MY STARTING POINT
From a trading bot
to an equation hunter.
Before Beyond Brute Force, tournament poker was my craft. I spent around fourteen years developing my game on PokerStars. Later, I put months into building and refining a trading bot. At about the same time, I was introduced to ChatGPT.
I used it to help develop the bot, then began pursuing a different question: could I adapt those tools to investigate mathematical relationships and explore physics? That became the centre of the work: scientific exploration and problem-solving, with coding as a tool for those investigations.
Over roughly three weeks, I turned the earlier work into what I thought of as an equation hunter. My questions ranged across quantum mechanics, general relativity and how our descriptions of the universe fit together. An experiment aimed at making the system more efficient produced an unexpected formula. That became a starting point for the work I have continued developing.
THE RESEARCH BEHIND THE SOFTWARE
Questions first.
Working testbeds followed.
I did not begin with the aim of collecting professional labels. I was new to this kind of computing when I decided to try building a trading bot. AI gave me a way to explore questions, build tools and keep improving how I checked the answers.
My research papers and protocols grew around those investigations. I began building games, animated systems and expanding worlds partly to test whether the computational models produced coherent, repeatable behaviour. Useful archiving, language, reconstruction and learning technologies grew from the same work.
- Physics, observers and context
- Questions about quantum mechanics, relativity and how observations, reference frames and context can be described in mathematical and software models.
- Mathematical problems
- Continuing investigations into P versus NP and other Millennium Prize problems, alongside the practical computational ideas those investigations suggest.
- Symbolic language and meaning
- How richer representations can retain identities, relationships and context across software tasks and language interfaces.
- Memory and reconstruction
- How useful information can be retained and reused, with selected detail reconstructed when an application needs it.
- Worlds, graphics and learning
- Interactive environments, varied creatures, animation and defined learning tasks used as computational testbeds and development tools.
The question that keeps bringing me back is how to turn that understanding into intelligence that can learn and be useful in the world.
FOLLOWING THE CONNECTIONS
One investigation
opened the next.
I kept coming back to the same possibilities: retain what is useful, represent relationships clearly, reconstruct what is needed and check whether the result actually fulfils the request.
Those questions led into archiving, retained working state, interactive worlds, graphics and learning. The same curiosity takes me into ancient construction and the mathematical and physical knowledge behind it. These are investigations I am pursuing; the product evidence pages describe the particular demonstrations and results I can show today.
WHAT DRIVES ME
From understanding worlds
to building useful intelligence.
I wanted to develop the building blocks for general intelligence: the language, memory, learning and world models a useful system might need. My investigations into quantum mechanics, general relativity and the role of an observer led me to explore how objects, relationships and change could be represented in software.
That work grew alongside continuing mathematical investigations, including Millennium Prize problems. The fascination is in the connections: how a compact description can reconstruct something richer, how experience can become useful memory, and how a system can apply what it has learned in a different situation.
My ambition is to build exceptionally capable robots and assistants that learn and work usefully with people. Fictional assistants such as JARVIS capture some of the appeal: intelligence that can understand a situation, remember what matters and help you get things done. That is an inspiration for the long-term research, rather than a description of a finished product.
HOW I WORK
Curiosity needs
something to test.
Working intensively with AI also made its errors impossible to ignore. A convincing answer could send an investigation in the wrong direction. I began developing protocols and tools to make source checking, mathematical constraints and experimental validation part of the process.
I look for tests that could disprove an idea, blind tests where appropriate, repeatable experiments and a record of what passed, failed or remains unresolved. I use AI as a research and development tool, while taking responsibility for the questions, design choices and claims I make from the results.
That approach carries into the software. A compact archive still has to restore the exact file. A local operation still has to complete the task. A faster update still has to produce the required result.
WHAT I AM BUILDING NOW
Ideas people
can explore and use.
Over the past twenty months, I have been turning this work into software, prototypes and a connected family of technologies. Beyond Brute Force now has eight filed UK patent applications, all pending, and a free Mac trial of GLYX Archive.
The wider family includes TokenHoloFold, Seed State OS, Expanding Worlds, HoloRenderer, Glyphtronix, Prooftronix and CogniGlyphonix. Each has a distinct role and its own current stage. My longer-term interest is in systems that can learn usefully, retain context and help people explore more ambitious questions.
I want visitors to understand what I am building, see the demonstrations, inspect the results and try the tools that are ready.