Principles
Machine assistance makes fluent reinvention cheap. The research practice must therefore reward prior-art discovery, correction, precise attribution and reuse of mature intellectual machinery.
Novelty discipline
Novelty is a conclusion of search, not a tone of voice.
Independent rediscovery can be intellectually valuable, but it should be named honestly. Finding that an idea already exists is a successful research outcome because it lets the project begin from a higher floor.
Defensible stewardship · local operating maxim
Epistemic authority should not exceed defensibility.
This formulation is retained as an Agalmic Research operating rule while explicitly acknowledging close antecedents in claim accountability, guarantorship, epistemic dependence and AI-assisted metacognition.
Contribution & authority
Origin deserves credit. Authority requires warrant.
Use standard contributorship roles where possible. The experimental addition is a separate, claim-sensitive authority record describing who can defend a claim and the evidence on which that authority rests.
Treat existing theory, standards and prior art as inputs. Classify a result as antecedent, rediscovery, synthesis, extension, application, operationalization or novelty-unassessed before using stronger language.
Do not coin a local term where a mature field or standard already has a suitable one. New terminology should purchase real precision rather than merely make familiar machinery sound proprietary.
Machine cognition may expand what can be produced faster than it expands what a human curator can understand. This maxim builds on existing claim-accountability and guarantorship traditions; it is an operating rule, not a claim of first discovery.
Use mature contributorship machinery such as CRediT where it fits. Contribution does not automatically confer authority over truth or interpretation, and later validation should not erase provenance.
When an initiator cannot evaluate a promising idea, preserve its origin and route it toward missing expertise, evidence or verification rather than forcing a choice between false authority and dismissal.
Record candidate generation, significance recognition, validation, integration and realization separately. Do not decide by vocabulary alone whether a machine or human “discovered” the result.
Use W3C PROV for ordinary provenance, RO-Crate for research-object packaging and CRediT for contribution roles. Extend locally only where the research problem is not adequately represented.
Knowledge compounds when others can inspect, reuse and challenge it. Technical mechanisms with plausible IP sensitivity should receive an explicit publication/protection review before disclosure.
Early public disclosure creates a timestamped record and invites criticism. It does not prove novelty, correctness, mastery or legal priority over unknown prior art.
When prior art or criticism changes a claim, revise the public object and preserve the reason. Do not silently smooth the history into a story in which the project was always right.
This principle has deep antecedents, especially Herbert Simon. The research task is to identify how machine-generated candidate abundance changes the allocation problem.
The objective is not to maximise the number of ideas. It is to improve the rate at which valuable possibilities are recognized, validated, routed and realized.
Software, models and reproducible analyses force conceptual claims into contact with constraints, but a working implementation is not itself proof of theoretical novelty.
Patenting is an instrument, not a default. It is justified only when temporary exclusivity is plausibly necessary for difficult real-world implementation and after prior art is properly considered.
Agalmic Research should be judged by the quality, lineage-awareness, openness and durability of its work rather than by prematurely imitating an established institute.
Standards stack
A useful default
Open publication is a research method, not a declaration of authority or novelty.
Before a paper enters the numbered sequence, it should have an explicit lineage statement and novelty classification. Where a contribution outruns the initiator’s competence, recruit the missing expertise. Where prior art outruns our formulation, use the prior art.