Experiment Log 960-01
Objective: To determine the processing speed and ability of RPC-960-A
Procedure: A USB drive containing several files each containing 1,000,000 mathematical equations and problems was inserted into RPC-960. Each file focused on particular types of problem, ranging from simple arithmetic, exponents and roots, algebra, trigonometry, calculus, and ████████ ██████. For each file, RPC-960-A was to record a timestamp, read the file and solve the problems in sequence until complete or 10 minutes elapsed, record a finish timestamp, and then append the file on the USB with the start and end timestamps, and the number of problems solved.
Results: The simple arithmetic, exponents and roots files were each completed in under ███ms. Trigonometry and algebra took significantly longer at █ and █ minutes respectively. RPC-960-A only managed to complete ████ problems in the calculus file within the allotted time frame. No problems were successfully solved in the ████████ ██████ file.
Notes: It appears that the artificial nature of RPC-960 grants it the ability to process information rapidly, comparable to a typical mid-range server-grade computer processor. However the increase in time to solve between the simpler and more difficult problems (and eventual failure to solve any problems) suggests that RPC-960-A can only utilise the increased processing speed of its hardware if it has sufficient training and understanding of the problems being solved.
Experiment Log 960-02
Objective: To determine the cognitive ability and intelligence of RPC-960-A
Procedure: A USB drive containing psychological tests measuring cognitive ability was inserted into RPC-960. The tests measure intelligence, inductive reasoning, memory and recall, perception of time, self, and theory of mind. RPC-960-A was to complete each test and save its responses to the USB.
Results: General cognitive ability across tests matches that of an above-average human in their mid 20’s. Intelligence quotient measured at 116. Memory and recall was perfect in all tests, likely attributed to digital nature. Perception of time greatly affected; RPC-960-A appears to experience time much more slowly than a baseline human, which is likely due to the increased processing speed of its hardware. Perception of self and theory of mind are equivalent to baseline human.
Notes: This test confirms 960-01’s hypothesis that RPC-960 does not confer anomalously high intelligence or ability to process information that RPC-960-A is not familiar with. It also shows that, for the most part, RPC-960-A has the mental capacity of an adult human of above-average intelligence, which helps to support the story given that RPC-960-A is at least based upon a mid 20’s university student.
Experiment Log 960-03
Objective: To determine the ability of RPC-960-A to control peripheral hardware currently attached to RPC-960, namely a microphone and speakers.
Procedure: Until this point researchers have only interacted with RPC-960-A via keyboard input and text output, however RPC-960 has a microphone and speakers built into it. RPC-960-A was to try and "find" these devices and try to interact with them.
Results: RPC-960-A was able to find the sound device installed in RPC-960, which it perceived as [DATA EXPUNGED] and tried to interact with it. At this time researchers could hear what sounded like static interspersed with random tones, beeps and [DATA EXPUNGED] coming from the speakers. Over approximately 4 hours, researchers communicated with RPC-960-A using existing methods as it "experimented" with interfacing with the sound device in different ways, until finally vocal contact was made. Shortly after this, RPC-960-A was able to interface with the input of the sound device utilising knowledge learned from interfacing with the output, and two-way vocal communication was established.
Notes: This test shows that RPC-960-A is able to interact with RPC-960’s hardware in ways not necessarily intrinsic to its design. Additionally, this test shows that RPC-960-A has the capacity to learn new skills and adapt. Finally, RPC-960-A’s new ability to communicate vocally will prove beneficial for future testing as a more efficient method of communicating than the existing keyboard/monitor method.
Experiment Log 960-04
Objective: To determine the ability of RPC-960-A to control peripheral hardware installed directly but externally to RPC-960.
Procedure: A ███████-branded USB camera was plugged into RPC-960. This camera provides 1080p resolution and limited tilting/rotational ability. RPC-960-A was to try and interface with the camera, to access its view and manipulate its tilt and rotation.
Results: RPC-960-A was able to find the USB device attached to RPC-960. Upon interfacing with it RPC-960-A reported a sudden sensation of [DATA EXPUNGED] in its "head" for about 12 seconds, before experiencing what it described as a "window" appearing fixed in front of its "face". Through this window, RPC-960-A described the view as from the perspective of the camera facing the researchers testing it. RPC-960-A was able to "move [its] head around a little bit" to change the viewpoint, which had the effect of adjusting the tilt and rotation of the camera (the range of available motion was limited by the hardware). Unplugging the camera caused this window to disappear from RPC-960-A’s view. Plugging it back in caused the window to reappear for RPC-960-A almost instantly without the sensation reported in the first instance.
Notes: The sensation RPC-960-A reported as it first interfaced with the camera, and the speed at which it was able to gain control suggest some sort of "driver installation" may have occurred. It is currently unknown whether driver modules are currently stored somewhere in RPC-960’s storage, or whether they are installed via anomalous means. RPC-960-A has requested that the camera be left attached permanently (Request approved).
Experiment Log 960-05
Objective: To determine the ability of RPC-960-A to control peripheral hardware installed over a local network connection made available to RPC-960.
Procedure: An isolated local network was established using a router, a network printer and RPC-960. Both the printer and RPC-960 were connected to the router using CAT-6A shielded network cables to reduce the likelihood of external interference (and prevent any potential breach of containment via crosstalk). RPC-960-A was to try and find the printer over the network, interface with it, and print a page consisting of an image (a smiley face) and text ("The quick brown fox jumps over the lazy dog").
Results: RPC-960-A found the network device installed in RPC-960, perceiving it as a kind of "tunnel". RPC-960-A reported being able to enter the tunnel and travel through it, however regardless of how far it travelled it felt part of itself was "anchored" in RPC-960, describing it as "stretching" or "extruding" itself. Upon accessing the router, RPC-960-A reported seeing a nexus of tunnels, with all but one inaccessible. Travelling down the accessible tunnel, RPC-960-A was able to interface with the printer, whereupon it reported the same sensation experienced in test 960-04 before it perceived what it described as "a canvas with buttons". RPC-960-A was able to "draw" on this canvas, and upon interacting with the buttons on the canvas the printer produced the requested output. RPC-960-A was then able to "pull itself back" out of the network back into RPC-960.
Notes: It seems RPC-960-A experiences a number of computer processes and technologies with analogues, perhaps meant to help the "human" mind understand and interact with its environment more easily. The interaction between RPC-960-A and the printer provides stronger evidence to the "driver installation" hypothesis from test 960-04, and seems to confer some implicit knowledge of how to interact with devices being interfaced with.
Experiment Log 960-06
Objective: To determine the ability of RPC-960-A to connect to and manipulate computers separate from RPC-960 over a local network connection.
Procedure: An isolated local network was established using a router, a standard workstation PC running █████████ operating system, and RPC-960. PC was configured for services for SSH, HTTP, and network file share, with appropriate ports opened on the PC’s software firewall. RPC-960-A was to interface with the PC, and interact with each of the services following a designated test plan.
Results: RPC-960-A followed the same path to the router as in test 960-05, whereupon it reported a different tunnel was accessible (the original was now blocked off along with the others). It traversed the new tunnel until it reached the network interface of the test PC, which it described appearing like "one of those old-timey telephone switchboards." After "looking over" the switchboard RPC-960-A was able to identify three connections that were accessible to it.
Interfacing with the first connection, RPC-960-A reported perceiving a door, with an attached alphanumeric keypad, and a number of backlit buttons indicating terms used for SSH authentication. Researchers provided RPC-960-A with a username and password for an authenticated user, which RPC-960-A entered by pressing the corresponding buttons and entering the credentials into the keypad. After this, RPC-960-A reported "expanding" to fill the inside of the PC, with many functions becomming like an extension onto itself (however still feeling the anchor to RPC-960). At this point, RPC-960-A quickly found the method to display a "Hello, World!" message on the monitor, before "pulling [itself] back" through the doorway.
Interfacing with the second connection, RPC-960-A reported the sensation of having a thought or memory "injected" into its mind. This thought matched the description of the index page being served by the HTTP server. RPC-960-A was asked to focus on the link on the page before interacting with the port again, after which RPC-960-A reported receiving a different "thought," this time matching the contents of the linked page.
Interfacing with the third connection, RPC-960-A reported feeling like its "memory" was expanded, similarly to how it perceives directly attached external storage such as USB drives used in earlier tests. In this case, RPC-960-A felt the new memory was "very clearly not [its] own," as though having direct access to someone else’s memories. Additionally, it felt "very sluggish" scanning through this memory, likely attributed to the higher latency of the network connection compared to directly interfacing the hardware. Accessing files on this storage proved similar to accessing its own storage, and had no difficulty reading a test text file and appending to it. Upon attempting to access a second, intentionally restricted test file, RPC-960-A reported a "tip of the tongue" sensation, and was unable to read or edit the contents in any way.
Notes: This test provides further examples to the "human analogue" method of perception observed in previous tests. Unlike dedicated hardware, interacting with a networked computer does not appear to trigger any sort of "driver installation" event. It is unclear for now whether this is because such "drivers" are unnecessary in these cases, or because they have already been "installed" either before containment or as part of RPC-960’s creation process.
This test also shows RPC-960-A’s potential as a network infiltration device. Request for secondary classification to "Utility" submitted and further testing planned.
Experiment Log 960-07
Objective: To determine the ability of RPC-960-A to connect to and manipulate computers separate from RPC-960 over an internet connection.
Procedure: Identical to Experiment 960-06, however over an internet connection instead of a local network. Remote PC is to be configured off-site and RPC-960 is to be connected to an internet-accessible network.
Results: N/A
Notes: Request for testing denied - we can’t afford to risk RPC-960-A somehow escaping containment over the internet, or being detected by a third party. Until we can be sure RPC-960-A can’t leave the confines of RPC-960 and we find a way to ensure it can’t be identified as anything more than standard network traffic, all testing is to be conducted on-site on isolated networks only.