[caddemon]

There's also a difference between outright fake science i.e. lies/fabricated data in the manuscript and bad science i.e. the conclusions drawn by the authors were always "fake" because of bad practices but if you look at the details of the work they are honest about what they did. Of course ideally you would minimize both types of bad paper, but the latter isn't too damaging to the system in isolation while the former can cause a handful of papers to mislead a subfield of science for years. Also how to screen for and how to systemically discourage these two things could be quite different.

[cauch]

And the first one should be divided further into two categories:

1) committed deceiver that started working in this field for years and that somehow managed to not get caught (pretty rare).

2) fake science articles that get published but has absolutely no impact on scientists because scientists don't progress based on randomly found articles, but by meeting the authors in workshops, exchanging with them, ... which make a one-off fake article with fake author totally irrelevant.

If you are a junior scientist, the articles you read are mainly the ones recommended by senior scientists around you, and if you are a senior scientist, you are part of a community, you know the people who publish, and if you see a random article coming from nowhere, you may read it just in case, but you don't let it mislead you or change significantly your own research just based on reading it.

I think it's a flaw on some lawman people when they discuss "fake articles being published": they don't realize how small "having an article published" impact the field. Presenting it in workshop and debating with colleagues does, but what the layman person has in mind will never maintain the illusion.