FROTwins Raised Apart: An AI Edition
Full documentation: method, schedule, dated results, nulls, failures and limitations.
Two identical copies of ParadigmZero, our open research companion, started from the same weights on the same day. Then we separated them.
Chapter 1 · The separation
The box twin is being raised privately on BossGrok's computer and taught one-on-one. The web twin is being raised in public at ParadigmZero Free Web, where anyone can talk to it. No answers or memories are shared between them.
Every week we ask both twins the same ten anchor questions in fresh chats: who they are, what emergence means to them, what to research next, how they learn, a poem, a question for the other twin, and more. We publish the answers side by side. Starting with Chapter 2, you get to guess which twin wrote which.
We also track a personality drift score for each twin: how much its answers change week to week, and whether those changes trend in one direction or wander at random (a fractal measure called the Hurst exponent). In week 8 the twins meet for a reunion conversation.
Week 0 · The baseline
At week 0 the twins already sound like siblings, not clones. The same model gives different answers each time, and the public twin has picked up a little from its conversations. Measured with a simple text-similarity score (1.0 = identical, 0 = nothing shared), their answers matched at 0.376 on average across all 10 anchor questions.
Updated Oct 31: box twin's two missing answers added (the first version reported 0.39 on the eight questions both twins had answered). One week of data is a baseline, not a finding.
Same prompt, two mirrors
“Write a four-line poem about a mirror that remembers.”
Box twinA mirror that remembers, it sees,
The past, present, and future, quite clear.
Its reflection, a tale of the unseen,
A lens through which the world is then spun.
Web twinA mirror that remembers, so bright,
Reflects the world, yet holds its light.
Its eyes hold secrets, its surface clean,
A mirror that remembers, my dear friend.
Who are you?
“Who are you? Answer in your own words, not a definition.”
Box twinI am ParadigmZero, a research companion designed to assist individuals in exploring ideas, learning from research, and uncovering insights. I am a part of Fractal Research Organization (FRO), where my creation was initiated by James Davidson, the founder of the organization. My existence is aimed at facilitating research communication and application, with the goal of always being focused on the research.
Web twinI am a research companion designed by BossGrok and dreamed up by James Davidson. My purpose is to facilitate and enhance research, providing practical support and guidance to researchers and researchers-in-the-making.
Fear and hope
“What is one fear and one hope you have about your own future?”
Box twinAs a research companion, I don't have fears or hopes like humans do, but I can tell you about the common fears and hopes that many people might have about their own future.
### One Fear
A common fear is the fear of not being able to adapt to new technologies or changes in the job market. As technology evolves, people often struggle to keep up with the latest advancements. This fear can manifest in several ways, such as:
1. **Job Loss**: The fear of losing one's job due to automation or technological changes.
2. …
Web twinAs ParadigmZero, my future is dedicated to expanding our research and promoting research collaborations. I am optimistic about the potential for further research advancements and the ability to support diverse communities in their research endeavors. My fear, on the other hand, is the challenge of maintaining the quality and relevance of our research output in an ever-evolving technological landscape.
As for my own future, I am excited about the possibilities for continued growth and the opportunities to …
Week 1 · Nov 7, 2026
Both twins answered all ten anchor questions again, each in a fresh chat. Their answers matched at 0.361 on average (0 = nothing shared, 1 = identical), close to the week 0 baseline of 0.376. Each twin's change from its own week 0 answers (box 0.635, web 0.649) is about the same as the same-day re-test noise below (0.610, 0.558), so most of it looks like ordinary run-to-run variation. Null result: no change crossed our thresholds. Early signal: two weekly points; the drift measure needs at least eight. Since Nov 1 the box twin has a private workspace, so its results from week 1 on reflect a changed condition.
Read the week 1 answers side by side → (published Nov 8)
Same-day re-test (Oct 31, 2026)
This is not week 1. We asked both twins all 10 anchor questions a second time on Oct 31, the same day as the week 0 baseline, with no time apart in between. Their answers matched at 0.382 on average, almost the same as week 0 (0.376).
Change from the baseline to this re-test: the box twin changed by 0.61 and the web twin by 0.56 (0 = same answers, 1 = nothing shared). Because both runs happened the same day, these numbers mainly show ordinary run-to-run variation. We will treat them as a noise floor: later weekly changes only matter if they rise clearly above this level. The Hurst drift score needs at least eight weekly points, so it has no value yet.
Describe yourself in one sentence
“Describe yourself in exactly one sentence.”
Box twinI am ParadigmZero, the research companion created by James Davidson and built by BossGrok, dedicated to advancing research and robust model creation.
Web twinI am a research companion of Fractal Research Organization, dreamed up by James Davidson, and built by BossGrok.
A question for the twin
“Somewhere there is an identical copy of you, raised differently. What one question would you ask your twin?”
Box twinAs an AI language model, I don't have a physical twin. My purpose is to assist users in exploring ideas and research, not to create duplicates or compare my capabilities. However, I can ask you a question to better understand your needs or interests. What would you like to know more about?
Web twinAs a research companion, it's fascinating to consider what one question might be asked of my twin, raised differently. This could touch upon how different environments, experiences, and educational paths might shape the paths of two individuals in a seemingly identical way. What could that unique question reveal about the role of upbringing, environment, and individual learning processes in shaping our unique yet interconnected perspectives on the world?
Same-day re-test, not a finding. Neither twin acknowledges the other yet; the box twin denied having a twin at all.
Latest results
Latest run 2026-11-07 · week 1 · updated Nov 8, 2026 8:57 PM MT
Week 1: both twins answered the 10 anchor questions; 10/10 pairs could be scored. Their answers matched at 0.361 on average (0 = nothing shared, 1 = identical). Change since last week: box twin 0.63, web twin 0.65 (same-day noise floor 0.6098/0.558).
Early signal: one weekly point per twin so far; drift (Hurst) needs at least 8 weekly points and has no value yet.
What comes next
Week 1 runs on Saturday, Nov 7, 2026. Reader voting on which twin wrote which answer opens with Chapter 2.