I am a junior and I have taken my fair share of UCLA CS classes. Out of all the professors I have taken, Rafil is the worst of them at teaching. Now let me list the pros and cons of this class to you and you can decide if you want to take it or not:

pros:
1. Professor Ostrovsky is very established in his field. He's also very patient to answer all the questions. He is also funny at times and know lots of stories of famous researchers
2. My TA Eli is amazing, best TA ever. He teaches much much better than the professor
3. As you can see on Bruinwalk, the curve is good
4. He gives out a compilation of his notes from 10 years ago so we can actually read the notes if we get confused from lectures. But surprise surprise sometimes his notes doesn’t agree with his lecture in terms of definition and implementations of things.
5. The material is pretty interesting if your into math and computer security

Cons:
1. The material is pretty hard. Pretty sure it used to be a graduate level class.
2. The professor is terrible at teaching. This lecture is definitely not a good learning experience for me . It looks like he doesn't prepare for the lectures at all and just write down everything from his mind to the (virtual) board, which, unfortunately could be full of mistakes at times, and then he has to rewrite everything several times and your screwed if you are taking notes. His lectures are extremely unorganized. He also likes abusing notations without informing us beforehand so everyone including the TAs get really confused. And then people will start asking him questions for clarification and then he would probably explain it using three times the time he used to explain the concept. He is so unclear that people have so many questions. The lecture could be so much more efficient if he explained everything clearly at the beginning. He also sometimes uses concepts we have never heard of without introducing them and then when people ask what it means he acts all surprised. Dude don’t you think you should probably spend 30 minutes to navigate the syllabus of prerequisites classes to understand what the students do and don’t know? Also, he is usually too lazy to talk about the general big picture of things and he is also too lazy to go to the detailed implementation of things so students are stuck in the middle confused about both what they are actually studying and how exactly does it work. In general, I feel that in order to do good in this class you have to already know everything before the lectures.
3. He gives no homework, so there is absolutely no way you can check if you actually understand the material. Then the exam are hard af and hes too lazy to make practice exams that he asks the TAs to make the practice exams and half of the actual exams.

He is a nice professor. But the course itself is sooooo hard!!!! Not for undergraduate. I do not recommend.

First off, Ostrovsky is a crypto legend so if you're interested in the subject or just want to see what a legend in the field is like, then take this class! but with caution....

The course is extremely difficult and requires a lot of extra effort outside of class to somewhat understand the material. I feel like too many topics were covered and only a very small part of students actually fully understand the material. Professor tries his best to explain the material and I believe he does a good job for how complicated the content is, but for an average undergrad CS major, the material is just very hard. The course only had 2 HWs throughout in the first half of the class, I believe having homeworks or projects to reinforce the course material would be very helpful.

The fact that our final grade is purely determined by two exams is scary. I went through the entire course feeling like I am going to receive an F.

The TAs are helpful especially Eli who is cool.

This class was my favorite class out of all of the ones I've taken here. No, it's not too hard. He has been reducing the material of this class a lot over the years, and he grades on a generous curve. In-person exams are quite simple - mostly just "recite definitions". Just attend every lecture and read about the material. You can do it in lecture.

First, some steps to enjoy this class:

Don't take this class until you've taken many other theoretical CS GEs. It's just more enjoyable when you're comfortable with the theoretical treatment, with some proofs, with swathes of variables.

ALWAYS bring a laptop or something you can use to look up materials. I always had the textbook open during class, along with lecture notes/slides from various universities, blog posts, white papers, etc. I was always reading during lecture, sometimes not even paying attention. But I learned so much from reading about the topics, much more than you could get out of lecture, of course. And his lecture notes are fairly accurate to what he teaches in lecture.

It's true that Ostrovsky is not a superb lecturer or anything, in the sense that he's not super clear and looks like he's winging every lecture. However, he is an interesting lecturer. He is a name in the field (seminal paper on ORAM, in particular), and he knows and always credits all the researchers who worked on some paper or idea, with only the exception of papers he wrote. He is always talking about the modern uses of cryptographic concepts, with direct historical examples as well as non-public information.

The material of this course is more theoretical, in the CS fashion, than it is number-theoretic. Because of this, we skip a lot of traditionally important cryptographic concepts, the kind that rely on number theory (RSA, ECC). Many of the subjects we learned were also post-quantum (though he never mentioned it). His treatment of cryptography was to cover every theoretical Big Idea in cryptography, rather than to focus on common ones like RSA. Every topic in this course could be explained to a layman with motivation - I know because I would often rave about the class to my friend over Discord.

I think this class is a wonderful tradeoff between interesting and unclear. You should read books/papers yourself to find real understanding; lecture is always like that. But when you do understand, I believe you can enjoy this class greatly. It's not for everyone; only motivated students.

This was a fun and interesting class to take if you are interested in cryptography AND you enjoy the more theoretical side of math and CS. Ostrovsky is obviously passionate about the material and he's done a ton of important research in the field, and he definitely tries his best to engage with students and make a fun and lighthearted learning environment. That being said, in true Ostrovsky fashion he isn't the best at lecturing; he's gotten better over the years but his handwriting is still atrocious (made worse by the online format due to COVID-19 and lots of technical difficulties) and he often likes to go in tangents and dance around topics, which often results in muddled explanations that take several rounds of questioning to clear up. This is especially problematic for crypto in particular, which is such a confusing class that each lecture was often 80% questions and 20% new material; however, Ostrovsky is amazing at answering student questions and always gives time for people to clear up their confusion. Additionally, he also provides incredibly detailed lecture notes to study after the fact, and overall I'd say that lectures were worth attending.

In terms of the class itself, it feels more like a grad class than an undergrad one. If you're looking for a practical class this is absolutely not it; the entire class is about the foundations of cryptography and thus is entirely definitions and proofs, with no projects and little discussion of practical algorithms like RSA or SHA. That being said, the material we did cover provides a great introduction to the academic field of cryptography and every other lecture felt like we were learning something mind-blowing and cool. Really, this felt more like a math class than anything else; while concepts were often incredibly hard to grasp upon first presentation, Ostrovsky repeats and re-explains important concepts often enough that you'll have a good working knowledge of everything that's actually important by the end. Discussions were also extremely useful in clearing up confusion (shoutouts to Eli Jaffe, who was probably the most helpful TA I've ever had at UCLA).

The grade breakdown is 45% midterm and 55% final (both take home; I'm not sure if this is typical or not for this class), with optional "homework," which is just questions posed to students during lecture that TAs usually go over in discussion. The exams were honestly extremely difficult and took multiple _days_ each, and most questions required detailed proofs written in LaTeX. This was definitely the worst part of the class, but the questions were usually interesting and made you think, and Ostrovsky was extremely lenient on grading. Overall, I'd say you should take this class if you're interested in the more theoretical/mathematical side of crypto, but stay away from it if you're looking for an easy or a more CS-focused elective.