low-key week low-key post
Note: This post was imported and originally posted on leaflet on 2025-06-14
Well last post was pretty buggy, so I'm not 100% sure that I'll be able to stick on leaflet. (Although I really like the interface's simplicity.) In case this website stops working, I'll have to look for alternatives. I have a substack account, but to be honest I don't love that interface and lot of other things about that platform. So, who knows what'll happen. 🤷♂️
work
I finished the first draft of my DSO report! I'm sure I'll have to iterate and edit it, but I feel a lot better about this one compared to the first deliverable. It was pretty tough, but satisfying to write summaries of the different OPRF constructions.
I've been trying to focus on my other projects now, but honestly I've found it pretty difficult to focus. I might just need a little break to relax before I push to get as much research done as possible before I start at Stevens. Fortunately, I think I made some decent progress on a proof that is a little tricky to get precisely right, at least.
non-work
I am tantalizingly close to beating Blue Prince. I've found so many secret things, but just have a few more puzzles that I haven't figured out completely. (Although, most of the credit has to go to my friends who have watched me play through it.) I hope I get the chance to finish it up in the next week or so.
Otherwise, I've mostly just been organizing my life more and more settled. The weather has been pretty mixed, but I've been amazed at how nice it is to walk/run even when the weather is dreary. I really hope that my honeymoon phase lasts as long as possible with Hoboken, because it's a really great neighborhood. :)
things i learned this week
The 2025 Gödel Prize was awarded to a cool paper that shows how to extract randomness from low min-entropy sources
A blog post by Matthew Green that discusses some intricacies and shortcomings of the new E2EE chat feature on X
An interview with a former DOGE employee on Hard Fork
A Nieman Lab article about how Guardian uses CoverDrop to anonymize whistleblower information
An Axios article highlighting how Trump has undermined some of Biden's cyber policies
questions
Is there MPC literature that analyzes the types of inputs one should allow into an MPC circuit in general?
In general, one could use zero-knowledge and generic MPC to prove that an input satisfies some circuit
So, you can (in theory) come up with criteria, design a circuit to check the criteria, and then use generic solutions
But what should these criteria bed? E.g., in PSI there is work that restricts inputs to have bounded set size, but are there other "natural" restrictions?
What lower bounds exist around structured encryption? Can you prove that to avoid some form of leakage, there must be some inefficiency?
I remember this paper and other ORAM papers which prove some lower bounds about query-equality or related things
But are there weaker things? Like a lower bound to avoid leaking maximum response length? Does it require latency?
Thanks for reading to the end! Here's a picture of our cat hoping for a treat (that unfortunately didn't exist):
