As a computational chemist, I agree: a lot of computational chemistry studies are useless and just a bunch of calculations on a molecule nobody cares about and that will never be synthesized. In most cases, computational chemists get a good result, publish a paper and then delete the files and forget about it because now they have something else to calculate, generally the information of such results will never reach a laboratory. Then there is the other part of computational chemistry: calculating stuff that has already been determined experimentally. For... Reasons. Just a couple days ago I reviewed a paper of this kind: very nice setup, good calculations and so on. Then I went to check the list of molecules they used, and they had experimental results for every one of them. Mind you, they were not testing the computational methods for accuracy, they were genuinely trying to predict those values...

Well anyway, I'll go look my 50 GPUs burn now.