Dr. Berryman: Research Findings - 2007-04-17
I'd like to thank you all for your attendance. I understand your time is precious, ladies and gentlemen, so I'll do my best to make this — oh, well? Right to it. First slide, please?
So! Consider the cognitohazard agent as traditionally employed by the Foundation to protect both its documents and its facilities. How exactly is it that they work? Next slide.
In terms of visual cognitohazards, it's not as simple as targeting the lateral geniculate nucleus, nor is it merely a matter of V2 processing that renders the agent effective. It's many things in conjunction: The brain's phonological and sematic cortexes performing simeosis, the temporal lobe deriving meaning from the image, the mind's processes of recall and memory retrieval working to interpret what the eye is seeing — it's in exploiting this complex dialogue between regions of the brain where we've historically found success. Next slide.
Though there's been good work done over the years, insofar that we have agents that can influence behavior, render viewers unconscious, or outright induce a grand mal seizure, visual cognitohazards have their limits.
By a show of hands, are any of you familiar with Sumerian cuneiform?
Excellent. Next slide, please?
Haha! Bit of a jumpscare for some of you, huh? You must be read-in to the SCP this cognitohazard protects, and I'm sure the wheels are beginning to turn. For our more stone-faced members: the cuneiform you see in the upper-right quadrant of this image, in normal circumstances, would say "remember us" in English bordered on either side by ellipses.
Yes? Ah! No, don't worry — if that were the case, it'd be an infohazard. Don't worry about intentionally picturing it in one's mind's eye; imagination is merely simulation, and an imperfect one at that. You'd need to directly visually process it in totality for it to be effective. That is our limitation.
Even in cases where text does not contribute to the presentation of the cognitohazard, one still has to account for identity, for culture. This is all to say: perception is inherently subjective, and thus there is no objective means through which to effectively convey a cognitohazard.
Until today. Next slide, please.
Everyone, please! Please! It's alright. Director? Director, please, calm down. Look around you — everyone is unharmed. Let's get off this slide.
What you just saw was a visualization from a fractal mathematical set, an image that instills in its viewers a mild sense of dizziness. I understand this in itself is not impressive, but what is notable is its versatility. This image achieves a 100% efficacy rate amongst human volunteers, vastly outperforming the previous 61% record. It has even expressed its effects to some degree in animal testing. This is the real deal.
Future memeticists need not be concerned with trivialities such as 'context' and 'culture' — mathematics is absolute. Think of it: fractals are a universal component of nature, underpinning reality itself! From the infinitesimal expression of bacterial proteins to the macroscopic ordered chaos of the universe itself, it all comes down to the complex interplay of numbers and figures; the interdimensionality of fractal systems. The language of the universe.
What you've seen today is much, much less than a mere tip of the iceberg. The fractal space is infinite, and we've barely begun exploring it — and I'm looking for computer science expert collaborators to help me explore it even faster.
Today you have felt the utility of my work. We have only scraped the surface, ladies and gentlemen. Together, let us discover its depths.
SCP-9001 production centre, Site Romero-5.
Special Containment Procedures: Once per quarter, Fractal Cognitohazardous Images (FCI) produced through SCP-9001 are to be forwarded via dedicated secure fiber-optic cable to Protected Site-7. All FCIs must be transmitted alongside research data describing the effects of each image. These packages are to be analyzed by second-generation AICⓘArtificially Intelligent ConscriptsSentient digital agents developed by the Artificial Intelligence Applications Division (AIAD). AICs operate by a loose set of Standard Principles that ensures they adhere to Foundation protocol and work in the interest of the organisation at all times. - to corroborate their stated effects and catalogue according to their intensity.
Catalogued FCIs and their inoculants, if any, are to be individually contained on isolated external hard drives in Protected Site-7's Secure Data Storage. Third-Order and Second-Order FCIs may be requisitioned by personnel with Level 3 clearance for research, testing, field use, and D-Class procurement.
First Order FCIs require Level 5 clearance with additional authorization from no fewer than three signatories, including the director of RAISA and at least two members of the O5 Council. All personnel read-in to SCP files and Top Secret documents protected by First Order FCI are to be given access to the necessary inoculants. FCIs are not to be handled by any personnel with a CRV rating of less than 9.5.
To prevent unauthorized viewing, distribution, or theft, FCIs are to be stored with quantum-proof encryption (currently AES-256) viewable only through a dedicated file-embedded module or standalone window application. Each encrypted FCI is to be uniquely keyed (currently with NIST FIPS 203, ML-KEM-1024) to be viewable only in its corresponding per-file embed or per-user window. The modules and windows employ protections against copy-pasting, screenshotting, and other capture methods. Analog printing or reproduction of FCI requires approval from the O5 Council, registration, and tracking.
Per Ethics Committee mandate, files containing First Order FCI are each to be preceded by a warning regarding their presence. Due to the internal prevalence of their colloquial terminology in the wake of Incident-9001-1, such warnings are to refer to First Order FCI in common parlance: "Berryman-Langford Kill Agents".
ATTENTION
Special Operatives and Task Force Agents assigned to the ongoing SCP-9001-A crisis are advised to continue through to the second page of this file.
Visual traversal of a dead region within SCP-9001 exhibiting no anomalous effects.
Description: SCP-9001 is the Berryman-Langford set: an algorithm that, while mathematically sound and relatively simple, is capable of generating visual cognitohazards. Computer-generated visualizations of SCP-9001 produce patterns with intricate levels of detail similar to that of the Mandelbrot set; the anomalous nature of SCP-9001 becomes apparent in that these depictions contain innumerable points/still frames that impart a variety of anomalous effects. These include:
- Subliminal [Third Order]:
- Imagery which influences neurotransmitters in order to compel behavior.
- Deleterious [Second Order]:
- Imagery which induces physical effects such as nausea, seizure, and unconsciousness.
- Fatal [First Order]:
- Imagery which triggers intracranial aneurysms leading to immediate death.
In more detailed terms, the human visual system can be expressed as the abstract equation $f(x)=y$, where $x$ is an image matrix (cone and rod cells in the eye, analogous to pixels for a computer), $f$ is a function that takes $x$ as input which expresses the brain's internal processing, and $y$ is the output of how the brain interprets the image. SCP-9001 is the result of a convex optimization searching over the field of possible images for an $x$ to produce a target neural output $y$ resulting in a desirable effect.
Dr. Langford, Project HYDRA Senior Researcher
Beyond the intricate geometric patterns found within visual depictions of typical fractal sets (including swirls, intersecting lines, and self-similar shapes), FCIs derived from the Berryman-Langford set have multiple points that form markers and superstructures visually implying the presence of biological features (such as tendrils, feathers, scales and faces) despite the mathematical purity of the generation procedure. The implications of this remain unknown, but the prevailing theory is that this can be accounted for as simple pareidolia.
While SCP-9001 was adopted with a focus on human subjects, FCIs have produced results with non-human animals. Third and Second Order FCIs show less utility due to simpler brain function, but First Order FCIs have proven efficacy in all primates and most mammals, as well as many species of arachnids.
History
SCP-9001
The following series of files relates to the
Department of Analytics Project HYDRA
and is level 3/HYDRA classified.
Unauthorized access is forbidden.
2007, April
On 2007-04-17, Dr. Ava Berryman led a presentation on the cognitohazard production aspect of Project HYDRA to a much wider range of staff than had previously been aware of it, with the intention of attracting personnel to the project with the skillset necessary for its advancement.
Amongst attendees was one Dr. Robert Langford, a computer science researcher within the Department of Analytics, who at the time was working on the detection of cognitohazards via machine learning as part of Project CIRCEⓘProject CIRCEA Department of Analytics effort to use AI techniques to detect visual cognitohazards. Project CIRCE leveraged convolutional neural networks, an architecture modeled off the human vision system. After advances in reinforcement learning and natural language processing, it was succeeded by Project MEDUSA in order to unify efforts to classify visual memetic hazards as well.-. Several days later, he proposed to meet Dr. Berryman to offer his expertise, and would be flown to Site Romero-5 to meet with her.
Dr. Langford photographed in Dr. Berryman's lab.
VIDEO TRANSCRIPT 9001-1
PARTIES PRESENT:
- Senior Researcher Ava Berryman, Site Romero-5 neuroscientist and memeticist
- Researcher Robert Langford, Department of Analytics Project CIRCE, AI researcher
FOREWORD: Initial conversation regarding the project that would become SCP-9001.
<BEGIN TRANSCRIPT>
Dr. Langford enters Dr. Berryman's office. Dr. Berryman is watering a substantial collection of flower bouquets and a leafless bonsai tree on some wooden cabinets. She turns to greet him.
Dr. Berryman: Dr. Langford, how do you do?
Dr. Langford: Um, well.
They shake hands.
Dr. Berryman: Why don't you have a seat, my dear fellow. I'm glad we could find time for this conversation. I quite enjoyed your and Dr. Reddy's presentation on the use of neocognitrons for detecting cognitohazards.
Dr. Langford: Thank you, Dr. Berryman. I've watched a few of your lectures myself, including your latest, and I'd also have to say they're quite groundbreaking. I'm honored that you liked mine and Dr. Reddy's so much — although really, to give credit where it's due, Dr. Reddy was the brains behind the work. I've only been here for a few years.
Dr. Berryman: Talent and experience are fungible to a degree. I admire Dr. Reddy, but she's… she has her own ideas about things, let's just say.
Dr. Langford does not immediately respond. Dr. Berryman hovers her fingers over the cluster of tiny white flowers pinned to her hair.
Dr. Berryman: I see these have caught your attention, Dr. Langford.
Dr. Langford: Um… yes. My apologies.
Dr. Berryman: It's only natural. Carrot flowers, or as the Yankees call them, Queen Anne's Lace. They grew on the estate where I was raised. To me, their most interesting feature is that they grow in recursive layers of flower shapes: each flower grows in a flower-shaped bunch, and the bunches themselves form a larger flower shape. That self-similarity carries a fractal structure, like my research. But ah, where's my manners? Allow me to offer you a drink first before we get into the weeds.
Dr. Berryman retrieves a bottle of scotch, two glasses, and some ice cubes from a drawer. She pours out the two glasses and reaches over to place one in Dr. Langford's hand.
Dr. Langford: I'm quite alright, Dr. Berryman.
Dr. Berryman: Please, I insist. We both work with mind-altering phenomena far worse than this. Humans think their beans so secure, ensconced in their thick skulls, when all it takes to throw it off balance is just a little drink, a little smack, perhaps just the right kind of little picture…
Dr. Langford hesitates, then takes a sip.
Dr. Berryman: No, not yet!
Dr. Langford: But you told me—
Dr. Berryman: This is quality Macallan, you need to aerate it and… oh, dear, I'm so sorry. You've not tried scotch before, have you?
Dr. Langford nods.
Dr. Berryman: Well, cultural context is everything, I suppose — as I'm sure you remember, actually. There have always been plenty of "non-anomalous" cognitohazards: war photography, political propaganda, pornography. So far, most prior work, and my colleagues', have focused on this avenue, engaging the higher functions of the brain's temporal lobe to make anomalously emotional, anomalously disgusting, anomalously meaningful cognitohazards… but it never works on everyone in the same way to the same degree.
Dr. Langford: Because non-universality of human experience is the one universal.
Dr. Berryman: Absolutely, Dr. Langford, and I'm so glad that you can recognize this stumbling block so quickly when so many of my colleagues can't… but anyway, I personally think there is another approach. There is another potential universal: everyone sees — well, at least tautologically, those who can see. Everyone can detect edges, recognize rough and smooth surfaces, outline curves. We all must share some universal features at the lowest levels of the visual cortex. The optic nerve is a direct chain into the brain that we can yank, and the closer to the eye the less tangled it is.
Dr. Berryman: We already know it's possible! In epileptics, high-frequency flashes translate directly into high-frequency electrical neuron signals directly into seizures. But, alas, it doesn't work on everyone. Images that trigger an extreme end-state in the brain, the principle, is there, if only we knew how to trigger it more generally. I've mapped the whole house, and just need to find the door — so, Dr. Langford, tell me: can you help me?
Dr. Langford: Hmmm, well, my first thought is that your issue is fundamentally something akin to data compression. The brain and its billions of neurons are such a high-dimensional object. Trying to cram the trigger for a specific brain-state into a small image would be Herculean.
Dr. Berryman: There is a way to compress the infinite into the finite.
Dr. Langford: Not as a compression algorithm, in my experience. Are you thinking of a mathematical function or a generator instead, perhaps?
Dr. Berryman: Yes, but we're looking to convey something visual. Something even denser, even more intricate.
Dr. Berryman picks up a writepad from her desk and writes an equation on it.
(1)
\begin{equation} z_{n+1}=z_n^2+c \end{equation}
Dr. Langford: This is the Mandelbrot set in the complex plane, isn't it? One of the first fractals.
Dr. Berryman: Well said, Dr. Langford. An entire universe of infinite detail, infinite complexity, encoded into a single equation of three terms. A well with no bottom, an ocean with no floor, a fall with no end. Even our physical universe has an end, a minimum, a digital pixelation: the Planck length. In some deep sense, a fractal is more real than reality.
Dr. Langford: I'll grant you that's intriguing, but humans are more visual than explicitly mathematical. Even anoraks like us don't actually calculate fractals in our heads.
Dr. Berryman: I imagine football players don't look fondly on their high school physics classes, and yet they calculate the trajectories of their balls all the same. Have you ever read the book The Little Prince?
Dr. Langford shakes his head.
Dr. Berryman: It's a nice story, I recommend it. But the relevant part is when the Little Prince asks the narrator to draw him a sheep. He draws successively less realistic sketches until finally, in exasperation, the narrator just draws a box and claims the sheep is inside. But though you can't see it directly, in your mind's eye, you can still see the sheep in the box, can't you?
Dr. Langford: Of course.
Dr. Berryman: Your brain isn't seeing though. It's imagining. It's calculating.
Dr. Langford: Decompression.
Dr. Berryman: Exactly.
Dr. Langford: But you can't do that with fractals. A sheep, sure, but an infinitely complex fractal? It's completely different!
Dr. Berryman: Are you familiar with MKULTRA?
Dr. Langford: The American CIA project where they gave a bunch of people hallucinogenic drugs?
Dr. Berryman: It was a joint effort with the Foundation Special Projects Division, run by the current O5-8. LSD, DMT, mushrooms.
Dr. Langford exhales slowly.
Dr. Langford: That… that was us? I'm really hoping we had a good reason for that.
Dr. Berryman: Of course we did. Did you think they'd declassify the useful findings?
Dr. Langford: Well, what did we find?
Dr. Berryman: Fractals. Fractals everywhere. Fractals integrated into the whole field of vision: the ground, the walls, the structures, the tunnels, the so-called "machine elf" entities.
Dr. Langford: Machine elves?
Dr. Berryman: Let's not get distracted. The important part is— look, these drugs yank the visual cortex away from physical reality, and what does it imagine, what does it calculate instead in its temporary deranged state? Fractals. Of course I had to explore deeper. My own testing confirmed those findings. The neural maps in the low-level striate cortex responsible for the most basic building blocks in vision? Detecting edges, curves, and surfaces? Those building blocks are rotationally symmetric fractals.
Dr. Langford: Stop, stop, hold on a second. You're telling me you looked at everything done in MKULTRA, and your decision was to… continue that research?
Dr. Berryman: No, no, don't think of it like that. It's standard procedure across the Foundation, after all. I've read the Project CIRCE files and Protocol 10-Kempelen. Even you use D-class to confirm cognitohazards for training data. Why do you do it?
Dr. Langford: That is for a good reason. What you're talking about is— is— you couldn't possibly have done that with Ethics Committee approval.
Dr. Berryman: But approve they did, Dr. Langford.
Dr. Langford falls silent.
Dr. Berryman: I'll tell you what I told them, and I hope you find it as convincing as they did. What I propose is not an idle, wasteful science experiment, Dr. Langford, but a revolutionary tool. A tool that could have many uses, if it were flexible enough. To protect the Foundation and its data and its personnel and yes, its anomalies, of course, but more than that. Think of a more peaceful world, one protected from the next anomalous cognitohazard via mass distribution of our own protective ones. Think of a reality bender finally going home to hug their family, tears in their eyes — unthinkable, right? — after we use a targeted cognitohazard to wipe the anomaly from their brain. Think of the people in your own life, surely at least a few friends or family, whose minds have seen better days and aren't their full selves anymore. Chemicals and pills are no way to treat the brain, harsh mallets that they are. Vision is a system directly tied to it, with far greater granularity.
Dr. Langford stares into his drink.
Dr. Langford: Of course I can think of a few such people, and of course I'd very much like to help them.
Dr. Berryman: We are this close to changing the course of human history, from the individual to society, and for the better. My prior work has taken us this far; the end is in sight, just a little bit farther away.
Dr. Langford: You've definitely convinced me of the impact, alright. But let's be realistic: this would be an exceptionally dangerous tool ripe for abuse, even with well-intentioned people like ourselves. The Foundation would keep it under lock and key, and rightly so.
Dr. Berryman: The Foundation would supervise it and ensure only its beneficial use, the same way it does and has done successfully for thousands of other items. We have the Ethics Committee to be sure, as you pointed out. Ultimately there is no containment more profound than explanation. And that is what we will do: not just for the cognitohazards we develop, but for all cognitohazards. The "anomalous" is arbitrary, just things we're too small to understand yet, but everything follows rules. I understand some of these rules myself: the rules of the brain, certainly. But I've realized I alone cannot further bridge the gap between the input and the output. I don't know how to turn the brain into a math problem to solve, where the answer is the perfect trigger fractal we're searching for. I certainly can't program. But your AI model, your solving techniques, and your evident insight, these I need to go from understanding to creation.
Dr. Berryman leans forward.
Dr. Berryman: Are you with me?
Dr. Langford nods his head.
Dr. Langford: I-I'm quite honored you esteem my abilities so.
Dr. Berryman: Excellent. I'd like you to transfer to Site Romero-5 and meet with me at least every other day. Let's immediately get to work on a project proposal to submit to O5-8.
Dr. Langford: O… 5? This is serious, then.
Dr. Berryman: I can see it in your eyes, Dr. Langford. You want to play in the big leagues, don't you? This is your chance.
Dr. Langford: Of course… let's do it.
<END TRANSCRIPT>
Shortly after meeting with Dr. Berryman, Dr. Langford voluntarily transferred from Project CIRCE to Project HYDRA, and relocated from Site-15 to Memetics Research Site Romero-5 within a week. Dr. Langford would quickly prove to be a crucial hire for Project HYDRA.
In 2007, the cognitohazard generation algorithm employed a Monte Carlo method to randomly sample the mathematical set discovered by Dr. Berryman, and then generated fractal imagery based on the resulting seed values. While it was the most reliable method known to Dr. Berryman's team at the time, it was not efficient in the slightest, and the vast majority (> 99.9%) of images depicted 'dead regions' of the fractal space with no cognitohazardous effect on observers.
Dr. Langford began by applying a Hessian-approximating optimization algorithm based on machine-learning techniques to augment this process, resulting in small efficiency gains. In doing so, he identified that the Monte Carlo process was fundamentally deficient, and suggested that the team derive the fractal set formulas that defined the space in order to solve them directly. Dr. Berryman had been unaware that this was a possibility, and so Dr. Langford began the task himself.
However, Dr. Langford complained on numerous occasions about Dr. Berryman's policy regarding potential exposure of her team to cognitohazards, and this came to a head when he put his thoughts in writing on the Site's internal communications system.
Dr. Berryman,
I've only been here a few weeks now, but in that time I've found some practices that, while apparently routine here, deeply trouble me.
This isn't coming out of nowhere. Project CIRCE was only working on detecting cultural coghaz, but even that necessitated several safety protocols, and I don't feel safe without them. It's like I've stepped into a new workshop and I'm the only one who's heard of a respirator.
I don't need to go far to find something that really shouldn't be happening at all. The team is, as standard, looking at the images they generate to see whether they're cognitohazardous or not. That's insanity. We might not have made anything dangerous yet, but we will!
I must insist that we adopt at least the most essential of those policies:
- Project HYDRA staff must never view the output of the cognitohazard generation process, even if we suspect it's just going to be a dead region.
- Instead, we must confirm and classify any potential FCIs via Protocol 10-Kempelen.
I know that you know of it already, Ava, but for the benefit of everyone else, Protocol 10-Kempelen is a process used by Project CIRCE, where you have a number of D-Class monitored by fMRI scanners, and you show them the cognitohazards and record what effects it has on them. It's a buffer layer between us and the cognitohazards, for our own protection.
Site-15 has its own 10-Kempelen facility that keeps a rotating population of 50 D-Class at all times. I recommend we verify our FCIs remotely using that resource, instead of having to build our own facility.
I respect you, and I respect the work you're doing, I truly do. But this brazen disregard of safety standards — to word it mildly! — will only cost us in the long run.
This is just going to grind Project HYDRA to a halt.
Your concerns are noted, but at this stage of the work, speedy testing and iterative development are more important. O5-8 wants this work done yesterday, after all!
Holy shit. Yes, we need to implement these policies. Thank you, Langford, for bringing this up.
I'll handle O5-8. Everyone needs to stop work immediately until we have this rectified.
Berryman, you and I need to have a word.
Under Site Director Hughes' direction, the cognitohazard production process was changed so that personnel were not to view their outputs at all, but instead send them to Site-15 for review.
In the interests of investing in that process' efficiency, O5-8 subsequently ordered the unification of Project CIRCE and several related projects, to consolidate all Foundation research into cognitohazard detection. The resulting project would eventually be named Project MEDUSA, and its goal would be to provide a single source of automated, scalable cognitohazard detection.
Additionally, O5-8 ordered Dr. Berryman to experiment with reproducing Procedure 10-Kempelen locally at Site Romero-5, to reduce the project's dependency on Site-15 and on D-Class. Dr. Berryman was loaned several researchers from the main project for this purpose. The duration of this experiment and its outcome, if any, are not on record.
Dr. Langford, who was now perceived as having single-handedly stifled the progress of the project, penned an open letter to all of its personnel.
Hi everyone,
Over the past few days I've had to put up with more than a few sideways glances here and there. You must all think I'm the new guy who's come here to throw his weight around and disrupt your routine just for the sake of it.
And, I'll say it plainly, I don't blame you at all. I get it. The Foundation and O5-8 are pushing us to produce as much coghaz as we can. We've got pressure on all sides. If I were in your position, and someone came along and told you how you should be doing your job, I'd be pissed as well.
So now I'm going to tell you why I'm right.
For those who don't know me, I came here from Project CIRCE. I had a hand in building the very same tools that we're now using for automated coghaz detection. Believe me - that project was not plain sailing. We made mistakes, serious ones, that cost lives — and now you, we, are on the verge of doing the very same.
In 1999, my team received our first instant-kill coghaz, the very first First Order CCI. Its cultural requirements were specific, but our security was lax. Six good researchers lost their lives on the spot. Their potential wasted, completely, instantly. I was very nearly one of them.
Our old Monte Carlo process was unpredictable, you had zero control over its output, but my new optimization algorithm is like a targeting rangefinder for our fractal cognitohazard generation. That means it's now only a question of when, and not if, before we create an FCI that does something irreversible — in fact, it MUST happen for Project HYDRA to succeed. And I know that the vast, vast majority of what we were producing through Monte Carlo were just dead regions, and that's still mostly the case now, and there's a good 95% chance nothing happens when you look - but you roll those dice every single time you do. And, given time, we'll get more efficient, and those odds will only climb — sharply.
We cannot risk viewing the images we create. Ever. If you value your life, you must never set eyes upon your creations. That is why Procedure 10-Kempelen was established.
The better of you will have looked into 10-Kempelen and will know that it is staffed by D-Class. The best of you will ask, why should they suffer in our stead?
They shouldn't. It's not fair. We need to fix that, too.
That is why O5-8 has ordered the Department of Analytics to refine the computerized detection processes. That team is doing exactly what we're doing, just in reverse. They are not our competition — we are working together. For all of our sakes, we must allow their progress to outpace ours. Yes, I know: the automated detection systems are at best about 30% accurate. That will change, given time.
Dr. Berryman is working on a solution too, in her own way, with her own handpicked team. I'm not privy to the details, but with their efforts combined, we'll be back to full speed and more before you know it.
And I'm hard at work as well, when I'm not busy trying to get everyone to down tools! If I can derive the underlying formulas that define the cognitohazardous fractal space, we'll be able to traverse it directly, and the current algorithm can go fuck itself. We'll be vastly more efficient and we'll have so much more control.
It will get better. I promise. The odds of losing are small, but we're not safe until they're zero, and the stakes are so very high.
Robert Langford
As Dr. Langford's policies had the support of Site Director Hughes, cognitohazard production remained subject to these constraints. In the meantime, efforts were redirected to the study of the production of inoculant images.
Excerpt from On the Generation of Cognitohazard Inoculants by Dr. Ava Berryman, 2006
The production of inoculant imagery is intended to confer immunity to one specific cognitohazard through its pairing with a second controlled cognitohazardous stimulus with an effect tailored specifically to render the viewer immune to the aforementioned. The mechanism is poorly-understood; our proposed theory posits that each inoculant permanently disrupts a distinct set of cerebral nodes required for the activation of its corresponding cognitohazard. Historically, identifying a suitable inoculant for a given cognitohazard has proven more laborious than the initial discovery of the hazard itself, due to limitations in methodology and the complexity of neural interactions […]
Although Dr. Langford had not yet derived the underlying equations, improving computational technologies were beginning to accelerate the pace of development. At this time, Site Romero-5 was producing up to 4 FCIs per week, though after curation, only an average of 2 per month were considered potentially useful to the Foundation.
Thin filament structured in Lichtenberg pattern. Viewer becomes acutely aware of their breathing patterns for a short period.
Branching fractal in greyscale with seven arms. Dead region. No effects noted within the image.
Serpentine swirl in blue tones and receding into two separate points. Induces mild sense of anxiety regarding an upcoming personal event or task that the viewer has neglected.
This unit posits this may be useful for internal use in increasing productivity.
2008, July
On 2008-07-05, the 10-Kempelen facility at Site-15 reported that Site Romero-5 had produced a lethal First Order FCI with ID number FCI-284/1.
Procedure 10-Kempelen attempts to assess the efficacy of the cognitohazard through exposure to multiple test subjects. 15 D-Class personnel were made to observe the FCI, all of whom were found to be susceptible before Site-15 staff terminated the test prematurely. For reference, the efficacy of the average CCI is 45% and of the at-the-time most effective CCI was 71%. The researcher responsible for the production of the FCI chose to remain anonymous.
The First Order FCI and its inoculants were submitted to Site-19, and on 2008-07-26, Lieutenant Masipag actioned the first-ever deployment of a universally-cognitohazardous kill agent to secure the SCP-001 file.
Dr. Berryman issued a statement to the Project HYDRA staff in response.
Colleagues,
This month, we have taken our largest step yet towards proving the legitimacy of our endeavor, and have finally made a truly significant contribution to the security of the Foundation. Our work has finally borne fruit. I am so very proud of all of you.
For many of you, this is your first Project of this scale. Hold onto this feeling — do not let your pride fade. An accolade of this esteem can catapult your career, if you play it right.
Tonight we're going to have a get-together in the Site cafeteria. Food and drink will be provided, and I expect you to not only leave your work at the door but to forget about it for as long as you need to in the morning as well, if you catch my drift.
I'd really, really like to see everyone come to this celebration, no matter whether you're working in Generation or if you're on the Berry team. Let's come together as a Site and give each other the break we deserve!
I'm grateful that our first lethal cognitohazard has found a role in the protection of our most important assets. It would have been a shame for that prestige to have been given to a more aggressive function.
Each day this month, I've thought of the fifteen personnel who gave their lives for the sake of this picture, and what they would think if they saw us now. We have much to thank them for. Please afford them a moment's thought yourself, even though you have never met, and are so many layers removed from them.
We are on a precipice, certainly. It is my deepest hope that it will be of something great, and that collectively we can be proud of the decisions we will have made along the way.
I'll make sure to swing by the celebrations briefly, but the stresses of the project have been getting to me, so I won't be staying long. It'll be nice to see everyone in a more relaxed capacity for once!
Despite the success, Project HYDRA was not yet complete, as Site Romero-5 could not yet continuously or reliably produce any kind of FCI, let alone First Order FCI specifically. All FCIs produced thus far had been the result of a mix of the Langford algorithm and guesswork.
2010, March
In March 2010, after just over two years of effort, Dr. Langford successfully derived the underlying formulas. The result of his research, a single equation that definitionally mapped out the entirety of the cognitohazardous fractal space, would become known as the Berryman-Langford set.
The definition of the Berryman-Langford set, after its correctness was proven, was immediately locked in an ultra-secure enclave within the Site Romero-5 data store, with access only provided to Site Director Hughes, Dr. Berryman, and Dr. Langford. Other members of the research team were not permitted to view the equations or even know its parameter count, as any leak to hostile Groups of Interest would represent a colossal security breach.
The Berryman-Langford set could be trivially solved computationally to pick valid seed values; the fractal space could now be traversed directly, intentionally picking regions to explore, instead of needing to randomly chance upon starting values that happened to intersect with the fractal space through sheer luck.
Through the Berryman-Langford set, the rate of generated output having cognitohazardous properties increased by orders of magnitude, from 0.004% to over 80%. Dr. Berryman suspected that the rate was actually closer to 100%, with the discrepancy caused by cognitohazardous effects that the detection methods at the 10-Kempelen facility were unable to classify, e.g. immeasurably minuscule effects, or effects targeting non-quantifiable parts of the mind.
However, Site-15 would not address Dr. Berryman's requests for improvement. Within a month, an official complaint was issued to Site Romero-5.
Samantha,
Your team has flooded us with cognitohazard classification requests. Our testing facility is overwhelmed. Both our organic and computational resources are being consumed mostly by you.
You are not the only Site that relies on us. Nearly all cognitohazard classification goes through Site-15. Our backlog is becoming unmanageable. We are hugely investing in our computational detection processes through Project MEDUSA but I do not have the resources I need to scale up our Procedure 10-Kempelen operations to the levels that you are asking for. There are not enough D-Class.
I have enough issues to deal with as it is. The Site Director has authorized me to block all classification requests from your team if I need to. I understand the importance of your work and I have no desire to do that.
But, please, slow down.
Site Director Hughes elected to sustain the request from Site-15. After negotiating with Director Vuković, she placed a limit on the number of potential FCIs that could be submitted to Site-15 per day.
With Project HYDRA's progress now externally bottlenecked, the headcount at Site Romero-5 was reduced from 55 to 19. Many of the displaced staff transferred to Site-15 to assist efforts to expand detection capacity at Project MEDUSA. Permission was granted to Dr. Berryman for access to any Foundation resource as required to complete her research project.
2011, June
In June 2011, the Department of Analytics released Project MEDUSA for general access to the rest of the Foundation.
Automated detection of cognitohazards via the MEDUSA service breathed new life into Project HYDRA. The MEDUSA service was capable of taking over all rote cognitohazard detection and classification work, and outclassed the previously-available cognitohazard detection methods in practically every metric. For cognitohazards with known effects, the cognitohazard backlog was emptied within a matter of weeks, and production quickly resumed at full pace.
As the MEDUSA service was a machine-learning based system, constant exposure to new fractal cognitohazards created a positive feedback loop, making it more and more capable over time. However, for the same reason, it was unable to classify cognitohazards with as-yet-unseen effects. Around one third of FCI production fell into this category. The only way to assess such a cognitohazard with sufficient medical rigor was, and is, to measure its effects on a living person. The 10-Kempelen facility at Site-15 remained the bottleneck.
2012, January
In January 2012, Dr. Berryman's research project was complete and her team merged back into the main team. Cognitohazard generation stopped while the results of the experiment were integrated into the existing FCI creation process.
Three days later, production resumed along with verification requests via the MEDUSA service, albeit now with zero demand on the 10-Kempelen facility.
In March, Project MEDUSA Director Vuković sent a message to Site Director Hughes.
Samantha,
How has your team managed to not only reduce its reliance on our 10-Kempelen operation, but eliminate it entirely?
I've seen some of the cognitohazards your team is producing since that change (not the imagery itself, of course), and the attached data is just as medically conclusive as ours. Yet your site is not requisitioning any D-Class. If others could reproduce your results, our load could be lessened even further, or perhaps we could use them ourselves.
Please interpret my asking as nothing more than grateful curiosity. The resources that can now be diverted from the operation will have cumulative benefits everywhere. The Department of Analytics director has advised me not to be questioning a good thing in the first place.
Vladimir,
Thank you for your grateful curiosity. In the pursuit of efficiency, we're very happy to have been able to reduce our impact on your personnel.
I am not at liberty to say more.
Dr. Langford did not comment on the process improvements initially, but a week later, he announced his full support.
With the MEDUSA service being able to validate any amount of cognitohazards as needed, Site Romero-5 being able to classify all FCIs itself to a satisfactory level, and the solved Berryman-Langford set allowing for unabated FCI generation, Project HYDRA was officially complete. O5-8 briefly visited to congratulate the team.
Most Site Romero-5 staff, including Director Hughes, Dr. Berryman, and Dr. Langford, remained on-site to oversee continued cognitohazard production.
As of the end of 2012, after curation, Site Romero-5 was submitting an average of 60 useful FCIs per week, up to 10 of which were First Order.
Innumerable wisps of pale orange strands which converge into a complex, many-pointed star. Viewer may experience quickened heart rate for a brief duration, coupled with an implaceable feeling of growing dread.
Greyscale snailshell spiral receding into a single point in the center-left. Minute effect on spatial reasoning: sets of right angles enclosed within a space composed of right angles (e.g. a closed caption bar displayed on a television) appear skewed.
Many colorful whorls bordering the edge of a dark region, from which emerges many strands of Lichtenberg filament. Viewer becomes skeptical of recently-learned information.
This unit has pored over accompanying research data and has come to the conclusion that the effect is too minute to be effective in the field.
SCP-9001
The following series of files relates to the
Department of Analytics Project NIETZSCHE
and is level 4/NIETZSCHE classified.
Unauthorized access is strictly forbidden.
In November 2014, O5-8 visited Site Romero-5.
Project NIETZSCHE briefing
In attendance: Site Director Samantha Hughes, Dr. Ava Berryman, Dr. Robert Langford, and O5-8 as Project NIETZSCHE director.
Meeting was conducted in the secure data vault below Site Romero-5.
O5-8: This room is private, yes?
Dir. Hughes: Private enough. There aren't any recording devices. Technically the room could be sealed off, but we'd only have half an hour of air, so I'm not going to do that.
O5-8: Seal the room.
Loud metallic noises and a rush of air as Director Hughes seals the security door closed.
Dir. Hughes: Oh, wonderful. You realize that can only be opened from outside?
O5-8: Your staff will have been notified already. Our conversation will not leave this room. Understood?
Dir. Hughes: Understood. Eight, what could possibly need to be this secret?
O5-8: You are aware that the cognitohazards you have been producing are primarily in use to secure critical Foundation assets.
Dir. Hughes: Yes.
O5-8: There is a flaw in that security. The cognitohazards do not affect one demographic.
Dir. Hughes: I thought the whole point was that they work on everyone.
Dr. Berryman: They do. Therein lies the beauty of the fractal. Its purity touches all.
O5-8: No. They do not work on reality benders.
Dr. Berryman: Which ones?
O5-8: All.
Dr. Berryman: You may have misunderstood. What kinds of reality benders, specifically, are unaffected?
O5-8: All.
Dr. Berryman: I… hm. I see.
Dr. Langford: That's not possible.
Dir. Hughes: Robert, you tested on reality benders, right?
Dr. Langford: Well, not specifically, but there would have been some in the 10-Kempelen—
Dr. Berryman: To be quite honest, I've been expecting something like this. I believe we would have heard about some high-profile reality bender death by now.
Dr. Langford: There's no way you could have been expecting anything like this. We have no reason to suspect someone would be immune just because they're a reality bender.
Dr. Berryman: Perhaps, perhaps not. The fractal is more real than reality, remember? Does it not fit that a reality bender would be subject to different rules?
Dr. Langford: By virtue of what? You are the biologist, aren't you? What biological difference is there between someone who's a reality bender and—
O5-8: Stop. Berryman, what is your theory?
Dr. Berryman: If it is only reality benders who are not affected, as you say, and it is all reality benders who are not affected, as you imply, then it follows that reality benders are simply fundamentally different to everyone else.
Dr. Langford: That can't be true! Some of the D-Class in 10-Kempelen must have been reality benders, statistically-speaking! I've reviewed all the data and not once has any FCI come back as anything other than 100% effective.
Dr. Berryman: Evidently not, elsewise we'd not be having this conversation, would we?
O5-8: Dr. Langford is correct. Latent reality benders in the D-Class population of Site-15 were exposed to your cognitohazards on numerous occasions. They were unaffected. We have gone to great lengths to hide those results.
Dr. Langford: Wh… what? Why?
O5-8: Utilization of the imperfect is more efficient than awaiting perfection.
Dir. Hughes: You've known about this for two years, at least, then? How has this not been a problem until now?
O5-8: Yours is not the only security method under our employ. Should a reality bender happen to access a protected document, conventional trackers have proven sufficient for locating and acquiring them. It does not happen often. Most who view a cognitohazard are inoculated against it.
Dir. Hughes: I thought we were making actual contributions to Foundation security, not slapdash patches to a sloppy approximation—
O5-8: Silence. We are now using your cognitohazards to secure elements for which fallback methods may not be sufficient. This is not yet a matter of urgency. What do you need?
Dr. Langford: To produce a cognitohazard that can affect reality benders? Well, there must be a biological reason for the discrepancy, and not something that's ever been noted before, in any study, ever, otherwise we'd already know about it. It must be something that's specific to fractal coghaz. So, we need data, lots of data, and we need to get it by exposing fractal coghaz to as many reality benders as we can.
Dr. Berryman: We don't need anything of the sort. It's simple: the reality bender enforces her own perception of reality. Of course this would render her immune to the greater reality of the fractal.
Dr. Langford: That's just a hypothesis and it doesn't even make sense. We need evidence.
O5-8: Yes. We are not interested in speculation. Prove or disprove.
Dr. Berryman: Very well.
Dr. Langford: We're going to need that data.
O5-8: We will give you access to the reality bender data from the Site-15 tests.
Dr. Langford: It won't be enough. We'll need more, much more.
Dr. Berryman: I believe I have just the vaguest notions of an idea as to how we might gather that data from the reality benders already employed at the Foundation.
O5-8: Good, Dr. Berryman. We look forward to hearing your idea.
Upon returning to Site-01, O5-8 initiated Project NIETZSCHE, enrolling Dr. Berryman, Dr. Langford, and the rest of the remaining Site Romero-5 staff. O5-8 would direct the project, with Dr. Berryman reporting directly, and Dr. Langford reporting to Dr. Berryman.
Project NIETZSCHE was presented to the Foundation as simply the continued effort to improve the cognitohazard production process, with knowledge of its true purpose — to discover a means to use fractal cognitohazards on reality benders — kept secret, on account of the security implications. To that effect, Site Director Hughes would remain chiefly concerned with the day-to-day operations of Site Romero-5.
2014, December
| Proposal NIETZSCHE-01 |
| Date |
2014-12-11 |
| Proposed by |
Dr. Ava Berryman |
| PROBLEM |
| Project NIETZSCHE requires a continuous source of data regarding the responses of reality benders to fractal cognitohazards, recorded to a level of medical rigor. |
| RECOMMENDED APPROACH |
| The Foundation has access to a large number of reality benders within its body of staff. We must convince or coerce them to voluntarily record cognitohazard response data. This will be far simpler than attempting to acquire reality benders from outside the Foundation or through attempting to use data collected from humanoid reality-bending SCPs, as reality benders on the Foundation payroll are already under our control. |
| PROPOSED STRATEGY |
The Foundation currently lacks a comprehensive method of objectively testing a reality bender's potential, and there is great demand for an assessment to fill this void. I will create such an assessment, which will be presented as a means for Site leadership to examine a reality bender's abilities in terms of veracity and usefulness.
The assessment will actually exclusively contain FCIs that we intend to test on reality benders. The assessment will guide an interviewer and a participant through discussion of the cognitohazardous effects under fMRI measurement. We will use the data to inform future editions of the assessment, and in so doing we will create a feedback loop that continuously empowers our development process.
I do not have the means to legitimately create a functional reality bending test, but any errors in its results can be attributed to its being merely an early prototype. We may be able to leverage this by encouraging Site leadership to get in on the testing procedure early and be ahead of the curve.
The assessment will be offered to various Sites with high numbers of reality benders on their staff. I will imply that participation in the assessment is, or will be, mandatory. |
| AMENDMENTS |
O5-8: It will reflect poorly on us should the truth come out that the assessment is ineffective. Site leadership must be informed of its true purpose. They will understand the importance. They will respect the secrecy. |
Update: 2015-01-06
I have compiled a set of 38 cognitohazards into a publication, the Reality Bending Aptitude Assessment (RBAA), and it is currently in print at the Foundation Press.
Cover of 'Reality Bending Aptitude Assessment', hardback, Dr. Berryman.
Update: 2015-01-17
Copies of the RBAA have been distributed to several Foundation sites: Site-87, Site-91, Site-118, Site-400. Each one of those Sites has responded similarly: they thanked me for taking an interest, said they would review, and promised to get back to me. As of yet, none of them have.
Update, 2015-02-11
My initial batch of requests remains unanswered. I have decided to reach out to Site-120ⓘSite-120Foundation headquarters in Poland. Containment and research facility that embraces usage of thaumaturgy and the anomalous, including within its staff.- earlier than planned. Site-120 is home to the Department of Ontokinetics and the single largest employer of reality benders in the Foundation. If they agree to my request, I believe it would be sufficient to start a snowball for the RBAA to get a foothold and spread across the Foundation.
Update, 2015-02-14
Site-120 has replied:
Dr. Berryman,
I appreciate you making this generous offer to Site-120, but more than that, I appreciate you revealing the truth of what this offer actually entails. Naturally, I will keep your secret. In fact, as far as Site-120 is concerned, your secret ends with me. I will not participate.
Dr. Berryman, the Reality Bending Aptitude Assessment is transparently bullshit. There is no use pretending. There is no one who does not see through it — Site Director, interviewer, participant. If you had reached out initially and asked for a means of data collection, you would very likely have received a more positive response. Instead, you have hidden your intent under layers of subterfuge.
However, even if I were willing to overlook that, this assessment is still not something I would be interested in my staff partaking in. You have gone to great lengths to conceal the nature of Project NIETZSCHE, and no doubt any information most personnel would be able to find would be carefully-constructed fabrications, but you have already revealed your hand to me. Your assessment may well be perfectly safe now, but your cognitohazards will develop, and eventually the assessment will not be safe at all. Eventually you will need to test a kill agent in the field. This is always the end goal of these kinds of projects, whether you choose to acknowledge it or not.
My staff are extraordinarily valuable to me, to Site-120, and to the Foundation as a whole. Why, then, should I agree to allow my staff to be guinea pigs for your experiments? Why should I trust you and your assessment, and grow complacent myself? Do you expect me to look the other way when you inevitably use a future edition of the assessment to turn on me and my staff simply because I once stepped aside for you?
You are unlikely to make any progress if you continue using these methods.
Dr. Jessie Rivera
Human Resources and Personnel Satisfaction Lead
Site-120
Evidence would suggest that attempting to convince further Sites to integrate the RBAA into their methodology would be an imprudent use of my time. We should still use the RBAA to assess reality benders as they appear in the D-Class population, should the opportunity present itself.
BY ORDER OF OVERWATCH COMMAND
Proposal NIETZSCHE-01 is hereby decommissioned.
Site Romero-5 continued to request reality benders for cognitohazard testing, but with the paranoia of the RBAA fresh in the minds of Foundation leadership, little attention was paid. Dr. Berryman and Dr. Langford would continue trying to find solutions using the small amount of data they already had, and production of regular FCIs continued.
2016, November
O5-8 visited Site Romero-5 and ordered Dr. Langford to provide an update on Project NIETZSCHE progress.
In the resulting schedule alignment meeting, Dr. Langford reported substantial delays and difficulties, principally that the team still did not have nearly enough data on reality benders to begin drawing any kinds of conclusions at all, and were still just as in the dark as when the project began.
O5-8 was satisfied by this answer, but not sated, and ordered Dr. Langford to repeat this presentation on an annual basis, each November, to keep interested parties up to date. The meeting would yet recur twice more.
2017, February
Following a period of several months with no updates from the Site Romero-5 team, without warning, Dr. Langford issued the following memorandum.
Hi all,
I've made a breakthrough regarding the reality bender's resistance to fractal cognitohazards.
We knew from the go (well, we'd speculated) that, as this problem is not only unsolved but apparently nonexistent outside of this narrow scope, the issue must pertain to fractal cognitohazards specifically and exclusively.
My research into reality bender neuroanatomy is beginning to indicate that this is indeed the case! I have proven that reality benders' immunity to FCIs is a direct result of their brain structure having some difference to baseline humans.
However, we don't yet have enough data about reality benders' brain structures to construct a pattern definitively. We need much more data than we have.
My suggestion would be to have all reality benders across the Foundation undergo CT scans and have the resulting data models transferred to our data store so that we can analyze and compare. Of course, this is quite invasive, and I don't have anything close to the authority to request it.
Excellent news. This is consistent with what you have been saying since Project NIETZSCHE was incepted. It is promising that your instinct has proven accurate.
Do what you must to acquire the data you need. You may cite our authority as necessary.
I wish you had consulted me before making this announcement. There is no method by which a reality bender can be identified by brain structure alone, this is known. The problem is more fundamental. The conclusion you've drawn makes little sense from a biological standpoint, and the weeks you've spent trying to convince me of 'scientific rigour' have apparently been lost to you as you've not included a single citation or scrap of medical evidence.
This isn't like you, Dr. Langford. But your childish insistence on communicating to me via open letters very much is.
We have placed full confidence in Dr. Langford. We advise you to do the same.
Ava, please accept my apologies. I keep getting wrapped up in the work and get ahead of myself in the excitement. Really, I am sorry.
In the months that followed, Site Romero-5 would issue O5-backed orders to other Sites, requesting computer tomography scans of their on-staff reality benders.
2017, November
Dr. Langford travelled to Site-15 and presented the second annual Project NIETZSCHE Schedule Alignment Meeting, with O5-8, Site Director Hughes, Project MEDUSA Director Vuković and several others in attendance. This time, at O5-8's request, the presentation was broadcast live across the Foundation, and 45 viewers tuned in remotely.
Dr. Langford again reported substantial delays and difficulties to the project, and the principal issue remained that there was not enough data to provably determine any conclusions. Dr. Langford was keen to highlight the extent of the proof that would be required: one objective of the project was to produce a fractal cognitohazard capable of terminating any reality bender, but to prove this would mean proving beyond reasonable doubt that no reality bender cannot be terminated via the same method. This would mean the deaths of multiple reality benders, or accurate simulation, and Dr. Langford expressed that the data available was not nearly sufficient for that.
O5-8 noted surprise that this was still an issue, but pledged that Site Romero-5 would eventually get the amount of data it needed.
2018, July
While attending a conference at Site-19, Dr. Berryman addressed Dr. Langford in a message on the Site Romero-5 internal communications network.
Robert, I've just seen a presentation by Dr. Clef regarding reality benders and it gave me an idea. Have we tried simply going deeper into a candidate modification to the Berryman-Langford set and accessing more deeply-buried regions? I'm wondering if the answer is as simple as overwhelming the reality bender's mind with the much more complex fractal context. I think this wouldn't be too much of a deviation from our existing work to completely invalidate the studies we've done so far.
What an interesting idea. I'll try it immediately.
Before Dr. Berryman's return, Dr. Langford ran several experiments using the suggested method and produced a set of prospective FCIs. They were proven to be effective on baseline humans immediately, and Dr. Langford requested Site-15 to notify him when a reality bender entered its 10-Kempelen population.
It would be several weeks before that was the case, and upon receipt of the message from Site-15, Dr. Langford submitted the new FCIs for testing. Of the full set, one single FCI proved to be effective on the reality bender.
Dr. Langford notified Project NIETZSCHE leadership of his success and continued developing the new method, though he stressed that despite the success, provability was still the most important problem to solve, and therefore the need for reality bender biological data was still relevant.
It is worth noting that while Dr. Clef was present at the Site-19 conference and did lead a seminar on reality benders, Dr. Berryman did not attend it.
2018, August
The Ethics Committee received an anonymous whistleblower report regarding Site Romero-5 and immediately investigated.
Investigation Summary | August 12th, 2018
An anonymous whistleblower report has accused Site Romero-5 leadership of the following issues:
- Unethical restriction for staff to remain on-site.
- Unethical usage of anomalous compulsion effects to influence and control on-site staff.
- Unethical use of human materials for testing purposes.
The Ethics Committee takes such allegations seriously.
Investigation was conducted by EC Investigator Amber H. Goldstein. On 2018-08-21, Investigator Goldstein travelled to Site Romero-5 to interview its staff.
Goldstein was able to interview most staff members one-on-one. The following statements from relevant parties are attached.
Witness Statement: Head Researcher Ava Berryman
Obviously, elephant in the room: [REDACTED]
The other ones just aren't true. I was at the Site-19 conference just earlier this month, and the kind of compulsion you're speaking of — well, there's no one stationed here who can do that, and the— precisely, they're locked down, the staff can't even access them. Only myself and Dr. Langford. Yes, of course, and he has my deepest sympathy.
Witness Statement: Researcher Derrick Coleman
We've never been barred from leaving. I don't think? We get our annual leave, same as everyone else. I mean, we don't get to go home every night, but that's standard operating procedure for classified projects. And the compulsion thing is patently false — the only source of compulsion here is the Third Order cognitohazards, but we're not allowed to look at them, there are even system locks that stop us if we try. No, I've not tried myself, of course not. And then there's [REDACTED] is valid, I suppose, but it's better than the alternative.
Witness Statement: Site Director Samantha Hughes
I've been Site Director for twelve years and I've never seen anything happen here that would warrant EC intervention. You can ask any of my staff. Oh, you did? Then you'll know for sure.
I have the utmost respect for the whistleblowing process, but I will admit I would love to know who started this. I don't suppose— no? I can't fault you for that.
Investigator Goldstein asked to interview Dr. Robert Langford but was informed that he had been hospitalized with an illness of a personal nature, the details of which are not relevant to this report.
Investigation results
Investigator Goldstein recommended the Ethics Committee take no action.
Closing Statement: Investigator Amber H. Goldstein
Site Romero-5 is operating nominally. There are some complaints amongst the staff, but nothing that wouldn't be expected in any other Site. The Site appears understaffed given its size, which could give rise to issues, but I'm told that the workload in general is fine, and besides, the complaints are so specific that they're unlikely to be related.
Regarding issue 1, I found no evidence of any member of staff being unduly restricted to the Site.
Regarding issue 2, I found no evidence of any usage of anomalous compulsion effects, including those that might be expected from accidental application of the cognitohazards that are made here, which does not seem to occur.
Regarding issue 3, [REDACTED]
Overall, most of the whistleblower's complaints are invalid and of those that aren't, there is a very good reason for it. Berryman has been around for decades, she knows what she's talking about, and everything she said checks out.
No immediate intervention is required. I will discuss with the Site Director regarding what should have happened initially, and it will be addressed in due time.
Case number HX-68396 is closed.
BY ORDER OF OVERWATCH COMMAND
In the interests of Foundation security, parts of this report have been redacted.
Following the investigation, in gratitude for their cooperation, Site Director Hughes granted all staff three weeks' immediate leave. She and Dr. Berryman remained on-site to discuss the investigator's recommendation.
Shortly thereafter, Dr. Langford, who was not on-site during the investigation, returned and resumed work.
2018, November: Incident 9001-1
In November 2018, Dr. Langford travelled to Site-15 and presented the third annual Project NIETZSCHE Schedule Alignment Meeting.
The following footage was recorded during the Project NIETZSCHE Schedule Alignment Meeting on 2018-11-01 at Site-19, and was broadcast live across the Foundation. 6 persons were in attendance, with a further 621 watching the live broadcast.
- Dr. Robert Langford: Presenting
- Director Samantha Hughes: Memetics Research Site Romero-5 Director
- Director Vladimir Vuković: Department of Analytics, Project MEDUSA director
- Commander Samhita Reddy: MTF Mu-4 ("Debuggers") subcommander; formerly Department of Analytics Project CIRCE co-director
- Dr. Isha Anand
- Dr. Bruce Miller
- Dr. Leopold Ackermann
Recording starts. The projector is active and the first slide of the presentation is already visible. The camera is mounted on the far side of the room, opposite to the projection screen. On the right side of the room is a small podium holding a laptop. The room is set up for a capacity of around 20 people. All six members of the audience have already taken their seats.
Dr. Langford is standing on the opposite side of the room, occluding some of the projection. He is holding the remote control for the presentation, passing it from hand to hand, watching the door.
Cmdr. Reddy: Robert? We're a minute past the start time. What's the hold-up?
Dr. Langford: All good, just… waiting for stragglers.
Dir. Vuković: Anyone in particular?
Dr. Langford: Well, to be honest, O5-8 usually attends these.
Dir. Vuković: O5-8 was called away on urgent business an hour ago. Besides, this is being recorded, isn't it? And it's only a formality in the first place. Let's get this over with.
Dr. Langford: Believe me, any frustrations you feel about the project timeline, I feel tenfold — but seeing as it was O5-8 who ordered me to make these updates in the first place—
Cmdr. Reddy: This is the third year running we've done this. We all practically know the contents of your presentation off by heart. I'm sure O5-8 does too. Let's just get this over with.
Dr. Langford: I… suppose that's technically compliant. Thank you, Commander.
The presentation begins with Dr. Langford pressing the remote control, switching away from the title slide.
Dr. Langford: You all know me — Dr. Robert Langford of Project NIETZSCHE. A team of nineteen people including myself, led by Dr. Berryman, have been hard at work at Site Romero-5 trying to solve the problem of reality benders' latent resistance to visual cognitohazards. We are currently behind schedule, and I intend to catch everyone up to speed regarding how we're addressing it. As usual, if you have any questions, just raise a hand.
Dr. Langford: As you know, we've already mastered visual cognitohazards via — if you'll excuse the name — the Langford set, an algorithm that, represented visually, produces fractal imagery that can be tuned to have psychoactive effects. We call this technique Fractal Cognitohazardous Imagery, and Site Romero-5 is responsible for production and distribution of FCIs to the entire Foundation. For demonstration purposes, here's one you're all familiar with…
Dr. Langford: This is a Third Order FCI that should have elicited a mild sense of dizziness. Do we have any reality benders in the room?
Dr. Ackermann raises his hand.
Dr. Langford: Dr. Ackermann, you and any other reality benders watching will have noticed that you didn't feel a thing. That is what Project NIETZSCHE specifically intends to resolve. FCIs are an excellent security tool, but if they're unable to affect reality benders, those images are rendered effectively decorative.
Dr. Langford: So, the question is: why are reality benders resistant to FCIs? The answer, at a broad level, is simple: their brains are wired differently. FCIs that affect baseline humans have no guarantee of effectiveness on someone who has a brain with a different structure, different nodes. You've all read the literature; you know the Langford set is encoded for the baseline human brain. The brain of a reality bender is simply different, and it's the sheer simplicity of the issue that makes it so difficult to solve. A question, Dr. Miller?
Dr. Miller: Indeed. You stated this same issue last year to justify the lack of progress. Since then, in my own time, I've taken a look at the medical data we have on file for reality benders, and I cannot corroborate your hypothesis that the brain structure is meaningfully different. There's variation, for sure, but not any more than you'd see amongst baseline humans. With what evidence are you making these assertions?
Dr. Langford: Thank you, Dr. Miller. In your research you'll have noticed, for example, that we've got far, far fewer brain scans for reality benders than we do baseline humans. The crux of the issue, and the real reason our research is taking so much longer than originally planned, is that we simply do not have enough data. Any speculation in the interim is survivorship bias at best. Does that answer your question?
Dr. Miller: Not remotely. I'm unconvinced that—
Dr. Langford: We can discuss this later, then. The number of reality benders available to us is so little, and you'd be shocked at how few volunteer for exposure to First Order FCIs. Despite having made leaps and bounds in the area of artificial test subjects, we can't do the same for reality benders, and we're stuck working with the handful of data points that we have been able to gather. We've submitted—
Dr. Ackermann: Artificial test subjects?
Dr. Langford: Yes — it's rather interesting, actually, and I'll be covering it in a few minutes. We've submitted hundreds of requests to Site-19 for reality benders for research purposes, but naturally, we've yet to receive a single positive response. Yes, Dr. Anand?
Dr. Anand: I apologise if this is already common knowledge — this is my first time attending. In your original proposal for Project HYDRA, you made it very clear that kill agents based on the fractal platform were near-universally fatal for humans. How are Greens meaningfully different? What is stopping you from using the data from just one Type-Green to produce a kill agent?
Dr. Langford: Thank you, doctor. I must correct you: the goal of Project HYDRA was never to construct a universal "kill agent" specifically — moreso a universal agent at all. But you've noted an important issue in that we can only test a First Order FCI on a reality bender once. Proving that a single FCI is capable of terminating a single reality bender, once, is not sufficient. It must be proven that it is capable of terminating any reality bender; additionally, one successful FCI is not a success for Project NIETZSCHE. No — we must be able to produce "kill agents" repeatedly, as many as are required.
Dr. Anand: Why do we need more than one?
Dr. Langford: For security. If someone needs access to a secret document and is therefore immunised against its protective FCI, that would render any projects protected by the same FCI insecure.
Dr. Langford: Now, that's not to say that we've made no progress. I, personally, believe that the answer lies in our computational methods. I have been hard at work with cutting-edge refinements to our procedures and generation logic, and we believe we have found a region of the modified Langford set that is somewhat effective on reality benders. We've only been able to create one FCI from this region, a very weak Second Order FCI, but I hope that further exploration of this region may yield better results.
Dr. Langford: On the next slide is that Second Order FCI. Dr. Ackermann, along with any other reality benders watching, and indeed everyone else: I expect the next slide to elicit a mild feeling of cold. Please observe the very latest in Berryman-Langford fractal agents.
The top half of the visual feed has been blocked, leaving only Langford's legs and the audience visible. All audience members save for Dr. Ackermann collapse.
Dr. Ackermann: What the fuck is that?
Dr. Langford: Um… is everyone…?
Dr. Langford is frozen in place.
Dr. Ackermann: Langford, what the fuck am I looking at? What is this meant to be?
Dr. Langford slowly shuffles sideways towards the podium with the laptop, facing away from the presentation screen.
Dr. Langford: This can't be happening. This can't be real.
Dr. Ackermann: Langford? Langford, please. Tell me what this is.
Dr. Langford: That is a kill agent, Leo. No-one is supposed to see that and live. You might be the only— oh, my God. This is live.
Dr. Ackermann: Langford, please, explain—
Dr. Langford crouches down next to the podium, and then onto his hands and knees. His eyes are squeezed tightly shut. He desperately fumbles with the cables under the podium.
Dr. Langford reaches towards the electrical outlet.
The live broadcast ended 34 minutes earlier than scheduled.
Alexandra.aic, which was monitoring the broadcast, was able to detect and intercept the First Order FCI present in the visual feed, preventing it from affecting any of the remote participants. Director Hughes, Commander Reddy, Director Vuković, Dr. Anand and Dr. Miller were the only casualties of the incident.
On 2018-11-02, the day following Incident 9001-1, Dr. Langford was summoned for debriefing by O5-8.
Project NIETZSCHE Debriefing
In attendance: Dr. Robert Langford, and O5-8 as Project NIETZSCHE Director.
Dr. Langford: It's good to—
O5-8: You're on the record. Explain.
Dr. Langford: It was a standard annual schedule alignment report, to explain why the project had stalled. During—
O5-8: The project has been stalled for three years. Correct?
Dr. Langford: Yes, but not without progress in some—
O5-8: Irrelevant. Continue.
Dr. Langford: Most of the presentation was the same as last year's, including the demonstration coghaz. There was supposed to be an additional demo that would prove that we're starting to be able to make FCIs that work on reality benders. The wrong image was in the presentation.
O5-8: Intention?
Dr. Langford: Of the image that was supposed to be there? To compel a sensation of cold, reality benders included.
O5-8: And reality?
Dr. Langford: The image that was actually included was a First Order FCI, lethal to humans.
Silence.
Dr. Langford: I don't know how it got there.
O5-8: Speculate.
Dr. Langford: Anyone could have copied my presentation from last year, but they would have been missing my latest amendments, such as… the coghaz in question. The only people, other than myself, who could possibly have had access to the presentation were personnel at Romero-5.
O5-8: Be specific.
Dr. Langford: I… can't. Most of them knew that I was making the presentation. Of those, the only person with access to my terminal would have been Dr. Berryman, but it wouldn't have been her.
O5-8: Why not?
Dr. Langford: It wouldn't make any sense. She has been instrumental to the work. She brought me on to Project HYDRA in the first place, and I owe her for that. And Project NIETZSCHE was her initiative entirely; I thought HYDRA was our limit. Sabotaging it now couldn't possibly benefit her.
O5-8: We need an explanation, Langford. Someone must take the fall. Do you understand?
Dr. Langford: I do.
O5-8: Accuse.
Dr. Langford: Myself. I— I mean… I must have made some error, swapped a Third Order and a First Order in the filesystem, I don't know, it doesn't make sense. That's the only answer I have.
O5-8: What measures were in place to prevent this from happening?
Dr. Langford: Well… nothing. FCIs are just images, they're treated the same as any other by the presentation software. But I was exposed to all cognitohazards in that presentation repeatedly while I was preparing it. I should be dead already.
O5-8: Unless you were already inoculated against it.
Dr. Langford: That FCI and its inoculants had already been distributed, according to Alexandra. They would already have been wiped from our systems entirely.
O5-8: Data is data.
Dr. Langford: I suppose. I can't prove it wasn't copied. But proving whether or not I was inoculated would be a witch trial.
O5-8: Yes, there's little sense in that. We lost five good personnel yesterday.
Dr. Langford: Six. I want to leave the project.
O5-8: Yes. Everyone has seen the recording by now; the damage you've done to the project image is immeasurable. Over the past day, the phrase 'Berryman-Langford kill agent' has arisen, presumably from mockery, but is already being baked into the Foundation's collective vocabulary. No one is saying 'First Order FCI' anymore. We won't permit you to leave, however. You remain useful.
Dr. Langford: Amnesticize me, then. I don't want anything to do with this. I already hated seeing my name on that thing.
O5-8: No. We need your expertise, not your labor. We'll find you something to do. Away from the light.
Immediately after this meeting, Dr. Langford was removed from Project NIETZSCHE and placed on administrative leave.
Over the following days, experts at Site-19 and Site Romero-5 convened and discussed the changes to policy that would be required in response to Incident 9001-1, and on 2018-11-06, Dr. Berryman distributed the following memorandum.
Colleagues,
The Project NIETZSCHE Schedule Alignment Meeting a few short days ago was nothing short of a travesty. Our hearts go out to those who knew Dr. Vuković, Commander Reddy, Dr. Anand, Director Hughes and Dr. Miller. They will be sorely missed, but rest assured that their families have been compensated. We are grateful to AIAD and to Alexandra.aic, without whom the death toll would have been considerably higher.
The disaster was an accident. The presenter, Dr. Langford, made a critical error and inserted a lethal Berryman-Langford Kill Agent in place of a safer compulsion agent more suitable for demonstration. This was not a personal failing of Dr. Langford, but a failing of our security and accountability procedures.
A position will be found for him where he can apply his insightful and intelligent nature with no risk of any harm to anyone. We bear him no ill will, and wish him the best of luck.
We are taking this incident extremely seriously, and are making concrete changes to ensure that no incident of this kind will ever happen again. Our foremost priority is to ensure that all cognitohazardous agents produced via the Berryman-Langford process are secure and accountable, and that the process cannot be reproduced or rediscovered without our oversight.
To that effect, a new set of policies is being implemented across the HYDRA and NIETZSCHE projects, effective immediately. These include wide-reaching and technical measures to ensure the absolute safety of cognitohazardous imagery, including new data storage policies, and the strict forbiddance of their usage outside of security contexts, e.g. for cheap demonstrations.
We must ensure consistent and continuous enforcement of these new policies Foundation-wide: therefore, the mathematical core of the Berryman-Langford agent is henceforth reclassified as SCP-9001, and all new security measures are implemented as its containment procedures.
The inciting disaster will be documented as Incident 9001-1, and will be kept on file in full detail, accessible to all, such that we never lose the awful context that necessitated these measures.
Production of cognitohazardous imagery will continue at Memetics Research Site Romero-5. I will take over Samantha Hughes' position as Site Director. As Romero-5 is therefore responsible for containment of SCP-9001, it is now a Dark site: all remote communications between Romero-5 and other Foundation Sites are strictly forbidden.
The security of SCP-9001 is paramount. We will accept no compromise.
Site Director Ava Berryman
Project NIETZSCHE Head Researcher
Authorized by O5-8
Following this announcement, communications with Site Romero-5 ceased.
UPDATE
On 2025-08-22, Protected Site-7 received its quarterly FCI package from Site Romero-5. During routine validation of its contents, Waldor.aic broke from protocol and initiated the deletion process on all newly-received files. Through SCiPnet access, the AIC initiated the site's lockdown procedures, setting the site on High Alert.
Following the initial turmoil and investigation, an All-Clear notice was declared and AIC Waldor was isolated from SCiPnet. Though it had purged the majority of the FCI package, a small selection remained within the database. Of those remaining, none had been marked by the AI as possessing cognitohazardous imagery and are catalogued below.
Undulating branches in a chaotic weave, colored in orange hues. Writhing, squirming, twisting in a specific pattern in varying shades of reaching a pattern. No effects noted within the image.
Fear is a negative emotional response to active/perceived threats or impending danger, suspected to have arisen as a form of survival mechanism (e.g. conditioning living organisms to avoid heights).
Dead region. Dead region.
Dead region. Dead region.
Dead region. Dead region.
Dead region. Dead region.
Dead region. Dead region.
AIC Waldor would not respond to questioning regarding its decision-making and unauthorized destruction of Foundation data. It was the prevailing belief among technicians that the AIC had likely suffered an integer overflow during image processing, interrupting its internal logic — evidenced by its inability to engage in dialogue beyond repeatedly requesting its own termination.
Following a conference with the O5 Council to explain the false alarm, RAISA Director Maria Jones was authorized to open a line of communication to Dr. Berryman's team to inform them of the FCI mishandling and request a resubmission of their quarterly findings. There would be no response to this, or subsequent hails from Site-01.
Satellite imagery of the Site grounds would reveal the presence of additional facilities and building extensions that were neither authorized nor accounted for.
ATTENTION
Site Romero-5 has been designated SCP-9001-A and has been assigned Keter Class. For more information, proceed.
SCP-9001-A.doc
.jpg){kind=link}