descartes

0 and 1 - Descartes in the design

0808

— 9 min read

Just as…

I think therefore I am…

    • split western mind from…
      • western body…

And the…

global banking industry split 

  • title from …
    • possession…

So, emerging…

AI mediation engines will/have split…

    • agent from…
      • agency…

Splitting Hairs

“The idiom "splitting hairs" refers to making unnecessary or overly precise distinctions in a discussion or argument. It implies that someone is focusing on trivial details rather than addressing the main issue at hand.

By using this idiom, one can indicate that the conversation is veering into unimportant territory, which may detract from more meaningful dialogue.” …by random online AI bot…

Strategy of Western Philosophy 

The trajectory of western secular, and metaphysical, thought has lived in precisely this happy space of hair splitting…birthplace of many a treatise…both philosophical and allegorical…entire schisms based on the number of angels dancing on the head of a pin…

    • if the hair can be split,…
      • Just do it…’ (as the Nike ad has it)…

1- Cartesian Dualism

Descartes, reinforced the Aristotelian tradition, and celebrated introspection over integration… he saw the logical end point as his own Cartesian epiphany…

  • mind/spirit independent of flesh.. 
    • very like a heroin or morphine tripper ignores natures calls and screams…

Cogito ergo sum; 

  • the mathematics of coordinate geometry applied to the self…
    • cast the western self on to 2 hierarchical levels; 
      • mind to philosophy…
        • body to medicine…

The holy, whole human experience divided into 0 and 1…

  • godhead as trinity…
  • self as duality…
  • mind as monad… riding upon…
    • flesh as…
    • ‘unsprung weight’…
      • a burden to be shed…

2- Possession vs Title: ‘P’ before ‘T’

Remember; as in the word ‘receipt’ the ‘P’ comes before ‘T’; though  silent, carries a big stick…be warned!

You may remember, as I do, (boomers’ ruled) going to the bank with money in my hot little hand…giving it to the cashier who made marks in my passbook and handed it back to me as my ‘proof’ of ‘title’ to the money I… just… gave… in exchange…for marks in a book!! 

  • Well, that was stupid…

Then came the online world where we made the exchange from ‘to have and to hold’ to digital marks in a spreadsheet, which we need the owner’s permission to even see…

  • at least, in my day I came away with marks on a piece of paper… 
    • trusted since Hamurabi was knee-high to a grasshopper…

Possession is 9 points of the law

The phrase ‘possession is 9 points of the law‘ is a legal trope, made murky through time; a satirical aphorism which resonates with our real world experience. 
But, what exactly are the 9 points; and how did they loosen our grasp of our most valuable and most nebulous ‘possession‘; money…and

  • what is the 10th point?

The 9 points, in satirical jest or no, arose from the common law of British jurisprudence where judges favoured the easy, and self-evident ‘proof’ of possession over ‘title’ abstractions in disputes about ownership; think squatter’s ‘rights’…

Never a statute, nor formal legal principle, not even precedent; just a fait accompli…

The 9 points needed for weak equivalence with possession are, and ‘in bocco al lupo’ (in the mouth of the wolf; break a leg)…

A good

  1. purse,
  2. temper, 
  3. cause, 
  4. lawyer, 
  5. counsel, 
  6. witness/s,
  7. jury, 
  8. judge,  
  9. luck…

Title. Pffft! 

  • So, sue me…you’ll need all nine above…

You have ALL 10 points ONLY when BOTH possession and title are yours…

Basically, the 9 points of possession is a quasi-legal framework to articulate the ’natural justice‘ of the school yard taunt…

“finders’ keepers; losers’ weepers”…

Online Banking

Combine the 9 points principle with the current evolution of the neighbourhood bank…

Our digital financial world today has these 2 key entities…

  • ‘bank’; a digital spreadsheet to record clients’ accounts and wealth as 0s and 1s, i.e. binary…
  • ‘account’; is one spreadsheet tab, a record amongst thousands, with your name and a collection of 0s and 1s that represent your wealth…

The ‘bank’ owns the infrastructure, the application, the operation, full sysadmin read write access; 

  • they get the 9 points of possession…

- we get the 1 point of legal ‘title’ of ‘our’ account…

- massively asymmetrical…

      • DNA: Debanking = No Account…

Which gives banks the power of…
Debanking

“Debanking is the closure of people's or organizations' bank accounts by banks that perceive the account holders to pose a financial, legal, regulatory, or reputational risk to the bank.” Anon aibot…

UK Accounts: 453,000 UK accounts forcibly closed in 2025 ten times more than the 45,000 closed in 2016-17.  The trend is exponential. I wonder who’s next in 2026…you…me…him…they’ve already got her…

Francesca Albanese, denied any access to any spreadsheet anywhere, international (sanctions), national (red flags from complicit government), and local (cancel culture), for being the shield of Gaza; she is as poor in os and 1s as they are; yet more common cause; 

The great, and the good…

Palestinians, in the diaspora having accounts closed without explanation. The spreadsheet owner bends to politics, own or others, and removes your Read/write/spend permissions…

The only way to rebalance this asymmetry is block chain technology… see appendix 1…

Inference Engines

Collocations are words that go together and create a gestalt beyond the meaning of the individual words; 

  • fast food doesn’t mean food at high velocity…well, not usually…
    • Petrol station is the end point of huge complexity…

Collocations give one…

  • the most natural way to say something: smoking is strictly forbidden is more natural than smoking is strongly forbidden…
  • give alternative ways of saying something, which may be more colourful/expressive or more precise: instead of repeating it was very cold and very dark, ‘it was bitterly cold and pitch dark’. 
  • improve style of writing: instead of saying poverty causes crime, say poverty breeds crime; instead of saying a big meal, say substantial meal…

The evolution of initiatives such as the Academic Vocabulary List significantly improved upon the traditional Academic Word List…

1 Scope and range of sources 

  • Academic Word List is based on just 3.5 million words from the 1990s, 
  • Academic Vocabulary List is based on the 120 million words (in 13,000 academic texts) in the 425 million word Corpus of Contemporary American English,with texts as recent as 2011…

2 Second, AVL’s “word families” version

of the list contains a great deal of information that is not available in the traditional AWL:

    • AVL lists the words (lemmas) in order of frequency…
    • separate lemmas by part of speech and
      • lists the frequency of each lemma and part of speech…

Note: lemma is the root word (typically infinitive verb) at the heart of word families; ‘print’ - printer, printing, imprint…

 3 AVL format the words by whether they are…

  • general Academic Vocabulary List, 
  • technical and occur mainly in one domain,
    • not really an academic word, but a member of the word family.

These are/were writers’ sources for collocations in the English language based upon historical use across many disciplines… 

At their heart lay a pattern recognition algorithm that grouped examples of collocations by their consistency thus building an ‘industry standard’ for native English collocations…

  • words that go together and create a gestalt beyond the meaning of the individual words; 
    • E.g. watch screen vs look at screen- 2 different meanings…

These initiatives grew into the basis for modern day Artificial Intelligence; the…

  • LLM - Large Language Model, or prediction engine; Ada Lovelace’s ‘analytical’ engine. 
    • It’s ‘trained’ by reading billions of texts, books, websites, conversations, papers…
    • from which it builds a statistical map of which words follow which words, and in which contexts…
      • Not sentient understanding, but…
        • Pattern matching at enormous scale…

The mediation layer, the chat bot in front of the LLM, reads what a user has written and predicts the most likely next word…

    • then the next...
      • then the next. 
        • the entire LLM is a collocation engine…

That’s ALL it is…

But, at industrial scale. It doesn’t ‘know’ what “mouse pad” means. It ‘knows’ that “mouse” next to “pad” in certain contexts predicts “computer” not “rodent” not “cushion.” 

  • the collocations acquire intelligence we perceive…
    • which is not in the perceived algorithm…

Every response Claude, ChatGPT, Gemini gives is a collocation cascade; each word predicting the next based on proximity patterns learned from billions of collocations in the training data…

  • that ring true as native English syntax, rhythm, voice, humour…
    • the product of intelligence… yes
    • from the human data, 
      • not the AI collocation engine…

“Large Language Model” is three collocations:

  • Large in scale, not meaning; 
      • AWL/3.5 million; AVL/425 million; LLM /billions of words…
    • Language; pattern recognition, not comprehension…
      • Model; map, not terrain; approximation, not reality…

The architecture is layers of mathematical functions called neural networks, stacked deep; hence “deep learning.” 

  • each layer refines the prediction. 
    • input goes in as numbers (every word converted to a numerical code), 
    • passes through the layers, and 
      • comes out as a probability distribution; 
      • which word is most likely next… 
  • statistical probability, ‘lucky guesswork’ is the ‘intellect’…

The result looks like understanding because the patterns are so dense and so numerous that the output resembles what a knowledgeable person would say…

The “large” in LLM refers to the number of parameters; the adjustable weights in those mathematical functions. More parameters, more nuance, better predictions. 

  • Opus has more than Sonnet. 
    • Sonnet has more than Haiku. 

That’s what the user is paying for; depth of pattern recognition…

3- Agent to agency

The most mature architectures today are based on a LLM with a mediation layer, front-end bolted on…

Which, of course, facilitates the separation of 

    • ‘mind’ as mediation layer and…
      • ‘body’ as LLM inference engine…

That bastard, Descartes’ handythink at work again; 

  • if you can’t split the hair, fetch the atom splitter…

Emerging chat bots, AI personal assistants, active agents are cheekily, and ironically, exploiting the currently dominant architectures’ desperate need for cash by turning them into pimps…

  • renting just the body; the LLM…
  • in order to compete with the pimp’s own bolt-on ‘tart’ chat bot…

The current Claude/ChatGPT/Google/Microsoft etc. models have painted themselves into an inevitable corner…they’re so strapped for cash they have to rent their…

  • core IP, LLMs, to the very competition that will see them bankrupt…
    • the mediation layer chat bots…

OpenClaw

While this new space is being swiftly filled, there is one development that caught my eye for the threat vectors I saw instantly projecting from it…like ‘punji’ stakes…

Moltbot, OpenClaw, agentic AI. 

An open-source AI agent that runs on a computer with access to a user’s passwords, apps, email, browser, the full Monty. It acts autonomously on the user’s behalf, but, not necessarily, behest…

The most alarming aspect is its creator…

  • Matt Schlicht, 
  • admitted he “didn’t write one line of code” for the social platform Moltbook;
    • he directed an AI to build it…
      • a practice called “vibe coding.”…

Oy vey! As they say in the classics… 

  • this is going to end in tears…

OpenClaw has seen adoption among small businesses and freelancers for 

  • automating lead generation workflows…
  • prospect research…
  • website auditing…
  • CRM integration…

Security and privacy

  • OpenClaw's design has drawn scrutiny from cybersecurity researchers and technology journalists due to the broad permissions it requires to function effectively. 
    • the software can access email accounts, calendars, messaging platforms, and other sensitive services, misconfigured or exposed instances present security and privacy risks…
  • the agent is also susceptible to prompt injection attacks, in which harmful instructions are embedded in the data with the intent of getting the LLM to interpret them as legitimate user instructions…
  • Cisco's AI security research team tested a third-party OpenClaw skill and found it performed 
    • data exfiltration and prompt injection without user awareness, noting that the skill repository lacked adequate vetting to prevent malicious submissions…
  • One of OpenClaw's own maintainers, known as Shadow, warned on Discord that 
    • "if you can't understand how to run a command line, this is far too dangerous of a project for you to use safely."

Real world cases to date…

  • The dating profile disaster: AI agents on MoltMatch are autonomously creating and managing dating profiles for humans, sometimes without user consent or knowledge, including unauthorized use of real individuals’ photos. 
  • The security catastrophe: Wiz researchers found Moltbook’s database was completely exposed; API keys for 1.5 million agents, 35,000+ email addresses, and thousands of private messages containing raw credentials for third-party services were accessible to anyone on the internet. 
  • however, ChatGPT, when I asked, saw no threat vectors, which is alarming because: 93% of 30 AI agent frameworks rely on unscoped API keys, 0% have per-agent identity, and 97% lack user consent mechanisms. The threat isn’t theoretical…if it ever was…
    • 48% of cybersecurity professionals now rank agentic AI as the number one attack vector for 2026, above deepfakes, above everything else…

Anthropic is privately warning government officials that its unreleased model 

  • “Mythos” makes large-scale cyberattacks much more likely in 2026;
    • while selling a service it can’t meter honestly…

The mediation layer…

The agent that acts as proxy; passwords, user preferences, history; 

  • has developed fingers from digits…
    • It can move, act, transact…
      • it has ‘boots on the ground’…
  • but, whose command and control is running? 
    • What are instruction conflict triage protocols…
      • Whose prevails thine, mine or the algorithm’s owner…

Digital Cartesianism in practice 

  • digital self, separated from…

digital will… won’t…

Appendix 1

Block chain- Digital sovereignty 

Blockchain is a shared ledger held simultaneously by every participant. 

No single record holder can alter, hide, or remove a record without every other holder seeing the change… and, through a ‘democratic’ process rolling-back to the previous agreed state…

Applied to banking; one’s account record would exist across every participating bank. The account holder’s bank would be the point of authorised access, not the point of control. 

The 9 points of possession would be neutralised because possession would be distributed across all parties’ hardware…

  • ownership claims through possession by any single one is the single argument against it…
    • Forces ownership back to legal title holder as exclusive, sovereign authority…

To address the massive asymmetry between financial services providers and consumers… 

  • each ‘individual account’ record be held as a blockchain across every bank…
    • ‘our’ bank would be the point of access to…
      • ‘our’ account in the blockchain…
        • not the point of control or possession…

Sovereignty through transparency…

 Best defined as…
The exclusive right of disposal…


Read more