My last undergraduate exam (many years ago...) was for the course "Microeconomic Theory II" - not a 4th-year course, but we had freedom to schedule. I was aiming to get an overall "Excellent" grade in my BA (= above 8.5/10 as grades are measured in my country). Moreover, this was my last exam, so I wanted it to be a triumph (vanity is never far away...). We knew that the exam would require drawing diagrams, but also performing basic algebraic calculations (finding the extremum of a function -cost minimization, utility maximization, etc). So I went in with pencil, ruler, eraser, sharpener, millimetre paper to draw the diagrams, and glue-tape to glue them on the exam papers. But no calculator - we were allowed calculators only in Statistics, to perform linear regression.
And indeed, there were the diagrams, and I was excited to draw them in such a high-quality (and flashy) manner, and of course, there was the "find the extremum" part. I did -and the end result was not a nice looking, round number. I got suspicious: we knew the professors were usually giving numerical exercises with nice round solutions - as a gift to the students, and perhaps a little easier (to the eye) to grade afterwards. So I re-checked the whole calculations twice (so in all, I did them three times). I could not find anything wrong, so I thought, "hey, this time, no nice and round solution". Apart from this little worry, I was pretty sure I had answered everything perfectly (i.e. completely and correctly).
After the exam ended I realized what I had done, three times in a row: I had "divided a multiplication": there was in front of my eyes something like "$3 \times 4$" ($=12$) and I have repeatedly calculated it as $3/4$ (=$0.75$). Down goes your triumphant final exam...
I got a 10/10. I am not saying that this was a fair grade (because there were maybe other students that performed at the same level as me without making the silly mistake), but obviously the professor saw that it was a silly mistake, and decided to ignore it, (impressed, perhaps, by the unexpectedly executed diagrams). But, my point is, that the only reason he could see it as a silly mistake, was because I had written down all the steps leading to the solution.
So while the correctness of the end-result is important, it does not really convey anything about what the student knows, if it is presented alone (and leaving also aside issues of cheating, etc). It is the arrival method that gives the instructor something to evaluate (the journey and not the destination, as it is said in other contexts...). So I would agree that "answers out of nowhere" should get a "nowhere" grade.