03:12. The timer in the top right corner of the page reminds me how much time I've got left. These aren't minutes, these are hours. 3 hours and 12 minutes to go. In front of me: My 10-year-old laptop, its plugged-in charger; an iced coffee, a banana and two chocolate muesli-bars. I am sitting in room 116 in building 116 at the DTU Campus, sitting my exam in the course "02528 - Computational Data Analysis" - digitally, mind you.
Bring whatever you want
When I sat my first exams last year in December, I was surprised of how permissive the rules were. "Alle hjælpemidler tilladt - med adgang til internettet" - Danish for "All aids permitted - with access to the internet". And this really means: The world (wide web) is open to you.
So I did what everyone did: Merged all PDFs from the slides of the lecture, made sure I have looked at the exercises and had the solutions, went to my exam after skipping through the slides and coming to the conclusion that it "should work out".
And there I was, reading the exam questions, answering what I could answer right-away and searching through the slides for the rest of them.
The contrast
Only a few months earlier, I had written my last exam at ETH. The only aid allowed: "two A4-pages (that is, one A4-sheet of paper), either handwritten or 11 point minimum font size." No other documents. No open-book policy. No computer. Just you, your tightly packed notes, and the exam booklet.
That was the broad standard for my Bachelor's degree. Some exams were slightly more permissive; particularly towards the end, or for courses like Numerical Methods where the lecture scripts ran to 900 and 1,000 pages respectively. But for the most part: one sheet, and good luck.
Both can be designed terribly
Before I come to my opinion, it's worth saying that both formats have real advantages and disadvantages - and you can mess up both.
If you want to test for understanding, then straight asking for definitions is useless. In an open-book exam, the answer is one Ctrl-F away. In a closed-book exam, it is "hit-or-miss" whether someone studied the fact or not. In both cases, I ended up either guessing or searching. Neither felt like understanding.
The other failure mode is using exams primarily as a filter. This is common in first-year courses at highly competitive programmes - it's not unusual to see failure rates of 40–50%. But as a lecturer, this leaves you with three possible conclusions: your course and its contents is generally overwhelming, the teaching wasn't clear enough, or you deliberately designed a hard exam. None of those conclusions reflect well on what classes or exams are supposed to be for.
My experiences on both sides
There was one moment (or let's say, many) in an ETH exam that still stings. I was reading through a question block and felt that specific, horrible recognition: I've seen this. I know this exists. But it wasn't on my cheat sheet. In the two, four or six sides of that summary, I had packed in everything I thought mattered - and this hadn't made it to the sheet. So I sat there, knowing the answer was somewhere in my head, unable to reach it, and watched the points slip away. No recovery. No lifeline. Just the quiet confirmation that I hadn't understood it deeply enough to reconstruct it from first principles. Missed the entire question block. Felt like a beaten dog.
Contrast that with a moment at DTU. Mid-exam, I caught myself opening Google to look up something I probably should have known cold - a basic concept about proteins, the kind of thing that should just be there when you need it. I had to look up a very basic thing (shame on me!). With the knowledge acquired, I (presumably) got the points. But afterward, the question was there: would I have gotten there on my own? It's a strange feeling - performing well on paper while quietly doubting whether the performance was really yours.
As an engineering student...
... I believe open-book exams make more sense - most of the time.
How realistic is the scenario that you are sitting at work, without access to the internet, or other sources? Exactly: almost zero. Very little that you need in engineering needs to be on the top of your head. Instead, you need to have a problem-solving mindset. How to approach the problem? Which techniques or family of techniques am I looking for? The rest of it is something you have seen before and should ideally know where to look up, but there is no need to know everything by heart. Great if you do, but not the point.
That said, this doesn't mean closed-book exams are without value. There are subjects where recall genuinely matters, and there's something to be said for the discipline of knowing material well enough to work without a safety net.
What does "understanding" actually mean?
This is the tension neither exam format fully resolves. Closed-book exams test a narrow, high-pressure version of understanding - can you retrieve and apply this right now, from memory, under stress? That's real, but it's also incomplete. Open-book exams test something closer to professional competence - can you navigate a problem, find what you need, and reason your way to a solution? Also real. Also incomplete.
True understanding is probably somewhere in between: knowing a concept well enough that you could explain it to someone, reconstruct the core logic without notes, and know where to look when the details escape you. The honest truth is that open-book exams make it easier to never find out which side of that line you're on. At ETH, the line was brutal and clear. At DTU, it's blurry - and the blurriness is comfortable in a way that should probably make you a little uncomfortable.
The conclusion
I am really fond of how DTU structures its examinations and evaluations. Be it (graded) hand-ins during the semester, project work or final (applied) exams, or a combination thereof, it comes to the point where it tests exactly what it should: Have you learned to use what was present in the lecture, and what was goal of this course? The transparency of how grades assemble is often a bit unclear, though, but this is something for another day..
All that being said: I sometimes "miss" my ETH exams. They tested me in one of the harshest and most terrifying ways I could imagine, leaving me with a very exact feeling of what I did and what I did not understand of the topic.
Here in Denmark, I sometimes feel a bit like an impostor.
Did I actually understand it? Or do I only believe I understood it and was lucky enough to reconstruct it from the slides, and running the commands right, lacking the big picture?
I'm not sure I always know the answer.
And maybe that uncertainty is the most honest thing either system has taught me.
01:43 left on the clock. I have answered the 18 multiple-choice and 2 open questions in my exam. Some were tricky because they included ambiguous words in their answers. I am looking through my answers one last time before handing it at 12 o'clock with exactly 60 minutes left on my timer. This was my last exam for this exam period. Summer holidays can come.
The author of this post studied Computational Science and Engineering in his Bachelor's at ETH Zurich and currently studies Bioinformatics in his Master's at DTU.
