Mechanical engineering
Mechanical research, evaluated on what the hardware did
How engineering judgment shows in built systems rather than in published models, and the standard the founding cohort will hold it to. Everything below is a draft in public, on purpose.
What counts as evidence of mechanical research skill
The characteristic artifact of this discipline is not a paper but a thing that either worked or did not. A test rig with its instrumentation documented, a failure analysis that names the mechanism instead of the component, a design revision whose rationale is traceable to a measurement — each of these carries more information about its author's judgment than a citation count, because each one was checked against a physical system that does not care how the argument was framed.
Simulation is evidence only when its validation is. A finite-element study is a claim about reality, and the claim is testable: what was the mesh convergence, which boundary conditions were assumed, and where did the model disagree with the rig? Engineers who report the disagreement are demonstrating the skill; engineers who report only the agreement are demonstrating that they did not look.
Much of the strongest work is proprietary, and the founding rubric is designed around that rather than in denial of it. A test methodology can be described without the part geometry. A tolerance-stack analysis can be shown on a substituted assembly. What an evaluator needs is the reasoning and the numbers that constrained it — not the drawing.
The skill hardest to see from a CV is knowing which uncertainty dominates. Every mechanical system has a dozen error sources and usually one that matters; an engineer who instruments the one that matters, and can say why the other eleven were bounded rather than measured, is showing the judgment this rubric is built to find.
The mechanical research evaluation rubric, first draft
This rubric reads engineering work the way a design review reads a proposal: for whether the numbers support the decision, and for what the author did when they did not.
- Physical validation
- Claims about a system are tied to measurements of that system, with the instrumentation, sample size and conditions stated. A model validated only against another model is reported as such.
- Failure reasoning
- Failures are traced to mechanisms rather than to parts, and the analysis distinguishes what was observed from what was inferred. A root cause the author could not confirm is named as a hypothesis.
- Tolerance and uncertainty discipline
- Error sources are enumerated and ranked, not averaged into a single figure. Margins are justified against a load case, and the dominant uncertainty is identified rather than assumed away.
- Design-decision traceability
- Each significant choice can be followed back to the constraint that forced it. Requirements that were traded away are stated together with what was bought by trading them.
Founding mechanical evaluators will attack this draft first — including against their own test reports and design reviews.
What founding mechanical evaluators will do
Take the rubric apart: where it privileges published simulation over shop-floor evidence, where it cannot be scored from work that stays behind an NDA, where it rewards documentation volume instead of judgment.
Run calibration rounds on public engineering work — open-hardware test reports, standards-body submissions, published failure analyses, teardown studies — scoring independently and comparing spreads, so the first published scores arrive with known uncertainty rather than borrowed authority.
Settle the confidentiality question in the open: what an evaluator can legitimately score from a redacted artifact, and what the community should refuse to score at all.
Who this is for
The founding cohort is looking for engineers whose judgment is already being exercised on real systems, wherever they sit:
- Test and validation engineers whose best work is a report no journal will ever publish.
- Design engineers who want evaluation to read the rationale behind a revision, not the CAD that resulted from it.
- Academic researchers in dynamics, tribology, thermofluids or structures who want their rigs judged as evidence.
- Open-hardware and standards-body contributors whose artifacts are already public and reproducible.
Who this is not for
Self-selection matters more than any filter we could write, so here is the honest version:
- Anyone after a credential for its own sake — the founding stage produces standards, not badges.
- Anyone who cannot discuss methodology at all without clearance; the rubric is built for redaction, but not for silence.
- Anyone uncomfortable having their evaluation accuracy tracked — the calibration record is the point of the design.
- Anyone who needs a live scoring platform today; the mechanics on this page are in design, and the tense is deliberate.
Apply to evaluate mechanical engineering
Mechanical engineering is pre-selected on the application. Link to work we can read — a test report, a failure analysis, a standards submission, an ORCID or repository profile.