Chapter 16 — The Context File
You have read fifteen chapters about how single-variable optimisation compresses human cognition. You have read twelve interludes that show what that compression looks like from the inside. You now understand — intellectually, and perhaps viscerally — the mechanism by which systems narrow the bandwidth of the people inside them.
Now what?
This is the question that every book like this fails to answer well. The diagnosis is sharp. The prescription is vague. "Think differently" is not a prescription. "Be more aware" is not a method. "Design better systems" is true but unhelpful to a person sitting at a desk on a Monday morning with a KPI review at ten o'clock.
Here is the practical move.
Get the Context File
Download: algorythmics-context.txt
If you just downloaded it, keep reading. This chapter explains what it is, why it works, and how to use it well.
How to use (60 seconds):
- Paste the context file into your AI assistant’s system prompt / custom instructions.
- Start a new chat.
- Ask something real: “Help me think through a decision I’m facing this week: …”
If you want the full text printed inline (for copy/paste), it’s in the appendix.
The Tool That Compresses Can Decompress
You are probably already using an AI assistant. A large language model. A chatbot. A copilot. Whatever the branding, it is a system that takes a context and generates responses calibrated to that context.
The word that matters is context.
Every AI assistant operates inside a frame. That frame is set by the prompt, the system instructions, the conversation history — what the industry calls the context window. The context determines what the model pays attention to, what it treats as relevant, and what it optimises for in its responses.
Most people use their AI assistant with the default context. The default context is: be helpful. Answer questions. Be concise. This is single-variable. The assistant optimises for answering your immediate question and nothing else. It is Goodhart's Law applied to thinking itself — the measure (helpfulness) becomes the target, and everything outside it (long-term consequence, boundary conditions, cognitive depth) disappears.
But the context is programmable.
You can give your AI assistant a different frame. A frame that encodes multi-objective thinking. A frame that includes boundary conditions. A frame that, when you ask it a question about work or money or time or decisions, reflects the question back through the AlgoRhythmics lens before answering.
This is not a gimmick. It is architectural intervention at the point of highest leverage.
If the argument of this book holds — that high-frequency systems compress cognition by rewarding single-variable thinking — then the most practical counter-move is to insert a decompressor into the feedback loop. A system that, every time you interact with it, gently expands the frame. That asks: what are you optimising for? What's the boundary condition? What are you not seeing? What does this look like on a longer time horizon?
Not a guru. Not a therapist. Not a book you read once and shelve. A persistent thinking partner that carries the framework in its memory and applies it to your actual decisions, your actual week, your actual life.
How It Works
The appendix of this book contains the AlgoRhythmics Context File. It is a document — a few pages of structured text — that you can paste into the system prompt or custom instructions of any AI assistant.
Once loaded, the assistant does not change what it does. It changes how it thinks about what you ask. It becomes a boundary-aware reflector.
Example 1: Work
Without context file:
You: How do I increase my team's productivity?
AI: Here are 10 strategies to boost team productivity: time-boxing, daily standups, eliminating meetings...
With context file:
You: How do I increase my team's productivity?
AI: Before we optimise for productivity, let's define it carefully. Productivity measured how? Output per hour? Quality of output? What's the time horizon — this quarter or this year? And what's the boundary condition — what breaks if you push productivity higher? Is the team already near a compression threshold where faster output means shallower thinking?
Same question. Different frame. The second response doesn't refuse to help — it expands the question before answering it. It forces you to think in more dimensions before choosing a direction.
Example 2: Personal
Without context file:
You: Should I take the higher-paying job?
AI: Consider the salary difference, career progression, benefits...
With context file:
You: Should I take the higher-paying job?
AI: What are all the variables, not just the financial one? What does each job do to your time, your cognitive bandwidth, your relationships, your learning rate? What's the 1-year answer vs. the 10-year answer? Is the higher pay compensating you for something it's also taking away?
Example 3: Business decision
Without context file:
You: We need to cut costs by 15% this quarter.
AI: Here are areas to consider: headcount, vendor contracts, office space...
With context file:
You: We need to cut costs by 15% this quarter.
AI: Let's map this before cutting. What's the objective function here — quarterly margin, or long-term viability? Some costs are load-bearing — they don't show up as revenue but they prevent collapse. Where is institutional knowledge concentrated? What's the Facilities Coordinator problem — is there someone whose value is invisible to the spreadsheet but structural to the operation?
The Feedback Loop
Here is the mechanism that makes this more than a novelty.
Every time you use the context-loaded assistant, you practice multi-variable thinking. Not because you are disciplined. Not because you are virtuous. Because the system you are interacting with reflects the framework back at you, and over time, the reflection becomes internalised.
This is the addiction function in reverse. Instead of:
You get:
The bandwidth widens. The feedback loop runs at thinking-speed, not scrolling-speed. The agent does not replace your cognition — it scaffolds it. It holds the framework steady while you learn to hold it yourself.
Eventually, you won't need the context file. You will ask those questions automatically. The frame will be yours.
But until then, the context file is a prosthetic for cognitive depth. A boundary-aware exoskeleton for your thinking.
The Strange Loop
There is something beautifully recursive about this.
This book argues that technology compresses human cognition. The solution it proposes is: use technology to decompress human cognition. The tool that narrows becomes the tool that widens — if you reprogram its objective function.
That is the thesis of the entire book, applied to itself.
AlgoRhythmics is not anti-technology. It is the argument that technology, like any optimisation system, does what its objective function tells it to do. Give it a narrow objective and it compresses. Give it a wide objective — one that includes boundary conditions, long-horizon thinking, and cognitive depth — and it expands.
The context file is the proof of concept. It is the book, distilled into a system prompt, fed into the machine, and reflected back as a way of thinking.
Not a revolution. A feedback loop.
Take it. Use it. Modify it. Share it.
The context file is available as a download — a plain text file you can paste straight into your AI assistant's system prompt. It is also printed in the appendix that follows.
From Prompt to Persistent Partner
The Context File is a one-shot prompt. You paste it, start a chat, and the frame holds for that conversation. But when you close the window, the context resets. You start again from zero.
Bob is the persistent version — a private AI agent that carries this framework continuously. It ingests your email, your bank transactions, your calendar. It builds a memory of your financial and administrative life. It holds the context not for one conversation but for every conversation, and it applies it to decisions that are actually yours.
Same thesis. Same architecture. One runs for an hour; the other runs for a year.
Context is harmony.
The species that built the machine can reprogram the machine.
Start with your own.