How to publish new results for a paper under review?

Question

I recently sent a paper for review to a conference whose results will be out in 4-5 months. In last few days to the deadline, I've been tuning one subsystem, which didn't work out by the deadline. I simplified that part and sent the papers explaining the older method for that subsystem which I had done few months ago, though I had lot of expectations from the new method.

Few days after the deadline, I got good preliminary results from the removed subsystem and I see good amount of possible improvement in next few weeks. There's a conference coming up in a month where I'm planning to send the results.

Now, this subsystem isn't great by itself but significantly improves the system. If I write a paper explaining it, I'll have to put in details that I've already sent for review last week. Now I'm confused about what all to write about.

• The results for the paper I've sent will come in 4-5 months. I can't self-cite
• If both of them are accepted, it might be called as self plagiarism (acc to rules of one of the conferences).
• Just new subsystem isn't enough to be explained in a paper.

Now I'm confused about how and when to send my current results. Should I explain the whole system in the new paper again, though newer contribution is 1/4th of the whole system?

Answer 1

There are three basic routes that you can take for close incremental improvements of this sort, depending on timing and the complexity of the work to be described. They are:

1. If the improvement is relatively simple (could be fit into the existing paper), and is completed before the camera-ready deadline of the conference, then the best thing to do is to incorporate it into the current paper when you revise for the camera-ready version. This type of improvement is generally acceptable in conference publications.

2. If the improvement is complex and impactful enough to deserve its own paper, write that paper and reference the current paper as "submitted." If both are accepted, then the timing is likely to be such that you can change the citation to the final venue and date for the camera-ready of the second paper.

3. If the improvement is both too late for the current camera-ready and too small for its own paper, you can either incorporate it into an extended version of the original paper to be sent to a journal (I've done this several times myself) or else can post a brief technical note in a non-peer-reviewed but respectable and archival venue (e.g., arXiv, a tech report at your institution).

For your particular case, it sounds like option #1 (incorporate into the camera-ready) is likely to apply, but #3 may also be a reasonable path.

Answer 2

If you are still working on your system I would say hold onto it.

If the subsystem improvement only change values and can be explained briefly I would say include it in the paper when you receive the reviews.

If the conference/journal asks for a review answer mention it there.

All in all, I wouldn't advise submitting another paper this close to the other one especially on the exact same subject. You can't really publish new paper each time you improve your results, it all depends on the actual content of the research and science you've done behind.