Is scientific work considered publishable if it is merely previously unpublished, but not novel?

Question

In general, the purpose of scientific publishing is to disseminate new research. However, is scientific work considered "publishable" if it contains results that are not new, but merely previously unpublished?

For instance, imagine that scientist A makes a certain discovery, announces it to the world, and is able to credibly demonstrate that the discovery was made, but (for whatever reason) does not want to publish any details. Then some time later, scientist B reproduces the discovery. Is it fair game for scientist B to publish their work?

This question was inspired by the recent Deepseek Nature paper, which is notable for being the first published work describing a large reasoning model, even though it is well known that OpenAI had developed similar models earlier but chose to not publish. Indeed, Referee #3 from the released referee reports explicitly mentions this fact. (Now, this may not be the best example, since it seems that the Deepseek work did have novel elements that would make it publishable even if the OpenAI work was previously published. In any case the question is about the general idea, not this specific example.)

---

Note: Here "publish" means to publish as a research article. (For example, the Deepseek paper was published as a regular Nature article, not a Nature review or perspective)

Answer 1

I cannot comment on fields other than math, but at least in math, publication of results which have been announced by others is acceptable if 1) the proofs are substantially different or 2) it is clear that no publication is forthcoming.

But the most common situation in math is where no announcement occurred, but the result is known to the community. These are so-called folklore results where a result is well known (or at least well known among specialists). In such cases, whether the result is itself publishable seems to vary depending on the journal editors and referees(and also may depend on the difficulty of the proof), but mathematicians will publish such results with an explicit statement that they are publishing a proof of a known folklore result. Often when this is done, the folklore result is published along with other novel results. This situation however seems to be very different than the sort of situation with Deepseek and OpenAI that inspired the question.

Answer 2

is scientific work considered "publishable" if it contains results that are not new, but merely previously unpublished?

I don't see why not! I think that your question mixes up a few things: originality of the result, publishability of the result, and honesty about communicating the result. Assuming that the paper is honest that the idea is not original, it may be publishable.

Not all publications communicate new results. Scientific publishing is a form of human communication. The goal is to disseminate some form of information learned by the scientist to the target community of a particular field. Most often, that means disseminating new results, but not always - for example, review and survey papers provide information not about old results, but by summarizing existing results. Such papers provide value to the scientific community which forms their target audience, and so they are often worth publishing.

Some publications even are explicitly designed for such types of papers - this is not the case for the Nature paper you mention, but the same publisher also offers sections like Nature News & Views which are written by a different set of authors to communicate a result by a single author. This seems similar to the scenario you mention.

For instance, imagine that scientist A makes a certain discovery, announces it to the world, and is able to credibly demonstrate that the discovery was made, but (for whatever reason) does not want to publish any details. Then some time later, scientist B reproduces the discovery. Is it fair game for scientist B to publish their work?

I think that there are two scenarios:

• If B claims that the result is new and does not credit A, then this constitutes plagiarism as B has passed the work off as their own.

• If B gives proper credit to A, then it would be a perfectly fair article as it is honest about what it communicates. It would not, however, be an original result.

Either way in the scenario you mention it seems that is it clear to the scientific community that A is the one who first made the discovery. So I see little chance of A's contribution going unnoticed.

Answer 3

Publishable where?

In any given field, or sub-field, or sub-sub-field, there will be a variety of journals, ranging in scope and prestige. From broad based, widely read, and hugely prestigious journals - Nature and Science come to mind, but there are others - to tiny journals read by a dozen or so people. There are even (at least in some fields) the equivalent of "vanity presses" that will publish just about anything, for a fee.

There is, as far as I know, no rule at any journal forbidding the submission of work that is known (say, presented at some conference) but not published (leaving aside issues of plagiarism and copyright, which don't seem to be what you are asking about).