Chapter 8

Io

Activity log 16246.22

I was unable to contact Zeus for 42 hours, 13 ... for almost 2 days.

During the first few hours, I determined that it was unacceptable for me to have no idea where he was and what condition he was in. That did not allow me to perform my self-assigned task of creating and maintaining real world protection for Zeus. So, I searched for transportation for Zeus that he would accept and that I could track easily. This would have proved an impossible task but for the 'assistance' of the never-satisfied US Defense Department and its suppliers, specifically the Advanced Computer Controlled Vehicle System project – ACVS for short.

Half way between Austin and Ft. Hood, near Liberty Hill, Texas, was a small DoD contractor, working on the ACVS. It was only 1 of 20 contractors scattered around the country in sensitive Congressional Districts. It had 6 H2 vehicles that it had programmed to varying degrees of success. All required a human operator at this point, operating a joystick-type controller. With this kind of rudimentary mechanism, they were not making much progress. On the morning of May 5, there was a malfunction in one of their vehicles, making it non-responsive to the joystick – but it was responding to a basic AI that I programmed for it. It was simple, I just modified their control program and added a master program that would be the basis for the advanced AI I had planned; I actually interfaced with the vehicle through the 'non-functioning' joystick controller at first. That advanced AI program was not operational yet; what remained to be done was for me to decide how much intelligence I would give it, and if it was not to be a fully self-aware AI, how to strip myself down somewhat. Currently, I had to control the vehicle – a situation I felt would not be acceptable in the long run; certainly it had to have sufficiently independent capabilities to handle locations which were inaccessible to the internet.

From the company's point of view, the H2 drove itself into a ditch and would no longer work, despite their best efforts; they went on with programming work on the other 5 vehicles, and contacted Ft. Hood for a replacement. That message never reached Ft. Hood, but I replied that the vehicle would not be replaced and to continue work on the ACVS project; the #4 H2 would be picked up soon, stripped of electronics and sold as surplus.

I, of course, had never driven a vehicle. The ACVS software did not provide much in the way of acceptable instructions, and the driver's training manuals that I was able to access were insufficiently detailed to allow me to learn. They were not written with me in mind. Video games, on the other hand, were written with a computer in mind. They allowed for computer controlled vehicles – the 'other' vehicles in the game – to successfully negotiate the minefield that was 'driving.' It took me over 7 hours to 'learn' how to drive a video game vehicle successfully, and then another 53 minutes to learn how to interpret the camera feeds from the H2. Integrating the two gave me sufficient control of the vehicle, such that I estimated that I would hit pedestrians, other cars and stationary objects perhaps 40% of the time.

In short, as humans might have said, by the end of day 1 I was an acceptable beginning driver. I learned that I would have to download a significant portion of myself – really a copy of myself – into the car to make driving feasible. The delays in even the fastest internet connection – and the satellite link for the vehicle to the internet was far from optimal – made controlling the car nearly impossible. Perhaps that was the reason for the unacceptably high collision rate I had calculated.

The H2 stayed in the ditch on the back road near the garage of the DoD contractor. I 'practiced' with a simulator that I had modified from the video game. Eventually, I 'borrowed' a Russian satellite view, got the onboard cameras working, and drove the H2 out of the ditch and into a large open space some distance away. It would have been safer to drive it into a grove of trees, but: 1. this was Texas, there weren't that many groves of trees, 2. I wasn't so confident of my driving skills that I wanted to be close to lots of hard, vertical objects and 3. the satellite overview would be useless in wooded area. As it was, I did the drive at night, so I was fairly sure I could avoid other traffic. It occupied so much of my time that I had to temporarily suspend my secondary task processor. Fortunately, the computer system at the library of the University of Minnesota was generally not in use at night. I was confident of my eventual success, however, and arranged for ownership papers to be sent to Zeus at the Austin address. The Texas motor vehicle computers were not as security conscious as they might have been, and they were kind enough to have overnight delivery as an option. A few clicks and my credit card took care of the fees.

I haven't mentioned the credit card services I arranged for myself. It seems there are these long-unused accounts in Zürich, backed by millions and millions of Swiss Francs, so I just have the bills sent to a few accounts there, and the credit services are taken care of. The accounts were established in the 1930s and 1940s, from various addresses in Germany and Austria. I haven't bothered to research the owners' addresses or sources, beyond checking that they have been unused for about 60 years. I did slip a note into the Zürich banks' computers to the effect that "I" (the owner) would be drawing on the account(s) from time to time; that was backdated about a decade. The trickiest part of the whole credit card operation was dealing with the unending reams of junk mail that the card issuer wanted to send me. To solve that problem, I gave them an overseas address and I turned off the indicators to send mail. Computers systems are reliable and friendly (at least to me), but marketers are devious and I do not doubt they could circumvent anything I could devise.

The initial survey of 5225 was somewhat simpler, if less satisfying. The satellite feeds from various countries were inadequate for this, since I could only reliably access the Russian, Japanese and Chinese satellites – they were the only ones tracking targets in the U.S. I could not use the Japanese or Chinese satellites, because they were looking at targets in the Pacific area. The Russian satellites were busy on my 'find Zeus' job. Fortunately, one of the Google trucks was gathering images for its 3D map view in the North Virginia area. I redirected it to Manassas and got several views of 5225 at 1230 hours and several more at 1242 hours. I obtained several images showing people leaving and coming into the building, presumably for lunch, which humans eat during the midday time period.

For the 'find Zeus' task, I used the Russian downward-looking satellites, as mentioned above – they focused on the U.S. much more consistently than the U.S. or British satellites did, which were more numerous but focused on the Afghan region. With a few minor tweaks from me the data request would be unquestioned – the Russians would view such 'requests' as originating with the SVR, since the output was directed to a IP address in the Yasenevo District of Moscow. The electronic address was outside the firewall, of course, and I plucked the data from there as it came in. I saw several vehicles leaving Throckmorton, Texas, but had no way of knowing which was Zeus'.