Arvind Narayanan

There's a great op-ed by Stanford's Jonathan Mayer and Princeton's Ed Felten explaining this problem. http://www.slate.com/blogs/future_tense/2013/12/13/nsa_surveillance_an… Their two labs also wrote a solid academic paper on the topic. http://randomwalker.info/publications/cookie-surveillance-v2.pdf

I'm the lead instructor. Let me know if there are any topics you'd like to see in particular. Here's the syllabus from the version of the class I taught at Princeton: http://randomwalker.info/teaching/fall-2014-bitcoin/ (P.S. about 200 students enrolled in the first couple of hours. Great to see the level of interest.)

Arvind Narayanan (who is also on HN as randomwalker) has published papers and given many talks about this fundamental concept: "the level of anonymity that consumers expect—and companies claim to provide—in published or outsourced databases is fundamentally unrealizable." It sounds like this is an example of that principle. It's not feasible for the public database to be both useful and preserve complete anonymity of the doctors. You can check out much more from his two websites: http://randomwalker.info/ and http://33bits.org/

HN strips trailing periods from URLs; add a period at the end, or click the Search link at the page the above links to, to see the case. The raw opinion is here: https://docs.justia.com/cases/federal/appellate-courts/ca9/09-15932/92… It's a very interesting finding. The appeals court actually did reverse parts of the decision, including the hard-drive-to-RAM argument, ending up in favor of MDY for many of the claims! But the devil is in the details. Even though MDY wasn't found to be infringing copyright by breaking the covenants of the license agreement against botting (simply copying static code to RAM was not considered sufficient "reproduction" to be infringement)... it was found to violate the DMCA because it was circumventing Blizzard's Warden framework designed to protect access to dynamic server-side materials. MDY therefore had the lower court's "permanent injunction" against doing so upheld. So if you were to create an ad-blocker detector that detected all current ad-blockers, and required that detector to digitally sign a request for dynamic content, but then someone circumvented that detection by subsequently creating an ad blocker that wasn't detected, would it fall under this precedent and violate the DMCA? And there's also fuzziness around whether MDY was tortiously interfering ( https://en.wikipedia.org/wiki/Tortious_interference ) with Blizzard's contracts with its subscribers. In this case, it wasn't, because MDY arguably was behaving as a good…

If you find something wrong with it, consider this an open offer for me to revise or withdraw the paper as a result. That being said, I've been studying blockchains for years, and this paper has been reviewed by some very high caliber designers of blockchain protocols in both academia and industry. It also uses very simple models as in http://randomwalker.info/publications/mining_CCS.pdf which in my view are extremely unlikely to be incorrect (of course, anything is possible).

Indeed, [1] shows inherent issues with coins that rely purely on transaction fees for eventual security. [1] On the Instability of Bitcoin Without the Block Reward http://randomwalker.info/publications/mining_CCS.pdf

How does this tool compare to other (libre) anonymization software programs, such as ARX [1]? From what I understand, there are only basic routines so to sample records and coarsen a few attributes (e.g. ZIP code, dates) implemented so far. This might also not be sufficient to truly anonymize data, as a large body of research has shown so far [2,3,4] [1] https://arx.deidentifier.org [2] https://www.uclalawreview.org/pdf/57-6-3.pdf [3] http://randomwalker.info/publications/no-silver-bullet-de-identificati… [4] http://arxiv.org/abs/1712.05627

I don't think your scenario is realistic. Are there examples of machine learning used to break into a system? I've heard of other machine-supported techniques which seem more effective, like fuzz testing. But not deep learning. Wouldn't it be more useful to talk about more realistic examples? Like how deep learning methods end up encoding race or gender bias (eg, https://motherboard.vice.com/en_us/article/nz7798/weve-already-taught-… or http://randomwalker.info/publications/language-bias.pdf ) because they are trained on data created by humans with human prejudices? Or using deep learning to keep people "engaged" with a web site or app with in-app purchases, by making making it more addictive? (eg, https://boingboing.net/2018/02/07/literal-sociopathy.html )

It's possible to hide that as well, funnily enough, by also overwriting Object.getOwnPropertyDescriptor (and similar tricks). As far as I know, it's theoretically possible to use this trick to completely 'sandbox' some code so that there's no way it can detect certain functions being overwritten (by overwriting all functions such as Object.getOwnPropertyDescriptor, Function.toString, etc. and making them hide the overwritten functions, including themselves). For some more information: http://randomwalker.info/publications/ad-blocking-framework-techniques…

GA is present on 46% of web traffic [1] or 65% of top web pages [2], so you could be justified to claim the majority... [1]: https://whotracks.me/trackers/google_analytics.html [2]: http://randomwalker.info/publications/OpenWPM_1_million_site_tracking_…

"Anonymized" aggregate data is not as anonymous as you might think. This study [1] was able to accurately de-anonymize users based on internet history and public social media 70% of the time. [1] http://randomwalker.info/publications/browsing-history-deanonymization…

Even without the patient's name attached it is easy to identify people because of the necessary metadata in the record. If you expect to get useful information about, for instance lung disease the record will have to contain information about exposure to likely causes, age, occupation, region of the country (possibly town), sex. It will also contain marital status, whether one has children, drinking and smoking habits, weight, ethnicity. This is pretty close to unique, just like a browser finger print. See for instance: http://randomwalker.info/publications/no-silver-bullet-de-identificati…

here's a direct link to the whitepaper: http://randomwalker.info/publications/ad-blocking-framework-techniques…

Has squatting gotten worse on Namecoin? Squatting is fairly hard to handle in decentralized naming systems in general. Namecoin got a lot of squatting issues mostly because of the pricing function (price of names dropped over x years, and now it's almost free to register names). Here is another paper from WEIS'15 that studied squatting in Namecoin: http://randomwalker.info/publications/namespaces.pdf

The original study: http://randomwalker.info/publications/OpenWPM_1_million_site_tracking_… The relevant section of the study is "6.4 AudioContext Fingerprinting", on page 15. Notably, they link to an online tool they developed that demonstrates this fingerprinting technique [1] [1] https://audiofingerprint.openwpm.com/

See "Why King George III Can Encrypt" (2014) for an alternative metaphor to 'private key' and 'public key': http://randomwalker.info/teaching/spring-2014-privacy-technologies/kin… >We present the user with four items, a key, lock, seal and imprint. The key and lock serve the purposes of encryption: Alice distributes her locks as widely as possible so that others can send her messages that only she can open with her key. Similarly, the seal and imprint handle signing: Alice passes out copies of her imprint so others can verify her as the sender of messages she has stamped with her seal. Collected together, we refer to these four items as a toolkit this abstraction handles the contingency where a user loses her key but not her seal: we insist that the toolkit represents an indivisible unit that must be replaced whenever any element is lost.

JSONB had issue with compression, because the compression is decided on the fly by checking if first part of the serialized data, and in the original format the first part was incompressible (thus no compression was used). But this was fixed in beta3, and so 9.4.0 works fine. Regarding size, it's more complicated. For example after loading the 'delicious-1250k' dataset from http://randomwalker.info/data/delicious/ I get 1322MB for JSON and 1415MB for JSONB, so yes, in this case it's slightly larger. FWIW, both JSON and JSONB use TOAST, because that's how PostgreSQL treats works around the 8kB page size (or whatever page size you use).

There was a really nice paper at the last IEEE Security & Privacy conference where they did an analysis like this on blog postings. http://randomwalker.info/publications/author-identification-draft.pdf I tried a similar analysis for a class project using Twitter data, and it worked surprisingly well considering the small amount of data in a tweet. Would-be-anonymous posters beware!

Now imagine if one of those cyber-stalking startups is successful at large-scale stylometric identification. :P On the Feasibility of Internet-Scale Author Identification: http://randomwalker.info/publications/author-identification-draft.pdf

This was my undergraduate thesis. In short, apart from the long sections where I explain the state of online ad-blocker detection, I built a prototype that can avoid ad-blocker detectors by temporally unhiding the advertisments when a script from the website queries the DOM for information about them. Basically, you can also look at it as having a "user DOM" that has the advertisments hidden and a "website scripts DOM" that has the advertisments visible. I did this by doing script rewriting through a MITM proxy and JavaScript function overwriting (hooray prototypes!). There are some rough edges to this work: Both in form (English is not my first language, and I was working full time while developing it so it is nowhere as polished as it should be) and in content (the 'undetectable' ad-blocker prototype is very basic, and there are some worrying doubts about real-world feasibility in regards to performance and privacy). I also haven't been able to continue developing this project further because of real life constraints. Additionally, a [very innovative paper](http://randomwalker.info/publications/ad-blocking-framework-techniques… on this area came out while I was working on it, which simultaneously made some of my stuff redundant, but also complimented my work very well. You can also see my presentation [here](https://bitbucket.org/joanbrugueram/content-filter-policy/downloads/in… for a quick and very rough summary.