Please take this in the constructive spirit that it is intended, and please understand that I am fully aware this may just be a result of my own ineptitude, but I have thus far completed about 70% of the new-ish Core Certification exam and I have been having a pretty miserable time with it.
Of all the vendor certification exams I've done, I've never quite seen one that leaves me scratching my head as much as this one does. To recap, we have:
A formerly free certification that is now $350 per attempt, no re-attempts;
An exam that can take 20+ hours to complete (on the high side - maybe not a huge deal for the self-employed but my boss expects me to get this thing done);
A pass mark of 75% in all sections, so getting 5/8 in the Gateway Config exam fails you even if you get 100% in the rest of the test;
No feedback on any sections you've already completed until you finish the entire exam, meaning you could fail the first section and still waste hours or days completing an exam you already failed;
A clause indicating that you could be found to be cheating if your webcam shows anyone 'near you' during the exam - very reassuring for those of us in an open plan office;
No ability to pause/save progress when midway through any section of the exam if anything urgent comes up, only a suggestion to contact support which for international users like me means waiting overnight for an answer.
Any/all of the above would be something I'd consider workable, if pretty annoying. However, some of these questions are written in such an obtuse or vague way that my only conclusion when combined with the above is that the intent of the exam is to be deliberately deceptive.
I obviously can't and won't reproduce the questions here but to give you an idea of what I'm talking about, in the modules I've done so far I've seen several fill-in-the-blank questions that are worded in such a way that multiple answers could be technically true such as 'which roles have the ability to do X in the demo project'. There have also been several questions that are along the lines of 'pick the option that best describes the flow of X feature' where none of the answers are completely correct and you have to do your best to try and figure out what they mean by 'best describes'.
To me this feels deliberately antagonistic, like the attempt to minimize the opportunity to cheat has instead resulted in a testing platform that is actively rooting for you to fail and have to pay again for another crack. It's often less of a test of technical aptitude and more a case of trying to decipher exactly what the question is asking and then guessing when it turns out what it's asking is pretty open-ended.