[00:01:46] too soon, dude ;p [00:02:05] anyway, this is gonna be fun. Because (1) an excuse to write C++ and (2) R does not actually have a "hash this string beyond recognition" library. [00:02:45] Oh, sure, it has digest. Digest is great. Digest /serialises/ the string by default so it doesn't actually compare to any other crypto output with the same input and algorithm, rolls its own implementations of the SHA library just for fun, and isn't vectorised. [14:47:09] morning! [16:17:15] good morning, Ironholds and halfak. [16:17:40] morning leila :) [16:17:45] guess what I spent my yesterday doing? [16:17:56] playing the game? ;-) [16:18:15] nope! [16:18:33] how do you hash data in R? Like, you have a set of unique IDS or whatnot and you want to release the dataset. [16:18:41] How do you cryptographically hash them so they can't be cracked? [16:18:51] I have never done it. [16:19:15] darn. Okay, my example fails. [16:19:28] But I wrote a binding into openSSL that's 0.5-1 orders of magnitude faster than the existing, insecure implementation [16:19:36] mm, but I'm assuming it's a painful process that you've simplified, significantly [16:19:36] and I'm pretty sure I can get it into the Journal of Statistical Software. [16:19:56] yeah, cryptohash(vector, algorithm). Takes about 80ms to apply a SHA-512 hash to 30,000 strings. [16:20:11] that's cool. happy Thanksgiving then. ;-) [16:23:41] I emphasised the giving, by giving back to the community! [16:23:47] and now we can has openSSL bindings. Mwahahahahahahhaa. [16:24:10] that's nice. :-) [16:24:23] how was yours? [16:24:43] spent some time in SF, dinner with relatives. [16:25:15] and today I'm working. :-) [16:26:01] but you have today off too! [16:26:13] I don't [16:26:25] I've asked to work today and get a day off later [16:27:54] ahhh [16:27:58] I should've done that [16:28:28] just enjoy it and do whatever you'd like to do [16:30:34] I guess [16:38:27] okay, I liked [16:38:37] the candidate who thought we were github is not the worst candidate [16:38:47] that would be the person saying they want to do this job in their spare time. [19:07:44] leila, ping [19:07:50] yes, Ironholds [19:08:00] how much do you know about information theory? :D [19:08:12] the next question is: would you fancy going in on a paper with me as second author? [19:09:13] information theory as in, for example, capacity bounds of wireless communication channels? [19:09:29] as in, theory around entropy and anonymity [19:09:35] or cryptography, I guess. [19:09:58] not much, unfortunately [19:10:13] I'm assuming the second question is related to the first. [19:11:06] eh, only slightly. I'm writing up this library I wrote for the Journal of Statistical Software and I could really do with someone familiar with publishing to check that I'm not being tremendously stupid, and point out flaws in my argument or things I might need to emphasise more. [19:11:54] I was wondering if you'd like to chip in :). Everyone gets a paper! [19:12:18] what's the title and abstract? [19:14:31] title, "Cryptographic hashing in R with cryptohash" [19:14:41] abstract: [19:14:42] Reproducability is a fundamental requirement of scientific testing As science becomes more open, [19:14:42] a clash occurs between the desire for transparency around datasets to encourage reproducability [19:14:42] and the ethical obligations a researcher has to avoid releasing private, identifiable information (PII). [19:14:43] Packages exist within R for the purpose of fuzzing this data, as a form of anonymisation, but there is [19:14:46] not currently a fast and cryptographically secure way to anomymise data through encryption. In this paper, [19:14:49] we present a new library, cryptohash, which is designed to efficiently generate cryptographic hashes [19:14:52] of values by providing R programmers with a binding into the openSSL crypto library. We demonstrate [19:14:55] both the speed of the implementation compared to existing mechanisms and several use cases to which [19:14:58] it could be applied, centred primarily around the open science movement and the publication of datasets. [19:16:33] if you're too busy or this isn't your thing I'll find the most amusing person I can :D [19:16:48] Quiddity. I'll get Quiddity to coauthor. He can copyedit like a demon. [19:17:02] except then I'll spend the rest of my life assuming the results are wrong :/ [19:17:04] * quiddity denies everything. [19:17:10] D: [19:17:53] Ironholds, what's your erdos number? I'll co-author anything for an erdos number.... >.> [19:18:15] ooh, good question [19:18:44] Ironholds, if you have the paper ready, send it to me and I'll have a quick look at it to see if I can contribute. [19:18:55] ditto [19:19:14] leila, okee! I'll write it up and send it over :). Expect it Sunday-ish? [19:19:28] whenever you have it. [19:20:38] HAH [19:20:43] quiddity, I /do/ have an erdos number! [19:20:53] woo! [19:21:05] now that's some geektastic resume material. [19:22:04] hahah [19:22:09] it's like 6, though. That hardly counts. [19:22:13] * quiddity highly recommends https://www.goodreads.com/book/show/714583.The_Man_Who_Loved_Only_Numbers [19:22:40] * Ironholds makes absolutely sure [19:24:33] aw, just looks like it's a common name :( [19:27:45] wait, no, I do [19:27:48] then you have to find your Erdős–Bacon–Sabbath number. https://en.wikipedia.org/wiki/Erd%C5%91s_number#Variations [19:27:48] quiddity, I have an erdos number! [19:27:50] and so does halfak! [19:29:13] ha, Pratchett has a super low Erdős–Bacon–Sabbath number http://timeblimp.com/?page_id=195 [19:31:20] 5 [19:31:22] my erdos number is 5. [19:31:23] ...woah. [19:32:03] sweet :) [19:32:38] Stephen Hawking's erdos number is 4. [19:32:40] leila, this is why you should collaborate :D [19:32:56] Erdos for everyone! [19:54:02] Ironholds, ;-) [19:55:27] quiddity, cool! So is halfak's [19:55:32] woah. Halfaker...Hawking... [19:55:42] one bikes for two people... [19:55:45] * Ironholds points at halfak [19:55:47] j'accuse! [20:47:53] Ironholds: very appropriate, these days. http://www.theguardian.com/commentisfree/2014/nov/27/bill-israel-judaism-democracy-president-israel-jewishness-reuven-rivlin [20:48:03] If only we had a Zola. :( [20:48:35] Nemo_bis, sorry? [20:48:59] https://en.wikipedia.org/wiki/J%27accuse [20:54:22] aha