You are probably right about these fake crystal structures. The moment someone tries to do some chemistry based on these structures, they will fail and eventually figures out that the structure is wrong.

However, your conclusion doesn't follow for fields where experimental results are few and far between. Fraud would be much harder to detect in a field like astrophysics or econometrics. There simply isn't enough data, so rather than contradicting the fraudulent results, you will probably just average truth with lies.