User:Ultra64Detective2401/The Search For More Anomalies, Part 2 (Draft)

Making the sixth post in my first sequence requires far more research going in than any of my previous posts, as it involves going into several reports and analyzing their reliability, while also attempting to get a sense of the capabilities and nature of the Personalization AI. Therefore, it will take longer for me to finish it. So, I figured I’d put my current draft of it here!

Note:There is a very real chance that this may end up getting split into two parts AGAIN. That would make three parts in all. We’ll just have to see.

———————————————-

Sequence:User:Ultra64Detective2401/The Basics of Anomalies

Previous:User blog:Ultra64Detective2401/The Search For More Anomalies, Part 1

Next:TBD

Last post, we tried to take a look at some of the most common reports, and found it quite hard to actually verify just what happened, making it seem like we had hit a dead end.

But, the reports aren’t all just anecdotes... Many of them are actual videos. And at the same time, we can learn some things about the AI, and use that to rule out some reports as fake.

First off, without even looking into any reports at all, we can set a maximum to certain capabilities of the AI. For one, the Nintendo 64 can only hold 4MB of RAM, and only 64MB of data can be on a cartridge. Only 256KB can be SRAM.

But, we do need to keep in mind data compression techniques.

Second, before we learn anything at all about the AI (or even the player for that matter), we already have a theoretical limit to how much data about the player it can collect.

Its only input is the controller, which consists of 9 possible D-pad combinations, 10 buttons, and a control stick with 65536 possible configurations (Okay, it’s lower than that because some can’t be achieved on an actual control stick, but nobody’s been able to calculate how many.) Together, that leaves a maximum of 29.169925 bits of information it can gather per frame. ...And there’s 30 frames per second.

875.09775 bits per second? No wonder it’s so good at personalization! Unfortunately that doesn’t exactly let us rule out that many reports. (Ceaveat: this is the theoretical maximum for a perfect AI, with a player whose inputs are perfectly correlated with their playstyle. We’re not sure how effecient Nintendo’s AI was, and the perfect correlation isn’t happening-You can’t be holding A, B, and Z one frame, switch to C-up and L the next frame, then switch to R the third frame, and if you press A, it’s probably because there’s a pit ahead of you. If the amount of info supposedly being collected by the AI is close to this maximum in a report... the report is still pretty sus.)

Third, the AI, like everything else in the universe, must be bound by the laws of physics. Everything it does must be physically feasible for a program within a Nintendo 64/Nintendo 64 cartridge. That means apparitions coming to real life is impossible.

Given that, now it’s time to start with reports. The best place to start is the earlier reports, because after anomalies became well-known, that many more people started making fake reports. (If something’s popular enough, more people want to fake having it!)

And of course, if these very sequences become popular enough, fakers might eventually figure out EXACTLY how to make a fake video that is indistinguishable from a real, actual report! (And we’d then need better verification standards!) Yet another reason to put earlier reports first.

———————————————-

Yes. It’s nowhere near finished. Hey, maybe it’ll be added to soon!

Also, if you somehow got here and haven’t read any posts in the “Basics of Anomalies” sequence... please go read them. In order. You’re not going to understand the draft of post six if you haven’t read one through five.