Kant: Mark I

Aligning in a complexifying world of atom-agnostic intelligences

This is an explorable in progress

Summary

In May 2024, Dr. Abby Innes, summarising Kant, said: “The universe is epistemologically ambiguous and ontologically indeterminant.” At certain population sizes, we re-circumvent this ontological indeterminacy problem by upgrading our fundamental precondition of truth. The intelligence explosion calls for such an upgrade, and this brief proposes a probabilistic, epistemic truth layer as the replacement for the Bible in this role. We believe, subject to further rigour, that we have re-satisfied the ontological indeterminacy problem for the intelligence age.

Background

The pursuit of truth is the only consensus mechanism that works across any group of people. We can think of improvements to this mechanism as follows:

1. Sift a given piece of content into its fundamental claims

2. Determine the extent to which these claims improve the system

3. Reward the contributor for this improvement with a greater reputation

4. Compensate contributors monetarily proportional to their reputation

Formally, we decompose our piece of content into a mini-graph of nodes and hyper-edges¹ representing claims about fundamental facts and fundamental causations. We determine the signal orthogonality, the uniqueness of each claim, and aggregate these orthogonalities inside their respective markets, as well as the impact of these signals in connected markets, to calculate the impact of the content across the system as a whole, which allows us to determine the total contributed signal and resultant reward. Then, we have a propagation step to appropriately weight the meta-graph edges - maintaining the system’s coherence.² We then have a reputation attribution step where the content creator is rewarded non-exchangeable reputation points in return for strengthening the system with this signal. The creator is then compensated monetarily on a regular basis, proportionate to their contribution to the system. We keep reputation currency separate from monetary currency so that reputation cannot be bought and sold - maintaining system integrity.

“It isn’t that the Bible is true, it’s that the Bible is the precondition for the manifestation of truth, which makes it way more true than just true. It’s a whole different kind of true.”
- Jordan B. Peterson

Maths

Here, we will work with the simplest scenario, one truth claim and one truth market, where the market has prior signal (existing claims), and the contributor p creates a piece of content containing a single piece of evidence for that market:

• Market: Atmospheric Measurements

• Claim c₁: “Mauna Loa monthly CO₂ for July 2025 exceeded 425 ppm”

1) Decomposition

Single market, single claim:

\(M_1=(V,E),\;\; V=\{c_1\},\;\; E=\varnothing\)

• c₁: “Mauna Loa CO₂ (Jul 2025) > 425 ppm”

• Contributions K(c₁) = {k} with raw scores s_k ∈ ℝ.

2) Determination

For new batch Ɲ with (s_j, ω_j), ω_j ∈ [0,1]:

\(S_{\text{old}}=\sum_{k\in K_{\text{old}}(c_1)} e^{\,s_k},\qquad S_{\text{new}}=\sum_{j\in \mathcal{N}} e^{\,\omega_j s_j}\)

3) Aggregation

\(\Delta S_1=\log\!\big(S_{\text{old}}+S_{\text{new}}\big)-\log\!\big(S_{\text{old}}\big)\)

\(L'(c_1)=\log\!\big(S_{\text{old}}+S_{\text{new}}\big),\qquad \sigma'(c_1)=g\!\big(L'(c_1)\big)\)

4) Propagation

Not applicable (no causation market).

5) Attribution

\(\Delta S_{1,p}=\log\!\Big(S_{\text{old}}+\sum_{j\in \mathcal{N}_p} e^{\,\omega_j s_j}\Big)-\log\!\big(S_{\text{old}}\big),\qquad R_p(t{+}1)=R_p(t)+\alpha\,\Delta S_{1,p}\)

6) Compensation

\(R_\Sigma(t)=\sum_q R_q(t),\qquad \pi_p(t)=\frac{R_p(t)}{R_\Sigma(t)},\qquad \mathrm{Payout}_p(t)=\gamma\,\pi_p(t)\)

Here is how we might build on our initial scenario:

• Scenario 1: A contributor uploads a report from NOAA’s Mauna Loa Observatory showing the July 2025 monthly mean CO₂ concentration.

• Scenario 2: A contributor posts a comparative analysis, arguing that measurements from Mauna Loa can be used as a global proxy for atmospheric CO₂.

• Scenario 3: A contributor submits a paper showing that increased atmospheric CO₂ is responsible for higher radiative forcing in July 2025 compared to June 2025.

• Scenario 4: A contributor combines NOAA CO₂ data and methane (CH₄) data from global inventories, arguing that both gases jointly explain the quarter-over-quarter increase in greenhouse forcing.

• Scenario 5: A contributor submits an IPCC-style synthesis chapter linking greenhouse gas increases to temperature anomalies and, in turn, to sea-level rise.

“You don’t spend reputation because ratings are orthogonal to exchange.” - Arthur Brock

Commentary

We encourage the reader to test the core model for themselves with the explorable.

First, some model questions:

• What happens when a contributor submits evidence that renders a claim as likely false?: We can either introduce a ‘mirror market’ for each existing market, which is normalised with each other, or we can leave our existing markets as is and introduce a ‘negative market contribution’, which yields positive reputation and compensation

• What happens when a contributor submits evidence that cannot be contained in the existing hypergraph?: We extend the hypergraph, creating new connected truth and causation markets, and the contributor yields all the initial gain from the creation of these new markets

• Doesn’t market naming require a quantum-level understanding of the universe? Potentially, but we will likely circumvent this through a dual mechanism of user-adjusted probability and detail thresholds, re-matching and re-weighting underlying market signals where necessary

• Where does this system fit into a social network’s broader technical stack? The truth protocol underpins a repository where the content is stored, a governance protocol for content production, distribution, and moderation, and the front-end application to access it all

Now, some broader questions:

• Doesn’t this mechanism eliminate the concept of citation? The system ensures that contributors are compensated reputationally and monetarily for their contributions to a given market, and that those contributions are public in a privacy-preserving manner

• Doesn’t this require all content to pursue the truth? Not necessarily, but a social network would run much more smoothly if the community note replaced the ‘retweet without endorsement’ - the content trias politica is a separate mechanism for the production, recommendation, and moderation of content

• Doesn’t this mechanism demonstrate that people cannot know what is true, and so, like robots, cannot be conscious? Potentially, but at this stage we can only claim that we have circumvented Gödel’s Incompleteness Theorem for the intelligence age - there is still a gap between the terrain (the ground truth) and the proverbial map (our protocol)

“AI could never be conscious because it cannot know what is true” - Sir Roger Penrose

Call to Action

We introduce a scale-agnostic, probabilistic, decentralised, open-source, and tokenised truth machine which solves the causal inference capability problem. The model works - it is time for governments to allow free speech and let us increase the truth of the internet 👊

1. The hypergraph’s hyperedges represent the causations, and its edges make the maths work

2. We have no causal claim in Scenario 1, so this step is empty in this piece