Hi folks, I’m Damien, a scientist at Tumult Labs =)
We’re a company focused on differential privacy, a technology that allows organizations like the U.S. Census Bureau, the IRS, or the Wikimedia Foundation to safely publish statistics from very sensitive data.
To support these use cases, we’ve built Tumult Analytics, the best-in-class framework for differential privacy. And today, we’re releasing it as open-source software
You can find more information about differential privacy in the linked blog post, and you can also directly dive into the software documentation: https://docs.tmlt.dev/analytics/latest/ =)
If you have any questions, don't hesitate to post them below! We'll be monitoring this thread today and we're happy to get feedback & answer questions.
This looks exciting, thank you for sharing! Implementing differential privacy correctly is famously difficult and easy to get wrong. Will you be making the privacy-critical part of your code available publicly for people to audit?
1) Note that the "min/max" example trivially leaks individual information: for example, releasing the max salary of employees of a company leaks the salary of the CEO. More generally, there have been numerous attacks on privacy notions purely based on aggregate data. One of my favorite is this one: https://blog.acolyer.org/2017/05/15/trajectory-recovery-from...
2) Typically, PII is not the only thing that can be used to reidentify someone, and matching records from different databases can sometimes infer sensitive information about people. One example: https://www.cs.cornell.edu/~shmat/netflix-faq.html
Hi, I'm one of the authors of the scientific paper¹ linked in this blog post. Incidentally, I wrote a series of blog posts explaining differential privacy in layman's terms. The first post might be "not-technical-enough" for HackerNews, but maybe the next in the series make up for it. Feedback welcome =)
Excellent point! That's one of the reasons why people understood that syntactic definitions (which try and tell you "how anonymous data is" by looking at the data) were not the right approach, and that you had to look at the mechanism to determine whether the process was sufficiently private.
This line of thinking led to the definition of differential privacy, which completely changed the perspective. I wrote two articles about it, one very simple [1] and the other a bit more detailed [2].
It's doable to compare syntactic definitions (k-anonymity vs. k-map vs. δ-presence, for example) and I tried to do that a bit. Differential privacy is something pretty much fundamentally different, as it applies to the mechanism and not the output dataset.
I'm trying to fix this gap: my PhD thesis is about rephrasing syntactic definitions in terms of relaxed versions of differential privacy. Typically, I'm trying to get to a point where I can say "hey everyone, remember this old definition that is easy to use but we don't really know what guarantee it gave? Turns out it's differential privacy with weakened assumptions, so here, have a formal guarantee for free". These are (I feel) natural questions that are still open problems. I will most definitely write about them when I have time (and solid results) =)
I also run a relay, and I don't have much contact with the Tor community. I'm not sure that most of the relay operators contribute to the IRC channels or other medias used by the community. There are ~5000 Tor nodes, and I just saw ~450 nicks on #tor - and many of them are probably users and not operators.
We’re a company focused on differential privacy, a technology that allows organizations like the U.S. Census Bureau, the IRS, or the Wikimedia Foundation to safely publish statistics from very sensitive data.
To support these use cases, we’ve built Tumult Analytics, the best-in-class framework for differential privacy. And today, we’re releasing it as open-source software
You can find more information about differential privacy in the linked blog post, and you can also directly dive into the software documentation: https://docs.tmlt.dev/analytics/latest/ =)
If you have any questions, don't hesitate to post them below! We'll be monitoring this thread today and we're happy to get feedback & answer questions.