11 KiB
Open Experiments Board
This page is a lightweight coordination board for the Tension Universe MVP experiment layer.
Its purpose is practical:
- show which Tension Universe experiment pages already exist
- show where the current published MVP work can be reviewed
- clarify how contributors should choose the next experiment target
- help the 131-problem field expand one page at a time without duplicate work
This board is intentionally operational.
It is not a full theory index.
It is not a fake placeholder registry.
It is a working contribution board for real MVP pages under TensionUniverse/Experiments/.
What this page means
The Tension Universe field points toward 131 problems in total.
However, this board does not list all 131 items up front with placeholder entries.
Instead, it does something simpler and more useful:
- it shows which MVP experiments are already published
- it points contributors to the current experiment collection
- it makes it clear that the next clean contribution is to choose a valid unbuilt problem that is not already covered
So the contribution logic is:
- if a problem already has a published MVP page, do not duplicate it unless you are proposing a scoped extension
- if a problem is valid and does not yet have a published MVP page, it is a candidate for the next contribution
- if sequencing matters, the maintainer may explicitly reserve certain items later
Current published MVP set
At the moment, 10 Tension Universe MVP experiments are already published.
These are the currently visible completed experiment pages:
If you want the current published set, the cleanest place to review it is:
That index is the actual source of truth for what already exists.
What contributors should do next
If you want to help expand the experiment layer, the rule is simple:
Do not start from the already completed items above. Start from another valid Tension Universe problem that does not yet have a published MVP page.
In practice, that means:
- check the published experiment index first
- confirm your target problem is not already covered
- choose a valid Tension Universe problem outside the completed set
- build one small, inspectable MVP page
- submit a focused issue or PR
This keeps the board useful and prevents duplicate work.
Where to check before building
Before you start, review these pages:
These links are enough to confirm:
- what already exists
- how MVP pages should be structured
- how contributor credit should be recorded
Contribution scope
The expected contribution target is still small and strict.
A clean MVP contribution should usually be:
- one valid problem id
- one focused experiment folder
- one main
README.md - optional notebook, screenshots, or Colab link only if they directly support the page
The intended home is:
TensionUniverse/Experiments/
The preferred pattern is:
TensionUniverse/Experiments/QXXX_MVP/README.md
Here, QXXX is only a formatting template.
It is not a real problem id by itself.
Replace it with a valid Tension Universe problem number before creating the page.
Status model
This board uses a small practical status model.
-
Completed
A visible MVP page already exists inTensionUniverse/Experiments/. -
In Progress
A real experiment page is being drafted, revised, or expanded, but is not yet ready to be treated as a stable published MVP. -
Open
A valid problem has no published MVP page yet and is available for contribution. -
Reserved by Maintainer
A valid problem is intentionally held for sequencing, dependency, or release-planning reasons.
These labels are only for coordination. They do not imply final scientific completion.
Example entry format
If this board later tracks individual items more explicitly, each entry should use a compact real-id format like this:
TU QXXX
Status: Open
Title: Replace with the real working title
Path: TensionUniverse/Experiments/QXXX_MVP/README.md
Scope: One page MVP
Difficulty: Easy / Medium / Hard
Claim: One sentence describing what the experiment is testing.
Claimed by: Unassigned / Maintainer / Contributor name
Important:
QXXX in this example is a template token only.
It is not a real listed Tension Universe problem unless replaced with a valid id.
Practical rules
Before starting a new experiment, keep these rules in mind:
- The page must map to a valid Tension Universe problem id.
- Do not duplicate the already published completed set unless the work is a clearly scoped extension.
- Keep the contribution inside MVP experiment scope.
- The cleanest contribution is one folder plus one main
README.md. - Small notebooks, screenshots, and Colab links are welcome when they directly support the MVP.
- Avoid large unrelated edits outside the target experiment.
Fastest clean path
If you want the fastest contribution path:
- open the published experiment index
- pick a valid problem that is not already completed
- create
TensionUniverse/Experiments/QXXX_MVP/README.md - keep the scope narrow and inspectable
- submit a focused PR
One real page is better than a large unfinished dump.
How this board should grow
This board does not need 131 placeholder entries on day one.
A cleaner long-term pattern is:
- keep the published experiment index updated as real pages are added
- use this board to clarify current contribution logic
- add explicit
In ProgressorReservedentries only when they refer to real problem ids - continue opening new ground by building valid unclaimed problems outside the completed set
That keeps the page readable, accurate, and useful.
If you want to start now
Start by checking the completed set first.
If your target is already listed, pick another valid problem.
If your target is not yet built, keep the scope small, make one inspectable MVP page, and open a focused PR.
That is the cleanest way to expand the Tension Universe experiment layer.
This board is meant to lower contribution friction. If you are unsure what “enough for MVP” means, start here.
Contributor FAQ
This FAQ is intentionally practical.
It is here to answer scope, merge, and workflow questions for contributors. It does not define Tension Universe theory. It only defines the cleanest way to contribute to the public MVP experiment layer.
Q1. Do I need to update every page that mentions completed experiments?
No.
For the MVP contribution path, the minimum required coordination update is this board:
TensionUniverse/Contribute/open-experiments.md
If your experiment is complete, it is enough to reflect that here. Other pages such as the experiments index, Event Horizon page, or root-level references can be updated later and should not block a small focused merge.
Q2. What counts as a valid MVP contribution?
A valid MVP contribution is a small, inspectable slice of a real Tension Universe problem.
The goal is not to solve the full problem. The goal is to produce one honest, usable experiment page at the effective layer.
In most cases, the cleanest MVP is:
- one valid TU problem id
- one focused experiment folder
- one structured
README.md - one small runnable Colab or notebook, if it directly helps verification
A narrow but working slice is enough.
Q3. Does the experiment need to cover the full TU problem?
No.
It only needs to cover one clear experimental slice of a valid TU problem.
That slice may be:
- a toy world
- one observable
- one comparison
- one probe
- one bounded demonstration
If the scope is real, clear, and reproducible, it is valid as an MVP.
Q4. Is one runnable Colab enough for an MVP?
Yes, in many cases.
If one small Colab can run and the README clearly explains:
- what the experiment is testing
- what the inputs are
- what the observable or result means
- what the current limits are
then that is already a strong MVP contribution.
The standard is “small but real,” not “large and complete.”
Q5. Can a README-first contribution still be useful?
Yes.
A clear README is always valuable because it defines the experiment shape, scope, and intended method.
However, for a Completed entry, at least one runnable or inspectable artifact is strongly preferred.
That usually means:
- a small Colab
- a small notebook
- or a clearly reproducible offline procedure
If the README is good but the runnable piece is not ready yet, the item may be better marked as In Progress.
Q6. Is a TXT plus strong-model workflow acceptable?
Yes.
This project is explicitly designed around TXT-based engine packs and strong LLMs.
An MVP may rely on the official TXT pack plus a lightweight notebook or evaluation scaffold, as long as the page clearly states:
- what is reproducible directly
- what depends on an external model
- what is being demonstrated at the effective layer
The workflow should be transparent, narrow, and honest about scope.
Q7. Should I work on an already completed experiment?
Usually, no.
For public MVP expansion, the first priority is to cover a valid problem that does not yet have a published experiment page.
If an experiment is already completed, do not duplicate it unless you are adding a clearly scoped extension. If you do extend an existing item, state that scope explicitly.
Q8. Should I open an issue first, or can I open a PR directly?
Either is acceptable.
- If the contribution is small, self-contained, and clearly in scope, a focused PR is fine.
- If the target, scope, or validity is unclear, open a short issue first.
The cleaner the scope, the less process is needed.
Q9. What should I update after an experiment is completed?
For the MVP merge path, update this board first.
That means:
- add the new completed item here, or
- move the item into the
Completedsection if this board is tracking it explicitly
This page is the minimum coordination layer. Other navigation pages can be synced later.
Q10. What if I can only complete a very small version?
That is still useful.
A small working fragment is better than a large unfinished draft.
If you can make one claim inspectable, one notebook runnable, or one bounded setup readable, that is already real progress. The experiment layer is intentionally built one small page at a time.