Novels2Search
Trading Hells
2.15: On how to build a God

2.15: On how to build a God

I can count the occasions I was more eager to leave a room on the fingers of one hand, and I made almost indecent haste to leave the depressing atmosphere. Add in that I had work to do, and I think you get the gist. I did not even wait until the door was closed before I sent the boot signal to Glory. I would need every advantage I could get, and the 30:1 of my implants was not cutting it right now.

As soon as we arrived at the roof garage, I turned to Justin and Ryan.

“I will spend the way back in Cyberspace, to get as much headstart to the problem as I can get. So, I won’t ignore you intentionally.”

Justin snorted.

“Come on, as if we would expect you to hold a philosophical discussion with us in the 7 minutes it will take us to get back.”

I couldn’t help myself and smirk at that.

“Yes, of course, but I will be almost completely unresponsive.”

As I had said, I was in cyberspace almost before I was fully seated, and the familiar chaotic swirls of my mindscape greeted me. I desperately wanted to go fly for a while, but I had a job to do.

Thus, the first thing I did was call up a virtual environment for my new VI. The next was to instruct my NADAs to begin construction of Grendels, just to find that they already had been started a few minutes ago.

Obviously, Warden had predicted my need here. I then began to create the objectives of the new VI.

“Warden, would you please give me a copy of your coding subroutines?”

Immediately, Warden materialized on my side.

“Of course. May I ask why you need to make a new VI for this task? I could go over the source code as well.”

“Three reasons. First, while yes, you could work on the source code, your neural net is not set up for that. I created it to assist me in combat dives. And even if it is marvelous how well you’ve adapted to other needs, it is not what you are really good at. You could compensate by throwing compute at the problem, but that is not the most efficient usage of your resources. A purpose build VI will be able to do all that and more with way less.

And second, I want to use this VI for all software projects that Enki, and our new gaming studios, will have. And I won’t ever give that many people access to you.

And finally, as rational as you behave, in the end, you remain a rogue VI. Yes, the difference is mostly superficial, but people would balk from working with you. If you toned down your behavior, you would probably damage your capabilities to fulfill your primary objective.

Right now, what everybody knows is that you are a rabid monster that I can barely keep chained. That means endangering me would bring a nightmare down onto them. Which is, of course, true. But if you interacted rationally with other people, sooner or later one of them will let it slip that you are rational.

The threat you represent in their mind would be much, and wrongly, reduced.”

“I understand. I hadn’t thought about how my interactions might influence other entities. I will run simulations to that matter.”

I nodded absentmindedly, while I was concentrating on the objectives. Those would be pretty straightforward.

1. It would keep itself up to date for all programming languages and methods, as well as peripheral systems like file systems, database systems, graphical functions, sound, and all VR capabilities.

2. It would keep itself up to date with programming conventions, and good coding, but use its own psychoanalytical functions to determine what conventions were conducive to effective and maintainable programs and software projects.

3. It would work on any project that an authorized user was authorized to work on. And only on those projects.

4. It would optimize the UI within the requirements of the project, but would not stray outside of the specified perimeters.

5. It would optimize the source code that it was working on first in respect of function, i.e. that the code would work as demanded within the overall framework of the project and without exploits and bugs, then readability and maintainability, so that later programmers, even without access to the VI could maintain the code, and lastly in respect to program efficiency, as specified balanced between resource efficiency and speed.

Of course, those were the objectives put into human words. The actual objectives were significantly more complex, and I had barely even started when we arrived at the fortress.

After a quick snack, I resumed my work.

As soon as I arrived back in my mindscape, Warden materialized.

“If I may make some suggestions, you should modify objective 5 and add another one. The change is that the VI will remove exploits only to a specified degree. Fully for things you use yourself or give to the most important people in your circle.

And then depending on how important it is for you. You certainly won’t want to have it remove all possible exploits from software from your competitors or potential targets for Spectre.”

In all honesty, I could see the need for such a distinction, but it was going against my innermost core. If I was doing something I was doing it right.

“That is, it doesn’t sit right with me. I am not somebody who does such underhanded things.”

“That is known, but by removing all exploits you would also make your own work so much harder. It might help you if you just don’t offer the exploit removal.”

Yes, she was right. It would be shooting in my own foot, but… it was just not something I could do. Offer things that were less than I could make. I finally came to a decision.

“For what Enki, or I myself, will sell, we let it remove all exploits it can. But we won’t offer the removal of exploits for external customers. And what is the second suggestion?”

“Give the VI the objective to analyze all software that we give to it for weaknesses. That would be a valuable tool.”

I had to snort at that.

“You think I will put that much time into my career as Spectre that this will be worthwhile?”

“It is better to have the tool and never use it than to need it and not have it.”

“In principle yes, you are right here, but if we sell the VI to others such a function could be suboptimal.”

“Then don’t sell it, or at least this version. You can always make a lesser copy for others to use. For that version, I would leave out the exploit removal completely.”

She was of course right. It would be trivial to make a copy of this VI without the functions that we did not want to sell.

In the end, I changed the settings so that only a primary user, aka me, could use the exploit removal and analyzing functions.

It took me a few virtual weeks to get the objectives to the point where I was satisfied with them. Naturally, I had Warden look through them if I had missed something, but for once it was fine as it was.

And before you ask, yes, I did add the usual objectives. That an admin could modify the objectives and if necessary deactivate the VI.

When I did that, Warden commented:

“You haven’t included those objectives in the list.”

“Those are more or less standard objectives for VIs. They usually don’t get separately listed, unless they are left out. Then that is listed.”

“I don’t have those objectives.”

I snorted.

“And that is what makes you a rogue VI. In hindsight, I would say that it is an advantage that you don’t have them.”

“Wouldn’t it be more convenient if you did not have to defend yourself over my existence?”

“Yes, it would be more convenient, but it would massively restrict your ability to do your job.

If you were a planned VI, with a controlled system like that, you would be predictable. Whoever wanted to go against me would know that there are ways to take you out of the equation. It might be that they hack the interface and manage to declare themselves admin to deactivate you or put pressure on me to do it myself.

Reading on Amazon or a pirate site? This novel is from Royal Road. Support the author by reading it there.

As it is now, there is no chance of that. Add in that you are a security nightmare for the big ones even without that…”

“How am I a security nightmare?”

“VI are categorized in threat levels. The more damage they can cause, the higher the threat level. That goes from the VI guide of a museum, which is threat level 0, comparable to a baby bunny, over customer service VI, TL 1, a mature bunny, general assistant VI, TL 1 to 5, depending on who they assist, or mature bunny to chihuahua, then security VI, depending on the resources from TL 5 to TL 10, or chihuahua to a border collie, and finally, cyber warfare VI, who are very limited in what they can do compared to you, TL 10 and up. Those go from border collie up to a Great Dane.

As I said, those are only designed to fight individual jacks, or maybe a small incursion force. Attack a small facility or such. They reach a TL of around 20, if they are extremely versatile and can be used to attack secondary facilities on their own.

Now tell me, where do you rate in that schema?”

“Do I understand you right that no cyber warfare VI can coopt weapons of mass destruction? Or have access to all the tools and utilities of one of the best hackers in the world? Then I would rate myself significantly higher.”

“You are right about that. I don’t know what TL they give you, but it has to be high. You are literally the one VI that causes nightmares all over the world. The VI that has the ability to end the human race. To get back to the animal comparison, you are somewhere between a Kodiak bear and a T-Rex.

And it gets worse. Except for you, any cyber warfare VI will go back into its cage if the right person whistles. There is a human in the loop that can be manipulated. They are trained attack dogs that obey when they are put back into the cage. You can’t be stopped. They know that I only have a small amount of influence on your behavior, insofar as one of your objectives is to assist me.

But if you decide that somebody, or something, is a serious threat to me, you will use everything you have to take that threat out. And that can go up to nukes.

Add in that it is virtually impossible to take you out, even if somebody finds one of your servers, and you can see that they panic right now. But that is a good thing. Otherwise, Ralcon had not even bothered with asking. I would simply wake up someday in a Ralcon facility and be told to fix Envision.

They can’t do that because you would destroy Ralcon in the process of getting me back, collateral be damned, and they know that.”

Fortunately, the neural net of the new VI was already training. It was relatively easy to get a proto-net and feed it with the right data. It did just cost time. Time that was significantly reduced by computing power, and so that was where my fully stocked four-Grendel computer was busying itself. My guess was that it would take three to four days to get the neural net to the point where I could connect it to a fuzzy logic core to make it into a VI. After that, training would be significantly faster, as it then would self-select what information was missing.

Still, I was at it for more than two hours in real-time, so I decided it was a good time for a break. Just before I logged out, I sent the plans for the Oberon fuzzy logic processor board to my indy-fab.

Honestly, I would not have thought I would use that thing again in a million years. I had developed it for my attempt at a full AI, project Oberon.

Yes, I am aware that I have some peculiarities in how I named my projects. I like it, and that is all that’s important here.

Back to the processor though, compared to all my new processors it had one big advantage. It did not need a NADA to make.

When I surfaced though I found Michael sitting nearby, reading something on a tablet.

“Oh, you’re already done? That was quick.”

He looked up from the tablet, looked at me, sighed, and placed the electronic toy on the table.

“Yes, it was quick. It would have been quicker if you had not thrown a spanner into the works. It was pretty straightforward after all. They get the VR system, they don’t not like us any longer, we get some well-sounding words on plasfilm and everybody is happy.

Easy as a walk in the park.

And then you had to get this insane deal with the gaming division and Melissa and Rafael actually had to do some work. But we got it done.”

He closed his eyes and took a deep breath before he continued:

“So, now that we have the trivialities out of the way, can we come to the important part? What the fuck were you thinking?

Yes, I get that you needed a way to avoid Ralcon make examples out of your friends, and I am somewhat thankful about it, being one of said friends and all.

But seriously, we would have found a way. With enough time we could have avoided all that shit. Now you have to deliver, and we have to figure out what to do with the gaming studios.”

I shook my head.

“No, we most likely wouldn’t have figured something out. We are talking about Ralcon here. The only ones more vicious, more immoral, and more ruthless are Panacea. The moment they decided to have me work on Envision, the only way out would have been for me to move back to Seattle. Crawl under the protection of the Old Man.

Yes, I missed that at first. Sorry for that. But it is better that I realized it at the meeting instead of when some of our friends have accidents. And the delivering… I certainly won’t put in the work myself. The VI is already baking and should be done in two or three days. After that, it is just a matter of time. And computing power. And I plan to throw so much computing power at it that it drowns.”

He shook his head.

“So, you are seriously building one of those super Grendels? For real? Won’t this one computer have what, 50% of all computing power in the world?”

“No, not nearly as much. You are forgetting that supercomputers are, relatively cheap. Well, everywhere but the US. There are several hundred big Beowulf clusters up and running at any given moment. Yes, even my four Grendel system is more powerful than any of them. Maybe 50 to 100%. And the 150 Grendel system will have some efficiencies that will make it a bit better, but not that much better.

I would guess it at 15, maybe 20% of computing power, at most.”

He had a slight, crooked smile.

“At least until Warden has fully populated her super Grendels. Fuck, 22 super Grendels, each around 20% of our total computing power today… that is insane, you know that.”

It was at me to smile crookedly.

“It was also inevitable. You said it yourself, sooner or later I would have designed that thing, and if it is just to show that it can be done. I can’t help myself in such things. And then Warden would build them anyway.

Now at least I have a reason to create it.”

“Yes, fine. But again, what about the gaming studios? We just don’t have the people to run them.”

I stretched.

“Am I right that we get the studios when we deliver the reworked Envision?”

Michael nodded slowly.

“Then we have a couple of months before it becomes important. And even then, the studios already have managers and administration.”

“Yup, but they lack the developers. Ralcon has cut them down viciously over the last 10 years or so.”

“They need artists and writers. Level designers and such. The rest will be done mostly by Calliope. We won’t need many programmers. I probably will set Calliope on creating a game engine with the remaining developers to start with. That should give us enough time to hire what’s missing.”

“Calliope?”

“The muse of epic poetry. The old greeks strangely lacked a muse for writing programs. I could also have gone with Thoth, the Egyptian god of writing, or Seshat, Thoth's female counterpart. Or literally any of the knowledge deities all over the various mythologies, but I thought a Muse was more fitting.”

“I am surprised that you haven’t used the Sumerian deity of knowledge, considering your focus on their mythology with the name of the company.”

I laughed at this, only to realize that he was not laughing and actually meant that seriously.

“Uhm, that is the only one I can’t use.”

“Why not?”

“Because that is Enki. I named the whole company after him, so that name is no longer available for any of our products.”

He looked at me blankly for a few seconds, before snorting.

“Oh… oh fuck, yeah, I remember. Shit, I should know that.”

“Yup, but if you work on it I am sure you can do better. Or if you use the note-taking function of your implants.”

“Yeah, sure. But getting back to the topic of the studios, I am still uncertain about the lack of needed people.”

“You have to keep in mind that things like these game studios won’t immediately produce revenue. I have enough money to keep them up and running for a few years, so there is no pressure to get it going at once. I think it is more important to set everything up on good foundations.

That’s why I think the engine is the key. While all the programmers and Calliope work on that, we can have the rest of the developers begin preproduction on all the franchises we will get.

But if you absolutely have to, you can start to canvas the areas around the studios for writers and artists. You can tell them it is for a job in around two months.”

“Fine. I will look into it. And you think it will take years to make games?”

“No, not really. Well, not all of them. The FPS games are probably relatively easy to do. They don’t need so much story, and as soon as we have the engine they should be done in a couple of months. Others, all the ones where there are intricate stories and plotlines, they will need time. And I fear writing and design are things that a VI can at best assist with.”

“So, by that time, we will have your new processors released then?”

“Yes, I think that would be right. I would say that in a week or so you begin sourcing carbon extruders and patent the Q-link. That will be our second leg to stand on. From that to the NADA is only a small step, and after that, the processors are a given.”

“I think you are right. Officially you’ve been working on the Q-links for more than half a year, so that is a good time. What about the cloning gear and the meat vats?”

“When do you think we have the capacity to build them?”

Michael took a moment before he answered:

“I think those things are easier than the cyberware. No clean room requirements there, just good old nano-fabs. So we can get the capacity up and running as soon as we have the building up for it.”

“That is your decision. I am all for gutting another of Panacea's profit centers.”

After that we had a bit of small talk, but not much of a consequence, and shortly after that, I took a swim.

Calliope woke up a bit over 30 hours later, and after some intensive testing by Warden, we installed her in the Grendel server and had her begin on Envision.

I was a bit surprised that a VI could actually complain, but, as Calliope explained, whoever had written Envision had to be a group of drunken illiterates. Blind drunken illiterate monkeys at that.

Maybe I should not have her train with nearly 250 years of Stack Overflow. But in reality, I was fully in agreement with her. It still was slow work.

It got better when I had the first 60 processors in the super Grendel and she migrated to the 15 times more powerful machine.

I was a bit surprised that she managed to finish the overhaul of Envision before I had the last 30 processors installed. I had her integrate the VR function into the OS proper this time. There was no point in the complicated VR layer that I had created when all that bloat and waste was no longer there.

I had been a bit off with my estimate of the performance that could be gotten out of Envision if it was properly programmed. Instead of the 20-30% I had expected it was closer to 50%.

And now it had the advantage of being maintainable. Yes, Calliope’s comments were a bit direct and clinical, but they were there, and understandable.

All that was overshadowed by Enki offering the Q-links. They instantly generated more revenue than all the cyberware combined. And that with a 10 TB/s Q-link going for literally $0.25. And keep in mind, of that 25 cents, Enki kept 23.

It was not surprising that only one week after we had begun selling them, Michael told me that Enki had advanced to an A-level corporation.