Titan Field Lab · governed public exploration

Test the method. Not yourself.

What turns a remarkable claim into a reliable human capability? Follow Titan’s eight gates, then use the four currently released G3 workbenches to examine one low-stakes task without creating a score, diagnosis, validation or Titan result.

Disposable by design

Nothing you enter is sent to or stored by Titan.

The workbench code makes no data requests after the page loads and uses no browser storage. Use fictional or low-stakes material only; do not enter health, legal, financial, safeguarding, employment-confidential or third-party information.

Low-stakes gate

Open the workbenches only inside their boundary.

All three acknowledgements are required locally in this page. They are not sent or saved. If the question is consequential, stop here and use the appropriate qualified route.

Confirm before enabling the workbenches

Workbenches are locked until all three boundaries are acknowledged.

How Titan tests a superpower

Eight gates. No dramatic shortcuts.

A phenomenon can be real without being trainable. Training can work without lasting. A lasting laboratory result can fail in ordinary life. Safety, accessibility, consent and cultural authority can stop a route even when performance evidence looks strong.

  1. 01

    Capture the claim

    State exactly what is being alleged—without silently replacing the original wording.

  2. 02

    Confirm existence

    Ask whether the phenomenon occurs under controlled observation.

  3. 03

    Declare the task

    Specify the actor, setting, configuration, outcome and measurement contract.

  4. 04

    Test acquisition

    Separate genuine change from familiarity, feedback, assistance and ordinary noise.

  5. 05

    Retest later

    Remove immediate practice or feedback and ask whether the result remains.

  6. 06

    Change the context

    Test a meaningfully different task, setting, device or support configuration.

  7. 07

    Test functional value

    Ask whether the result improves something meaningful outside the test itself.

  8. 08

    Pass safety and legitimacy

    Safety, accessibility, consent, attribution and affected-person rights are non-compensatory.

Promotion is a separate decision after the gates—not a ninth automatic step. Popularity, one successful user, a striking video, sponsorship or AI confidence cannot promote a claim.

The currently released set · four bounded G3 workbenches

Make one task observable.

Released at G3 means Titan may publish this reversible, low-risk exploration—it does not mean the underlying capability was newly validated. The broader Research Frontier separately identifies 22 records eligible for a bounded, non-prescriptive informational guide frame; only these four are released as interactive workbenches. Inspect the complete static library of 22 frames. Each workbench is bound to an existing low-risk operational exemplar. The readouts belong only to the task you declare. They are not ranks, diagnoses, assessments, capability levels or evidence that Titan’s method improved anything.

G3 · Explore

Workbench 01

Confidence-tagged quizzing.

Enter at least twenty low-stakes items across the fixed confidence bands and how many were correct. The calculator shows alignment only for this declared item set.

Can show: observed accuracy and task-specific confidence–accuracy alignment.
Cannot show: intelligence, honesty, character or global judgement. Never use it for employment, diagnosis or punitive decisions.

Locked. Acknowledge the low-stakes gate above to enable this disposable workbench.

Record one predeclared item set
G3 · Explore

Workbench 02

Cue-anchored task plan.

Configure one low-stakes intention around an observable event, a completion window, one fallback and a burden stop rule.

Can show: whether one declared cue plan is precise enough to observe later.
Cannot show: a general memory upgrade. Never use it as the sole safeguard for medication, childcare, driving, emergency or legal duties.

Locked. Acknowledge the low-stakes gate above to enable this disposable workbench.

Build the smallest observable plan
G3 · Explore

Workbench 03

Task-duration forecasting.

Compare estimates with actual active time across ten genuinely comparable, low-stakes tasks. Keep waiting time and invalid tasks out of both columns.

Can show: the direction and size of estimation error for this task set.
Cannot show: certainty or global forecasting skill. Never turn this readout into a delivery promise.

Locked. Acknowledge the low-stakes gate above to enable this disposable workbench.

Ten comparable tasks · minutes
G3 · Explore

Workbench 04

Claim-and-counterargument memo.

Separate one low-stakes claim from its evidence, direct observations and assumptions; preserve the strongest opposing case and name what could change the decision.

Can show: whether the reasoning components have been made explicit.
Cannot show: truth, persuasion skill or decision quality. Do not use argument structure to coerce, harass or overwhelm.

Locked. Acknowledge the low-stakes gate above to enable this disposable workbench.

Write a bounded decision memo

What happens next

A readout is the beginning of a question—not a verdict.

A useful result would still need a predeclared repeat, delayed retention, a genuinely different transfer task, functional value and an applicable safety review. This Field Lab performs none of those promotions automatically.