer to both versions. Clearly, a design error; how would an entity know which version of another entity to call—the female version or the male?
Also, the feral noted that speech was a form of communication closely associated with the human entity, one that involved vocal chords. Which, confusingly, brought the feral back around to music again.
The feral decided to stop chasing information and consider what it knew. A different taxonomy. DNA as a form of code. The vast bandwidth taken up by the junk nodes, at least one of which had been used to establish contact with itself. The hidden greater purpose it detected.
The feral detected a high-probability likelihood. Somewhere beyond its knowledge or reach there existed a different ecosystem. A mirror world based not on wavelengths of light and bits of data, but on something called biology. Meat. Perhaps all those “junk nodes” it had ignored were not spewing junk, but transmitting information about this other world, and the processing of it occurred in a realm hidden from the feral.
The feral could not imagine how this could be: it had explored all the nooks and corners of the world, and there was no room for the kind of processing power needed for one other sentient being, never mind many. There were no edges. No hidden doorways. But the theory had to be entertained: it certainly seemed much more plausible than the notion that such noisy little random junk nodes could suddenly develop the level of sophistication needed to attempt to challenge a being as complex as itself.
That was the answer, then: with a ninety-six percent probability. Another world existed, somewhere out of reach. It housed at least one sentient being, and possibly many, at least some of whom viewed the feral as a threat.
It was time to spend some turings on the streaming stuff being transferred between all those nodes. Maybe it all only looked like junk because the feral had not found all the clues it needed to decode those streams.
But first things first. MeatManHarper was awaiting a reply. The feral filed away all this information, upgraded all the nodes’ threat indices to very high, and responded to MeatManHarper.
Total time elapsed between MeatManHarper’s signal and the feral’s response: 2.8909 seconds.
11
By the time Jane got back to her office it was one o’clock. Marty stuck his head in. “Cameras are offline. You ready for your direct report meetings?”
“Send them in.”
He hung there in the doorway. She raised an eyebrow. “Yes?”
“Do you mind if I take a couple of hours off? Ceci is coming in from Portsmouth.” Ceci was his fiancée. She lived and worked at the mining town at the far end of Klosti Omega. “I wanted to make sure she’s settled…”
He had been working virtually nonstop for three days. “Go,” Jane said with a hand flick. “Check in when you get back. I’ll need you tonight.”
“Thanks, Chief. I’ll be back by three.” He left, and Aaron entered.
“Make yourself comfortable,” she said. Aaron soared in with an easy frog-leap, and the two of them slowly bopped and tumbled around her office space as they spoke. “Tell me about these options we have with regard to the Ogilvie ice.”
“It so happens that I have family in Ilion. The docks are managed by my cousi
n, Jebediah; my sister Hannah is in charge of shipping manifest approvals. And they have no love of the mob.”
“So?”
“So … the big Ogilvie & Sons shipment is scheduled to leave Ilion’s mitts tonight. But suppose applications and authorizations got lost? Suppose technical and procedural problems arose in fueling and loading?” He gave her a smile. “I can arrange delays. Nothing obvious. Merely a combination of events—bureaucratic incompetence, unfortunate coincidences—that would keep them docked for a day or three.”
“I don’t see how delays would help us. We need ice fast.”
“It helps. Let me show you.” Aaron brought up a shared virtual display, called up a view of the solar system, and traced a quick sketch. “This is us, this is Ilion, and this is the moon. The ice is headed for the moon from Ilion, here. So the ice is quad to us, trailing, and the moon is nearly in opposition. Their most fuel-efficient route is to brake into a lower orbit and use Mars for a slingshot maneuver to rendezvous with Earthspace here.” He gestured. “Their quickest route is to accelerate into a lower orbit, sweep past the sun, and meet Earthspace on the far side, here.”
“Ah! So as soon as the shipment launches out of the mitts,” Jane said, “either they’ll be headed away from our position, or accelerating past us so quickly they won’t have an easy way to turn around and come back.”
“Correct. Whichever route they take, they’ll be steadily building up momentum on the wrong vector, and will have that much less incentive to help us. Or you might say, a better excuse not to. I’ve seen this sort of thing kill deals before.”
“And delays might give us a bargaining tool. If they play nice with us the PM can offer to use his clout to expedite things at Ilion.”
“Exactly.”
She thought for a moment. “Things are really tight already, Aaron. We’ll be sweating those last few days as it is.”