13

Sal glanced at his interface as he attempted to place the deception. Felicia had tasked him. All he had to do was make a pencil appear in a glass of water as it would naturally, along with the apparent bending of light that would make the pencil look like it was broken in two. Normally, this would be a simple task, but after having chatted with Felicia, he knew it wouldn't be.

Sal, was, in a sense, fighting for his very survival with this deception. What he had learned from Felicia had shocked him. If what she had told him proved to be correct, and it looked like it probably was, it would mean a radical shaking of reality as he had understood it was about to take place.

Felicia had been working with Telray for about five years on quantum algorithms that used Bayesian inference. Bayesian inference had been largely ignored in many purist communities where statistics were used, mostly because the mathematicians liked the supposition made by the frequentists far more than the inelegant assumptions that the Reverend Bayes had made in what was now nearly ancient history.

The frequentists liked to look at reality as though it were a pure statistical curve--or a frequency diagram of probabilities. This allowed them to state that something could happen with a certain probability. Bayes had turned that thinking on its head, instead, he said that prior information had a bearing on future information. If reality was a curve, then if one had a prior hypothesis that prior hypothesis was somewhere on the curve. If one kept making hypotheses, then eventually one would arrive at the "correct" one with a certain probability. What probability was was influenced by the previous probability.

The frequentists had strength when it came to matters that involved purely mathematical problems free from the real world. However, in a world without full availability of information, Bayes had more applicability.

The easiest way to think of it was to consider the Old World Internet. The Old World internet was like a great-great-great-great-great grandfather to the holonosphere. It was more akin to the postal service in the now than anything else. On the Old World internet, people would often receive spam or unwanted commercial messages. The problem with a frequentist approach to spam is that one cannot easily assign a probability to a given message in a vacuum. You need a little more information. Who is the message from? What form does it take? What does it look like compared to other spam messages? What sorts of words does it use? If you can then draw on a database of previous spam messages, it becomes possible to conclude to a small degree of error utilizing Bayesian statistics whether or not the current message is spam. The prior messages matter. They act to prime the information--they inform the distribution. Without them, one cannot begin to assign any meaningful probability--especially not one that helps significantly reduce spam.

With the advent of quantum computing, it became possible to traverse multiple universes and their respective curves. In other words, in addition to being able to calculate whether THIS message was spam, it was also possible to calculate at the same time if in another universe the spam assumed some other form than the one here what the probability of THAT was too. When taken all together, one could compare the given message to a "universal curve" or "curve of all curves" in all possible universes, and so one could say with a very, very high degree of accuracy whether a given message was spam.

Felicia and Telray were not interested in spam, however. What they were planning to do with the technology Sal could only guess, but he knew the implications.

Sal wasn't now dead because Telray had hired him to be a guinea pig. Telray had implemented Felicia's algorithm and wanted to see whether a very talented deceiver such as Sal could break it. Instead of using the algorithm for the detection of spam, however, Telray intended to use it to make long-lasting adaptable changes to the holonosphere. In the instance that Sal had been employed, they were using the algorithm to cancel out changes. The reason Sal could not place the hill he had been hired to was because the algorithm was in the background calculating out prior states and nullifying his changes. He couldn't break through because the holonosphere along with the algorithm was being used in such a way as to reinforce physical reality.

When Felicia had told him this, Sal had grimaced.

"Even if I were to believe what you say," said Sal, "there is no algorithm that exists that does not have a weakness. It's just a matter of time before someone breaks it. More than likely, I will be that person."

Felicia had smiled a condescending smile. "Well, if you feel that way, Mr. Grimone, I will be happy to provide you with the extra time you need to make that happen. In front of us on this table is a glass of water. Why don't you produce something small and less challenging than a hill in it? How about a simple pencil?"

"Just give me the time and I'll get it done," replied Sal dismissively.

"I'm sure you will, Mr. Grimone. I suppose I will leave you to your work."

Sal made it a point NOT to pay attention to the swishing way in which Felicia left. He really would have preferred his adversary be male. It was easier to keep things in perspective that way and to keep his mind from wandering down evolutionary corridors that were only distracting.

Since she had left, Sal had been staring at his interface, trying to find inspiration. The one advantage he had now was that he knew what was going on, and how it was being done. He just needed to find a way to break it. He also needed to do it in secret.

As far as Sal knew, Telray wasn't paying him for his work this time. Perhaps if he proved useful and found some hole they might, but he didn't care to become their in-house debugger. If he found a hole, he would keep it to himself. More than likely, Telray would be monitoring everything going on in his interface, but that didn't concern him.

Since Sal had made a career out of deception, he had taken the appropriate steps to ensure that his interface would be guarded in case a situation like this ever arose. In a prescient moment, he decided that he might need to deceive someone someday concerning what he was doing. That time was now. He developed a system that was a subinterface to his normal interface which was only accessible if he provided the appropriate key phrase. He had thought long and hard about what his key phrase should be. It needed to be something hypothetically believable in the context of whatever it was he was doing so as not to raise suspicion. He decided it needed to be a common error--something someone might say then correct themselves on. For him, that phrase was "colon slash colon---correction colon slash" while in diction mode which he frequently used to rattle off key pieces of code. It was a common mistake, but one he did not make. If he did accidentally make it, he had another keyword to turn it off. Better to have it easily on when needed and experience accidents than unavailable.

Once this mode was entered, his auxiliary interface would be activated. This auxiliary interface did not run in the normal sense. It didn't even display. In fact, it was totally cut off from the rest of his system, as it existed only as a virtual interface. This virtual interface would then begin. Within this interface was an AI that Sal had nicknamed Belvedere. Belvedere's job was somewhat like that of a butler. He was to watch whatever Sal was doing completely silently and make some inferences in conjunction with what Sal appeared to be doing in his actual terminal. So, for instance, if it looked like he was trying to create a deception with a cup of water and failing, Belvedere would know the intent. Once this was established, Belvedere would clandestinely try different lines of attack to accomplish whatever it was that Sal was attempting to get done.

Sal waited a good hour before making the mistake that invoked Belvedere. The first thing Belvedere did on springing to life was to take a picture of the state of the holonosphere in the immediate area. This picture was a frozen moment in time which Belvedere could then run attacks against for as long as it needed to get the job done. If the job in question had a limited time duration, Belvedere did not have to worry about changing conditions as a result of the time expiring. He could just keep cracking at whatever the task was, and when Sal got done doing whatever he was doing, he could join Belvedere if it proved necessary. Belvedere had his snapshot and began working silently in the background.

Sal had additional problems of his own. He had to make it LOOK like he was actually trying to crack the problem. He tried exceedingly simple solutions at first that he assumed would not prove to be successful. He was very glad that none of them succeeded. As time elapsed, he tried more complex varieties that he felt would be equally unsuccessful. Nothing seemed to be budging it, but even the responses he got from the environment from trying what he knew would not work were telling him something about what would work. Belvedere made note of these things to narrow its own search scopes. It was almost like being in a chemistry lab where you simply kept throwing chemicals together in an effort to make rubber. What you learned by noticing what didn't work moved you toward what might.

After about six hours, of apparent non-success, Felicia returned.

"I see you are still working so hard, Mr. Grimone. Ready to call it quits?"

Sal wondered whether Belvedere had had enough time to work his magic. He had programmed Belvedere to signal him should he achieve success. Again, Sal wanted to be sure that by signaling he did not give away the covert interface, and he had to think quite a lot about what might be a good signal that no one would notice. He finally concluded that a visual subtle signal would be best, and the thing that made the most sense to him was for his cursor to blink rapidly twice. He had arranged some other sequences should Belvedere need to communicate something else, but a rapid double flash of his cursor once would tell him Belvedere had cracked the problem. He had not yet seen such a signal.

"I'm not sure I am, Felicia. Six hours really isn't all that long to solve something like this."

"Well, it is going to have to be long enough," said Felicia.

"You must be really scared I'm going to break your code," replied Sal.

"Hmmph. If it were up to me, I'd give you a solid week with all the meals you could eat to try to break what I've put together. Unfortunately, it is not up to me. Telray is satisfied that my work is good enough since you have not been able to find a way to crack my algorithm in the six hours that you have been here, plus the failure of the terrain modification."

"Well then, why don't you go tell your bosses to go fuck themselves?"

Felicia smoothed the lapels of her jacket for a moment before she answered. "Some of us have found that having manners goes a lot further than being a rebel without a cause."

"So what you mean is that if you behave like a good dog, your handlers are happy to give you pats on the head? asked Sal.

Felicia's brow furrowed. "No, that is not at all what I mean."

"Could have fooled me," replied Sal.

"I already have, Mr. Grimone. Twice."

"Listen Felicia. I don't know what your handlers have taught you other than how to be an obedient little bitch, but your code will break. If you have enough pride to think it won't, then you are not nearly as smart as you try to project."

"I do not need it to be unbreakable, Mr. Grimone. Just unbreakable enough. Nonetheless, your time is up, and Telray has no further need of you. I suggest you shut your terminal down so that I may escort you to the other door--unless you would prefer the slipstream travel that brought you here."

Sal glanced at his interface to close it. He kept his vision firmly on the cursor. The instant before the interface disappeared, it blinked twice.

Sal turned to Felicia, "Well, I guess you can't win them all," he said. "Better show me out before I prove how misplaced your boss's confidence is in you."