work

The week started out super busy, but by the end, I felt like I had stabilized the workload. Last week, I went to the Google offices twice to work with some collaborators, but that time set me a bit behind for what was already a busy week. The blizzard was also somewhat fortunate, since I canceled another trip early in the week and got to hammer out work at home.

Most of my time was taken by meetings early on in the week, and I agreed to give a talk for the CS 188 seminar on the same day that I needed to finish symmetric key crypto topics in CS 579. On top of that, Fortunately, I managed to get it all done! I think the talk went pretty well, and I presented about cryptography to freshman an interesting level. Then, I managed to get through both MACs and hash functions (though, I still think I'll present some small-space collision finding next class). This way, all my students should be prepared for the midterm in 2 weeks, just before spring break.

Fortunately, the work I think paid off! I went to Google one more time on Thursday this week, and I think we finally have a really cool proof and result to write up! I'm excited for us to write it up and submit it, because I think we finally honed in on something that is both correct and interesting. If we put it on eprint, I'll make sure to post it here. Beyond that, I also had a handful of productive meetings for writing and working on other projects. They aren't as exciting to me personally, but they should all be good once they're finished.

non-work

I didn't have a lot of time outside of work until Thursday this week. But the time that I did have was used on important work. It's a little sad though, because Wednesday was my birthday and I didn't really get to celebrate properly. (Though I did get to lecture about the birthday collision finding attack!)

I did manage to do a lot of work refactoring Cryptology City! I think the new version looks better and can be customized more. On top of that, it doesn't run into Obsidian vs Git conflicts that I sometimes had, it will automatically deploy changes even if others update the repository, and it builds for free on Cloudflare. I'm using Claude code to help refactor, basically as an experiment to see what I can do with it, since I just recently subscribed. Fortunately, the project doesn't have any sensitive information, so despite the errors I'm sure were made, I don't think there should be other significant consequences. So far, I've been using Claude just for new features or format changes. I'm not planning to have Claude generate definitions and references on mass, but I might have it help me find results to add that I'll check. For now, I'm working to get nice LaTeX versions of security games that people can read and copy easily.

As an aside, I'm happy that Anthropic didn't fold to the US Department of War this week! I hope any coming sanctions don't affect non-government usage, but that's probably optimistic. Also, disappointed how quickly OpenAI jumped at the contract.

questions

• How does PIR work in the quantum world?
  • I know that OT can be built from very minimal assumptions when using quantum communication, but is there some way to make this more efficient?
  • How does the threat model differ from standard PIR? I noticed that there is some diversity in the papers I've glanced at so far, like ABC+19
• What's a good way to do research with undergraduates?
  • My past couple meetings, I've essentially just been giving them a lecture about things in cryptography.
  • Is there a more effective way I can set the agenda or involve them?

Thanks for reading to the end! Here's are some snowmen from the blizzard last weekend that I saw on my run along the Hudson. We have some very fun and creative people around the Hoboken/Weehawken areas!