Shouldn't IEEE and other publishers make it mandatory for authors to upload codes for article's reproducibility, and reviews be made public, as done in top AI conferences, e.g., NIPS? The question is what is holding IEEE/others back from its implementation?
This way transparency increases, articles' citations would go up and there will be a fair comparison of methods/baselines. Open reviews and answer to reviews will increase transparency. It would also help other authors to see how a final paper evolved from its first version and learn from it. IEEE should start it with top tiered journals and conferences.
But company want to protect their IP.....
I agree to make revireviewsew and reviewers public.
Thanks Antonio Napoli! I completely agree with your thoughts on protecting IP. I am facing the same problem at DLR. I think there exists a middle ground. If some open datasets exist, we could use them for evaluation in adition to confidential datasets. Regarding reviews, it could be problematic if their names are mentioned before, but after the publication it would be nice for accountability.
I think there is a lot of uncertainty about licensing. You would have to balance the use of the dataset to verify the results with the interests of the company. I am sure that a balance will be found here, as the same has already been done for scientific publications in general .Efforts are already underway, for example with IEEE codeocean.
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude