Who may describe a change to a model, and why is an internal edit not something a public page should promote? This note answers with roles, using a feature change in a study of SPY as the public example.
A change is an internal fact first
A model change is a change to a rule, a feature, a parameter, or a piece of code that a study depends on. Inside the firm it belongs to the people who can read the test: the author, a reviewer, and whoever is accountable for letting the study continue. Promotion is a different verb. Promotion tells an outside audience that the change matters to them. Most model changes do not survive contact with that verb. They are repairs, reversions, or experiments that died. Policy keeps the audience small so a dead experiment is not laundered into a story of progress.
The person who may speak externally about the firm's work is whoever the firm has asked to speak for it, and even that person is limited to what the public pages are allowed to be. An educational note can describe a class of test, such as a replay or a regime check. It cannot advertise that a particular change improved a result. There is no result on offer in these pages. A researcher who wants credit for an edit gets that credit in review, not in a headline. /legal remains the page for the site's terms. It is not a release note.
An internal model change may be documented for reviewers. It may not be promoted on a public page as a performance claim, a personal credit, or a reason to invest.
Changing a feature in a SPY study
Suppose a study of SPY replaces one accepted feature with another version, after the acceptance gate any feature requires. Internally, the note names the old version, the new version, and the tests the new version still has to face. That note is the promotion that is allowed: promotion into the study's own file, visible to reviewers. It is not a public claim that the study got better, faster, or closer to a live order. The SPY example is public and dull on purpose. If the policy only restrained exotic models, ordinary edits would become the loophole.
Nobody outside that review is entitled to recast the edit as a performance narrative. Not the author in a public biography, not a recruiter in a campus talk, not a webpage that wants a concrete story. The allowed public sentence is the class of idea: features are accepted by a person, models face tests, failing models are removed. The forbidden public sentence is the scoreboard. This article does not keep a scoreboard. Personal details about who edited the file are also unnecessary, and personal information is handled under /privacy-policies rather than in a changelog on a market page.
There is a further limit when the change could affect an order path. A research edit and a live release are different events, as the note on research code and live orders says. The person who may approve the research edit is not, by that approval, the person who may turn the edit into an instruction toward a venue. Collapsing the roles is how a feature swap skips the wall. The SPY study does not need to approach that wall to make the point. The point is that even a harmless notebook edit is not public marketing.
The public page stays at the class of fact
Public pages may teach the class. They may say that a model change is reviewed, that tests can kill it, and that a public example such as a SPY feature is still only an example. They may not attribute a market outcome to the change or invite a reader to invest because the change occurred. If a sentence wants to do those things, no role is authorized to publish it here.
Who may promote a model change, on this site, is nobody, if promote means a public performance claim. Reviewers may accept an internal note. Everyone else may read the educational description and the governing pages at /legal and /privacy-policies.
This piece does not announce a model change, a result, or a status. The SPY feature is a hypothetical example. Site terms and privacy handling are at /legal and /privacy-policies.
