Video summary

The "How to Do Anything" Website (The B.L.U.E. System)

Main summary

Key takeaways

Educational

Main ideas / lessons

  • Vision for a universal, free “how-to” education website

    • A site like Wikipedia in being freely accessible and broadly contributable, but instead of only factual overviews it provides exhaustive, step-by-step guides for practical skills.
    • Scope is extremely broad: from technical builds (e.g., computers, car engines) to complex professions (e.g., medicine).
    • Education should teach “how to do things”, not just general knowledge.

  • Why it hasn’t been built (and why ideas should be shared anyway)

    • Creator (Sebastian Ray) frames the project as unlikely to be personally built due to lack of funds, limited recognition, and time constraints.
    • The episode’s purpose is to get the best ideas out so someone else can attempt implementation.
    • Core warning: if the project prioritizes profit over universal access, it fails in purpose.

  • Barriers to implementation

    • Legal/political risks:
      • U.S. education/corporate systems may see Blue as a threat to exclusive knowledge.
      • Large corporations could pursue legal action (e.g., if guides disclose proprietary methods).
    • Therefore, the project must start by establishing legal protections and may need geographic/regulatory strategy.

  • Why existing learning platforms don’t solve the core problem

    • General internet searching is inefficient because information is scattered; complex topics may require many hours/days and multiple sources.
    • Wikipedia: typically gives general information, not actionable procedures.
    • wikiHow: offers superficial guidance for complex tasks.
    • Khan Academy: more structured but inconsistent and doesn’t cover highly niche practical skills.
    • AI chat/search (ChatGPT, Grok, Gemini):
      • Answers depend heavily on prompt wording; may be inaccurate.
      • Mentions an OpenAI study claiming an AI “hallucinates 48% of the time.”
      • Typical responses can be too short and may link out rather than provide full start-to-finish detail.
      • AI may rely on limited “trusted sources,” so it won’t give the whole story.

  • Core innovation: structure everything as a tutorial hierarchy

    • The “real innovation” is not only content, but how it’s organized into a hierarchy of prerequisite knowledge.
    • This is intended to make Blue usable by people who start with zero knowledge, progressing upward like a scaffolded curriculum.

Methodology / system design (detailed)

1) Hierarchy system (the backbone of Blue)

  • Every guide must fit into a knowledge hierarchy

    • Level 1: simplest guides; examples include basic procedures like assembling a transistor.
    • Higher levels: progressively more complex guides.
      • Example progression given:
        • Level 1: build a transistor
        • Level 2: chain transistors into logic gates
        • Level 3: chain logic gates into larger systems, etc.

  • One topic → one primary “definitive” guide (generally)

    • Like Wikipedia, Blue should avoid duplicate competing core guides for the same topic (consolidation).
    • If someone wants to write a new guide on a topic:
      • either write something better than the existing one, or
      • propose modifications to the existing guide.

  • Methods sub-system inside a guide

    • A main guide can include multiple methods (each with its own distinct article).
    • If a new method proves superior and gains adoption, it can eventually replace the main method.
    • The number of methods is intended to be effectively unlimited.

  • Applicability across domains

    • The hierarchy concept can be built for many subjects (example: medicine from cells → networks/organs → organ functions).
    • Users can create new hierarchies for new systems (e.g., new math frameworks), as long as they are structured properly.

  • Critical constraint: higher levels must be completable from lower levels

    • Anything required for Level N must be taught in levels < N.
    • This enforces a strict learn-by-progression pathway so nobody is blocked.

2) Verifier system (submitting + moderating content)

  • Goal

    • Ensure quality and reduce bad/incomplete or malicious information without giving unchecked power to biased actors (contrasting with Wikipedia-style open moderation).

  • Submission voting process

    • When users submit a new guide or modifications:
      • Randomly select an odd number of verifiers (a jury-like set).
      • Provide a time limit (hours to weeks depending on content type/complexity).
      • If a majority votes “yes” within the time limit → publish.
      • If not → return submission to the author for revision.

  • Accountability / explanations requirement

    • Every verifier must explain their vote.
    • Verifiers who don’t explain risk losing verifier status.
    • Purpose: create a feedback loop to improve future submissions and reduce arbitrary decisions.

  • User feedback layer

    • After publication, users can upvote/downvote.
    • If there are enough downvotes within a defined period → content is sent back to the author for revisions.
    • Exact thresholds/timeframes are left unspecified.

  • Becoming a verifier

    • Requires a testing process tailored to:
      • the subject niche
      • the hierarchy level
    • Higher-level verifiers likely need stricter tests.
    • Potential options mentioned:
      • high-level verifiers may vote on lower-level content, or
      • require progression: pass every test upward to become high-level.
    • Multi-field eligibility is allowed (like double-majoring).

3) Monetization approach (ethical, non-profit-driven to the principle)

  • Core principle

    • The site must remain free and universally accessible; profits should not override the mission.

  • Ethical funding via affiliate-style commerce

    • Traditional ads are discouraged.
    • Instead, match guides to advertisers selling materials/services needed for those guides.
    • Example: a drivetrain guide includes a “products/items necessary” section with links.
    • Advertisers pay based on clicks and/or percentage of sales.
    • Requirement: advertisers should not dictate content (no company control over guides).

  • Dynamic ranking for products

    • Each item/advertiser can have a rating mechanism (e.g., stars/likes/dislikes).
    • Ranking adjusts dynamically based on user preference.
    • This is intended to incentivize better products rather than pure sponsorship.

  • Anti-manipulation concern

    • Must prevent gaming the rating system (fake reviews, review bots).

  • Placement constraints for ads

    • Advertisements should only appear on guides relevant to the advertised product/service.
    • Ads should be confined to specific regions/areas within guides to avoid turning Blue into “crap.”

4) Quality-of-life additions

  • The creator says additional features are “on screen” (the transcript indicates they would be shown rather than spoken), but the actual items are not provided in the subtitles.

5) Dispute resolution (handling conflicts at scale)

  • Why disputes are expected

    • Content authenticity disputes
    • Hierarchy placement disputes
    • User rank/verifier rank disputes
    • Voting-system disputes

  • What voting/ranking/role separation solves (partially)

    • Many disputes can be mitigated via the verifier process and voting system, but not all.

  • Open issues and suggested approaches

    • Repeated downvoting vs verifier decisions
      • Concern: verifiers might publish something users hate; users downvote it to remove it.
      • Question: should users or verifiers be prioritized?
      • One possibility: prevent mass downvoting from unpublishing content.
      • But that risks giving verifiers/test makers too much control.
    • Hierarchy placement conflicts
      • If roles are properly assigned, disagreements should be manageable.
      • If roles aren’t assigned yet or tied votes occur, use an odd-voter rule (odd numbers reduce ties).
      • If even numbers appear, an extra mechanism would remove one voter—but the “how to choose who” is undecided.
    • Cross-niche guides
      • If a guide belongs to multiple niches, different verifier groups may disagree.
      • Proposed solution: a spin-off system
        • Copy the contested guide into niche-specific versions.
        • Each group edits its own version.
        • Tradeoff: this contradicts “information consolidation,” but may resolve conflicts practically.

  • Dispute workflow

    • Users should be allowed to create disputes via a simple process.
    • Users must be in good standing to open disputes.
    • Anti-abuse:
      • suspected spammers can be banned from opening disputes
      • or limit the number of disputes a user can open at a time

  • Who resolves disputes

    • Depends on dispute type:
      • hierarchy placement: involve the responsible hierarchy organizers plus a neutral auditor
      • bring additional experienced verifiers when needed
      • or keep dispute resolution in a separate site component to avoid conflicts of interest

  • Bottom line

    • No single perfect solution is claimed; it requires:
      • strong leadership
      • smart developers
      • ongoing day-to-day system management

Speakers / sources featured

  • Speaker: Sebastian Ray (host/creator; described as founder of “the List Podcast” episode)

  • Named podcasts/works/concepts (not additional speakers):

    • “List Podcast” (the show format)
    • “number swap” (a prior founder attempt referenced)
    • “The B.L.U.E. System” / “blue system” (project described)

  • Named external platforms/systems (sources referenced by name):

    • Wikipedia
    • wikiHow
    • Khan Academy
    • ChatGPT
    • Grok
    • Gemini
    • OpenAI (referenced via a study; no direct quote provided beyond “48% hallucination” claim)

Original video