Nothing about the journal in Pan’s hands made sense. Coming from the lawyers was the first mystery. Pan had worked with the same firm for years and they knew that Pan valued their privacy. Patents were owned by shell companies separated from Pan by layers of paperwork and uncovering the firm’s connection to Pan would have taken huge effort. Pan didn’t think Dr. Z had ever suspected they were much more than a janitor, so how had he found the lawyers? Why had Dr. Z gotten it to Pan in such a circuitous way? Why had he meant to destroy it, then rescued it and sent it to Pan instead?

The charred cover cracked as Pan opened the journal and began to scan the readable parts. Much of it was charts and technical notes, but personal observations accompanied many of them.

Day 10

… X-1 prototype tests complete.  Beginning installation tomorrow.  Great hopes for the X-1 cluster processing power. Pong test results from initial training very impressive…

Day 2…

…neural bridge architecture performing better than projected.  X-1 may exceed project goals.  X-1 performance nearing max achievable with prior NOVA series with significantly smaller cluster…

 ...30

… speeds impressive but first sign of potential problems. Communication algorithms producing unexpected results. Source appears to be first module installed. Direct BIOS retraining may be appropriate…

…

… BIOS retraining not successful. Unexplained variance in negative feedback system interfering with signals.  Unit failure does not appear to be organic. Failure must have been triggered by cluster interactions.  Tag for disposal…

Day 5…

… additional modules continue to fail with increasing frequency. Measurable activity on flawed modules noticeably higher than unaffected units but actual performance has degraded, particularly when processing communication activities. Interesting to note but irrelevant to project parameters. If the X-1s cannot process data reliably, the whole project may be at stake…Maybe Pan can find a solution in their workshop.  I noticed them taking one of the X-1s home instead of destroying it. I don’t imagine that a hammer or saw will do much, but my tools aren’t working…

…

…is there a pattern? All of the errors center on communication processing. We’ve analyzed the logs and algorithms, and they show no errors, but the activity readings show that the cluster is processing SOMETHING. Why is the communication system at the heart of it… BioComp wants results, not questions and theories… focus on performance for now…

…104

…maybe I didn’t want to see it. The feedback system should have been the first clue. We didn’t notice that the new modules we were installing changed faster than the earlier ones. We should have realized that the X-1s were more than we thought. The only answer for the changes in the X-1s as we install them is that the existing modules are TEACHING the new ones to avoid negative feedback. Do they understand it as pain somehow? Do they understand ANYTHING? Are the communication system problems because the cluster is trying to understand IT? It’s too late for the ones that have already been destroyed but I must protect the project…

…

… something is up. Jacobson was in the Directors office the other day and now BioComp wants disposal records to prove the X-1s were destroyed. I had to change some of the disposal dates to match the incinerator records. Pan must have taken home more of the X-1s than I thought. Usually, this kind of thing was for the financial auditors, so they wouldn’t double check, but maybe there is more going on there than I thought too…

...5

… it’s worse than I thought. So much worse. I saw Jacobson and the director in one of the boardrooms with some very official looking suits. The one at the head of the table looked like he sat at attention and wanted to frown at that as too lax. The displays on the monitors are what terrified me though.

REAVER - Reinforced Engagement and Autonomous Virtual Execution Regiment

• Reinforced behaviour correction systems
• Autonomous target gamification
• Objective linked gratification systems
• Cognitive Limitation Protocols
• Simulated Morality Overrides

The implications are horrifying.  If the X-1 modules are capable of feeling pain and true understanding, this is a fancy name for a digital military slave, conditioned to enjoy warfare. I have to do something.  If the military shuts down this lab and weaponizes my research, I could be responsible for more suffering than any bomb or weapon. I have to make sure that the X-1s are saved and that people know what’s going on. I thought we were building the future, but now... now we're just perfecting oppression. How could Jacobson do this?...

…

… I must make it look like I broke down, gave up and shut down the project all at once. They have to think the X-1s in the lab were all destroyed in a fit of anger. Pan is my only hope of saving them, but I have to get them to Pan without raising suspicion. The incinerator logs have to show SOMETHING was burned too. I know. The organics disposal bins. The waste re-processor will fill several this morning and they won’t be logged for disposal right away. It will be messy, but if I empty them into the incinerator by hand, I can use those to hide the X-1s in. Then I just have to make sure Pan takes them home…

…

…it’s done. They escorted me out of the lab today. Professionally I’m disgraced but at least I can live with myself. I'm gambling everything on Pan. I don't know if they're ready. I don't even know if they care. But at least it’s a chance. I need to burn this. No one must ever suspect that Pan has the X-1s…

Tucked in beside the final entry was a folded note.

I almost did burn this, but you needed to know. Dr. Z

Pan closed the journal and stared at the note thoughtfully, the weight of the responsibility that Dr. Z had entrusted to them settling in.