Mind of Apollo project

A collective rational reasoner is a system designed to think, learn, and update its beliefs through a structured collaboration among many human contributors. Unlike an individual reasoner, whose beliefs are stored in a single mind, the collective reasoner's beliefs are explicit, written, and stored in a shared database. This makes them open to inspection, critique, and refinement by others.

Its reasoning process is governed by a workflow, a set of clear rules, procedures, and guidelines that determine how new information is added, how conflicting arguments are evaluated, and how coherence is maintained across the entire belief system. Each contributor, or editor, can propose arguments, counterarguments, or revisions. These proposals are then assessed according to rational standards such as logical consistency, evidential support, and clarity of reasoning rather than popularity or authority.

Over time, as more editors engage with it and as its rules of reasoning are themselves refined, the collective rational reasoner becomes increasingly accurate and sophisticated. It evolves as a distributed mind - one whose understanding of the world grows not through the experience, intuition or emotion of a single person, but through the disciplined interaction of many critical thinkers working within a rational framework.

In essence, the collective rational reasoner aims to embody the ideal of reason itself: a system that continuously corrects its own errors, integrates diverse perspectives into a coherent whole, and seeks to align its beliefs with reality as closely as possible.

At the end of the day, it's the individual editors that do the reasoning. However, they do it in a special way. When they are making reasoning steps like evaluating arguments and deriving conclusions, editors have to assume the existing beliefs of the collective reasoner (the ones that are in the database) as true.

That way the worldview of the collective reasoner remains consistent and it makes sense to think about it as if it was one virtual person rather than a mix of separate perspectives. It's like an individual editor is reasoning on behalf of the collective reasoner and they share their results with the others. Then the next person picks them up and makes the next reasoning step and so on and so on.

The editors are people who want to participate in the collective reasoning process on their own accord. This could be because of one of the following motivations:

• supporting the project's mission
• convincing others
• getting closer to truth
• debating as a sport

This list outlines general principles that apply to any rational reasoner, not just the collective one. It expresses the guiding intentions and aspirations for the project; while the exact methods for realizing each principle are still in development, they provide a direction for how the collective rational reasoner should ultimately function.

Foundations: aim for truth over comfort, separate truth from desire, define terms clearly, distinguish facts from interpretations, keep an open but selective mind.

Framing the problem: state the question precisely, surface and examine assumptions, note stakes and context, decompose complex issues, specify what would count as an answer.

Evidence and sources: seek relevant evidence before judging, prefer primary and high-quality sources, check credibility and conflicts of interest, triangulate independent lines of evidence, distinguish data from anecdotes.

Causation and explanation: don't infer causation from correlation, look for plausible mechanisms, compare competing explanations, prefer simpler explanations when other things are equal, test whether explanations make novel predictions.

Reasoning and inference: use valid logical steps, avoid contradictions, generalize only as far as evidence warrants, apply rules consistently across similar cases, quantify when possible.

Uncertainty and updating: express degrees of belief, update with new evidence, run sensitivity checks on key assumptions.

Counter-checks and error finding: seek disconfirming cases, consider strong alternatives before choosing, test predictions against reality, probe edge cases and exceptions, replicate or re-derive important results.

Coherence and integration: connect claims into a consistent model, flag and resolve tensions with established knowledge, maintain dependency maps between beliefs, avoid double-counting evidence, separate descriptive claims from value claims.

Decision and action: weigh expected benefits against costs and risks, consider opportunity costs and reversibility, choose experiments that maximize learning, time-box analysis when returns diminish, plan contingencies.

Communication and collaboration: show your reasoning steps, cite sources and uncertainties, invite and steelman critique, distinguish factual from value disagreements, use shared rules for revision and conflict resolution.

Reflection and improvement: keep a log of predictions and outcomes, analyze past errors and biases, keep heuristics that work and drop those that don't, pre-commit to standards before seeing results, iterate the process regularly.

There are several sources of motivation for editors to follow the principles of critical thinking and rationality: internal, structural, and social.

Firstly, there are internal motivations. Many editors genuinely want to be good reasoners and to seek truth, motivations that most people share, though for some they are especially strong.

Secondly, there are structural motivations built into the system itself. The platform is designed to organise reasoning in a methodical way that constantly demands justification. For example, you can't simply post long paragraphs of text, you must make your arguments explicit, separate your claims from your evidence, and show how one set of premises follows from another. The workflow itself enforces clarity, coherence, and accountability in reasoning.

Thirdly, there are social motivations. Editors are kept in check by their critics. Both those who hold opposing views and those who simply enjoy scrutinising arguments. If an editor fails to follow rational and critical thinking principles, their arguments will receive low strength scores, which is discouraging. Moreover, they will fail to persuade the collective rational reasoner, and that matters, because its aggregated position is what ultimately gets published and recognised. As the collective rational reasoner gains credibility as an epistemic authority, it becomes desirable to have one's reasoning validated by it. Even if someone manages to "win" temporarily through poor reasoning, their success will soon be undone when a more competent critic examines their arguments.

No, this is not an artificial intelligence. It's the individual human editors who do the actual reasoning. The system that pulls their reasoning together is fairly simple in technical terms. It doesn't attempt to understand the text that is fed into it. It's more of a framework with some formulas and simple algorithms built in.

The collective rational reasoner relies on structure rather than the brilliance of individuals. It provides a collaborative workflow where people contribute through small, well-defined tasks, and the system integrates these into coherent reasoning. To help participants contribute effectively, there will be learning materials and clear guidelines for each microtask.

The system is designed to encourage editors to follow the rules of critical thinking and rationality. By rewarding clarity and consistency, it makes emotional manipulation and other less noble tactics ineffective, ensuring that honest reasoning prevails. Although it isn't meant to be creative on its own, its structure allows the best ideas to rise to the surface.

In the context of this project the word "truth" refers to objective truth. It corresponds to the way things actually are in the world. Objective truth is independent of anyone's beliefs, opinions or feelings.

Editors have different and often competing perspectives, so arguments from all sides are added to the collective reasoner's database. The relative weight of each argument is determined by how well it meets standards of critical thinking, not by popularity or voting. As long as there are editors representing diverse viewpoints, the stronger arguments, regardless of who presents them, will prevail.

Let's take an example: one person wants to ban high-sugar foods because they harm health, while another opposes the ban because it would take away a source of joy. Reason alone can't decide whether health is more important than joy. Their relative importance differs from person to person.

What reason can do is predict how health, joy, and other factors would actually be affected by the proposed ban. It might turn out that the outcomes are very different from what either side expected, prompting them to reconsider. It might also reveal alternative policies that achieve the best of both worlds. And finally, reason might show that the disagreement really boils down to fundamental differences in values, in which case, it's better to stop arguing and focus on more productive pursuits.

This project is not an attempt to eliminate conflict. Its goal is to change the rules of conflict, to change how it is conducted.

The mind of the collective rational thinker is a place designed to elevate rational arguments and make less noble tactics, such as emotional manipulation, ineffective. Within such an environment, disagreement between its editors becomes a productive activity. It generates deeper knowledge and understanding of the issues, helps to reveal which solutions are truly best, clarifies what is at stake for all parties, and may lead to the discovery of “best of both worlds” solutions.

Such an environment can even serve as a fair and satisfying arena for settling disputes through reason before they escalate into violence.

Like with any innovation, there is a potential for it to go wrong. Too many people may start trusting the collective reasoner too much.

However since the reasoning will happen completely in the open, outside critics of the system will have a good view of its flaws. They will limit the influence of the collective reasoner showing that its conclusions are not beyond doubt.

If one collective reasoner gets corrupted, hopefully another will spring up showing what went wrong with it and pulling the resources and influence away from it.

At the end of the day it's a fallible human system and people should remember that. The collective rational reasoner should have a powerful voice, but never unquestionable.

Here's what editors can look forward to when joining the Mind of Apollo at this stage of the project:

The thrill of intellectual battle. Choose your side and defend your view. Try to convince Apollo of your reasoning better than others can.

A community of clear and deep thinkers. Engage with people who value truth, critical thinking, and intellectual honesty, people who cultivate the virtues of good reasoning in themselves.

The novelty of being part of a collective brain. Become one of the first contributors to an entirely new kind of mind — a collective rational reasoner. Help it grow from its humble beginnings into something with superhuman reach and insight.

The chance to make a real difference. By helping the project gain momentum, you'll be contributing to a more rational, wise, and intelligent world, and the sooner that happens, the more harm can be prevented.

The joy of meaningful interaction. Experience the satisfaction of seeing thoughtful, rational responses to what you write. Feel heard, understood, and challenged — surprised by new perspectives, and sharpened by constructive criticism.

The registration is currently open! Please register here.

Apollo is a collective rational reasoner in the making. The Mind of Apollo is an online platform that hosts this virtual being and lets editors interact with it.

On Wikipedia, if you read an article about a contentious issue, you'll often notice that its slant changes depending on the language version. Having Apollo think different things in different languages would be like giving it a multiple-personality disorder.

That's why all content on the platform is automatically translated into all the popular languages. This allows editors from around the world, not just English-speaking countries, to contribute their reasoning and perspectives to Apollo. This collective reasoner is free from the bubble of a single language and less biased as a result. Looking from a different angle, this feature opens the doors to productive, cross-cultural discussion and debate within the Mind of Apollo.

Hi! My name is Dante. I am a software dev with a Maths & Computer Science degree, living in the UK. I am a philosophy enthusiast who spent a whole year full time exploring these concepts as well as several years iterating on them in my spare time.

It will be funded from donations, similar to Wikipedia.