CODExSIMS Systems, made visible

AI-generated · source-grounded · interactive

AI-generated. Evidence-grounded. Built to explore.

Every CODExSIMS explainer is generated with AI. Trust comes from constraining that generation with primary sources, visible evidence boundaries and repeatable browser checks—not from pretending the model cannot be wrong.

The result is more than an illustrated article: change meaningful controls, experiment with animated simulations, answer prediction and quiz blocks, ask questions inside the lesson, or watch and read at your own pace.

Browse explainers Inspect the method
  • Ground the generation Specifications, source code, papers and maintainer explanations constrain every model.
  • Show the boundary Confirmed facts, inferences, versions and unknowns stay visibly distinct.
  • Keep it correctable Tests catch regressions; corrections revise both the model and its explanation.
From source to scene
Source evidence transformed into a visual model

  1. Controls with consequences

    Change a real input and see state, structure or downstream outcomes respond.

    Try it in Cassandra Gossip
    Replication lab Live input

    One replica is behind. A nearest-node read may be stale.

  2. Play with the system

    Use playful simulations to explore tradeoffs, coordination and failure modes safely.

    Try it in Raft Consensus
    Three-node commit Ready
    ALeader BFollower CFollower

    Advance the simulation to replicate a write.

  3. Predict and check

    Pause for predictions, quick quizzes and recall before revealing the answer.

    Try it in PostgreSQL MVCC
    Prediction pause Choose first

    Which version is visible to snapshot 42?

    Make a prediction, then check the model.

  4. Ask inside the lesson

    Request a slower explanation or ask a follow-up grounded in the current explainer.

    Ask inside PostgreSQL MVCC
    Ask at this scene Scene-aware
    Answer

    Choose a question to get an answer grounded in this scene.

  5. Watch or read

    Follow narrated pacing or move through the complete lesson at your own speed.

    Open the complete MVCC lesson
    Lesson 03 · MVCC visibility Watching
    Snapshot 42 started before xid 52
    v3xid 52too new
    v2xid 37visible
    v1xid 12older
    01:12 · visibility check

    Snapshot 42 rejects v3, follows the chain and finds v2 visible.

Inspect the method

Trust should leave a trail.

A trustworthy explainer should let you inspect what shaped it, see where certainty ends and report a correction when the model is wrong.

  1. Source trail

    Start with a complete explainer whose model stays connected to the implementation evidence behind it.

    Inspect a source-backed explainer
  2. Validation

    Build, link, layout and browser checks run against the public project before a page ships.

    View validation runs
  3. Corrections

    If a boundary is wrong or a visual model misleads, file it against the project so the explanation can change.

    Report a correction

Browse

Choose a topic collection