FYI, I am in CS and EE field. Relevant journals and conferences are organized by IEEE, ACM, USENIX and so on.
I've noticed that many paper submission system don't require neither source code nor experiment data.
In this situation, how reviewers tell if a paper under review is forged or not?
I can imagine that a cunning person proposes a novel-looking idea, but their performance and evaluation is not that good as he/she proposed. Then, I think he/she can forge his/her work slightly so that it cannot be suspected of being forged by reviewers.
Such papers/work make followers have hard time to identify if the work is forged or not.