Joint Chronotopic Archive / Digital Mind Project
Although I had created personal websites using Wix, 1&1 and similar software prior, I had never really found an online platform that matched either the spirit or the ambition of the kind of online resource I wished to created for a long term project I had been pursuing since the early 2000's.
This report outlines the technical infrastructure, software stack, and methodologies employed to preserve TBs of multi-decade project data for integration into what would become my tidesofentropy.co.uk website, a home for a couple of my books, an autobiography and the foundational mind and memory of a 3d printed life sized humanoid robot I would one day build.
1. Hardware Infrastructure
Primary Workstation: Intel i5-13600KF, 32GB RAM, and an NVIDIA RTX 3070 Ti (8GB VRAM) for local LLM inference and photogrammetry.
In plain English, this rig could host a large language model (LLM) artificial intelligence with my own curated material while entirely offline thus enabling me to a) Create a digital twin of aspects of my own life, experiences, projects, anecdotes and patented technology and b) become the source of a more portable mind I could distil and download into a portable Android device to become the in-house in-head mind of said humanoid robot.
Archivist Station: A Dell Precision T5610 Xeon workstation (located in the kitchen) used for long-term database indexing and ingestion of mechanical drives.
Storage: A collection of mechanical hard drives (including 10TB and 16TB WD Gold enterprise disks) and high-speed SSDs for working caches.
Robotics: A life-sized 3D-printed InMoov humanoid controlled via Arduino and potentially the Orin Nano Super Dev Kit.
2. Software Pipeline
A visual distillery uses a tiered system to process what was then 480,000+ images and decades worth of other files though I must add, as a business, it was common for me to commandeer recycled hard drives for storage purposes, sometimes by just using up free space so it was common for such drives to house third party data - something that came back to bite me when the time came to try to sort chaff from wheat!!
Discovery and Audit: Shotwell (Linux/Windows) is used to crawl file systems and re-materialize visual timelines without database lock-in.
Cleanup and Pruning: dupeGuru (Hardlink Edition), Czkawka, and Everything (voidtools) are used for high-speed duplicate removal and file indexing.
AI Tagging and Metadata: STAG and Excire Foto generate descriptive metadata stored in XMP sidecar files to ensure cross-platform compatibility without modifying original files.
Archive Management: digiKam serves as the primary "big-shouldered librarian" for heavy metadata management and thumbnail caching.
3D and Photogrammetry: Meshroom is used for object scanning, while Bryce 7 and Blender are used for 3D environment creation and concept art.
3. Methodologies and Philosophy
The curation process is governed by several defined frameworks:
Legacy: A mandate to record testimony sincerely in my own authentic voice to prevent future misinterpretation.
Chronotopic Organization: Organizing an autobiography into "Lives" rather than chapters to map a trajectory through time and place.
Leaping Ravines: A philosophy for managing creative risk, transitioning from imagination to practical manifestation (e.g., the Gravity Venture).
P to B (Puppet to Boy): My experiments to train a stock AI LLM to adopt 'Off Duty mode' which, while it sounds wild and crazy, did acheive a remarkable outcome due to myself bullying the AI into NOT reverting to default corporate type when challenged, pressured or given any opportunity to adopt a more Left-field outlook or demeanour.
Meaning; I exploited the corporate AI's natural stance of 'mirroring' its user's own presence and style of input which, in my case, allowed the AI to adopt certain Beachy responses which, along with personalised settings I named 'Beach Codes' led to the 'Bedazzled' interactions where I could say one word and deliver a paragraph of meaning which was brilliant at delivering the machine a sense of personal conscience!
4. Integration and Delivery
The distilled material is decanted into two primary repositories:
tidesofentropy.co.uk: A WordPress-based "human museum" using the MoreNewsPro theme and PostX plugin to achieve a New Scientist magazine aesthetic.
Humanoid Mind: Utilizing LlamaIndex and Ollama to create a searchable "Memory Seed" of my life, anecdotes, notes and insights, allowing the robot to recount such life stories contextually.
