1 00:00:11,360 --> 00:00:14,575 Mike Rugnetta: Friends, hello, and welcome to Never Post, a 2 00:00:14,575 --> 00:00:16,815 podcast for and about the Internet. I'm your host, Mike 3 00:00:16,815 --> 00:00:21,535 Rugnetta. This intro was written on Tuesday, 07/28/2026 at 4 00:00:21,535 --> 00:00:28,640 08:34PM eastern, and we have a dangling show for you this week. 5 00:00:28,720 --> 00:00:32,640 In this episode, I chat with security and privacy researchers 6 00:00:32,640 --> 00:00:38,640 Jan Penekamp and Johannes Lomola about the legions of academics 7 00:00:38,640 --> 00:00:43,435 potentially doxxing themselves in the process of publishing 8 00:00:43,435 --> 00:00:48,315 their own research. It is a wild story of unforced errors, 9 00:00:48,475 --> 00:00:52,075 unnecessary technological complexity, and platform 10 00:00:52,075 --> 00:00:52,875 complacency. 11 00:00:53,690 --> 00:00:56,650 But first, we're gonna take a quick break. You'll listen to 12 00:00:56,650 --> 00:01:00,250 some ads unless you're on the member feed. And when we return, 13 00:01:00,250 --> 00:01:02,650 we'll talk about a few of the things that have happened since 14 00:01:02,650 --> 00:01:08,425 the last time you heard from us. Never Post, the show you're 15 00:01:08,425 --> 00:01:11,545 listening to has some news and we're gonna put it right at the 16 00:01:11,545 --> 00:01:14,985 top of the show so you don't miss it. First, we have released 17 00:01:14,985 --> 00:01:18,940 oodles of new member content over the last couple weeks. 18 00:01:18,940 --> 00:01:23,100 A team chat about party planning platforms with much part of full 19 00:01:23,100 --> 00:01:26,940 slander and its own event invite. You heard about that on 20 00:01:26,940 --> 00:01:30,780 the mailbag if you listen to it. We've released also a dramatic 21 00:01:30,780 --> 00:01:35,475 reading by myself and Georgia of the legendary Tumblr post, Kurt 22 00:01:35,475 --> 00:01:40,035 Cobain will have his revenge on the straights. And most 23 00:01:40,035 --> 00:01:44,115 recently, we uploaded Paper Report, the first in an ongoing 24 00:01:44,115 --> 00:01:48,070 series where we read and discuss published Internet research. It 25 00:01:48,070 --> 00:01:49,670 is a blast. 26 00:01:49,670 --> 00:01:53,350 Those uploads and more, including a slow post and a WIGO 27 00:01:53,350 --> 00:01:56,630 from back in May, are all waiting for folks in the member 28 00:01:56,630 --> 00:02:01,355 feed right now. Second, Member Drive stream week starts on 29 00:02:01,355 --> 00:02:09,035 August 17 at 11AM eastern at twitch.tv/ the never post and it 30 00:02:09,035 --> 00:02:12,920 will continue through the rest of the week. We have many of 31 00:02:12,920 --> 00:02:15,880 your favorite guests from the extended Never Post universe and 32 00:02:15,880 --> 00:02:19,080 more. Joining us throughout the week, got games and live segment 33 00:02:19,080 --> 00:02:22,520 recordings, giveaways, and hopefully a bunch of new members 34 00:02:22,520 --> 00:02:25,960 so we can keep making this show. A schedule and further details 35 00:02:25,960 --> 00:02:30,085 will go out in two weeks, but as of right now, the week of August 36 00:02:30,085 --> 00:02:33,365 17, clear your schedule, cancel your plans, call in sick to 37 00:02:33,365 --> 00:02:37,445 work, and finally, we have a survey. 38 00:02:37,445 --> 00:02:42,640 But here's the catch, we don't want you to fill it out. No. Not 39 00:02:42,640 --> 00:02:46,320 you. Do not click that link. We would love it if your friends, 40 00:02:46,320 --> 00:02:50,240 family, coworkers, folks who do not currently listen to the show 41 00:02:50,240 --> 00:02:51,440 could fill it out. 42 00:02:51,520 --> 00:02:54,695 We are working with some of the wonderful folks at Radiotopia to 43 00:02:54,695 --> 00:02:59,495 do some audience development research and we wanna hear from 44 00:02:59,495 --> 00:03:04,295 some folks who do not know what Never Post is except maybe in 45 00:03:04,295 --> 00:03:06,535 passing because you won't stop talking to them about it. We 46 00:03:06,535 --> 00:03:08,455 wanna know what they're into. What's their favorite color? 47 00:03:08,455 --> 00:03:11,380 What's their ideal date? There's a link in the show notes. 48 00:03:11,380 --> 00:03:14,180 Please text it to your mom, your favorite barista, and the 49 00:03:14,180 --> 00:03:17,700 captain of your rival bowling league team. Thank you. Okay. 50 00:03:18,500 --> 00:03:21,380 Time for the actual news. Jason, start the tunes. 51 00:03:24,395 --> 00:03:28,075 Keep my feet on the ground, keep my head in the clouds, 52 00:03:28,475 --> 00:03:33,755 electrified by the sound. It comes from the six stories this 53 00:03:33,755 --> 00:03:41,110 week. HBO Max is launching shorts. Max shorts will 54 00:03:41,110 --> 00:03:43,750 apparently consist of clipped segments of longer form 55 00:03:43,750 --> 00:03:46,950 entertainment available on the streamer as well as trailers and 56 00:03:46,950 --> 00:03:49,750 behind the scenes content according to a press release, 57 00:03:49,750 --> 00:03:52,870 which also reads quote, the clips for this feature are 58 00:03:52,870 --> 00:03:56,875 powered by an AI tool built in house which helps human editors 59 00:03:56,875 --> 00:04:00,475 select and verticalize content by using machine learning 60 00:04:00,475 --> 00:04:04,075 algorithms that utilize scene level metadata to comb through 61 00:04:04,075 --> 00:04:07,595 thousands of hours of titles and suggest the most compelling high 62 00:04:07,595 --> 00:04:11,850 impact scenes for discovery, end quote. If there's one thing I 63 00:04:11,850 --> 00:04:16,090 love my content to be, it is verticalized. 64 00:04:16,650 --> 00:04:18,970 This comes on the heels of similar products announced at 65 00:04:18,970 --> 00:04:22,490 other streamers like Netflix, ESPN, Paramount Plus, and Disney 66 00:04:22,490 --> 00:04:25,985 Plus, each of whom platforms like Instagram and TikTok as 67 00:04:25,985 --> 00:04:29,665 direct competitors. A media consultant told Business Insider 68 00:04:29,665 --> 00:04:33,665 in February that quote, premium streaming is plowing into this 69 00:04:33,665 --> 00:04:37,770 space because of a need to own more lean back moments in a 70 00:04:37,770 --> 00:04:41,450 viewer's day, end quote. This in my opinion represents a 71 00:04:41,450 --> 00:04:44,970 fundamental misunderstanding of why audiences turn to short form 72 00:04:44,970 --> 00:04:48,970 vertical video feeds, but maybe I'm wrong and all of you do want 73 00:04:48,970 --> 00:04:53,415 to watch a hundred one minute clips of CBS's CSI Miami on 74 00:04:53,415 --> 00:05:00,775 Paramount plus. France will be the first European country to 75 00:05:00,775 --> 00:05:04,890 ban social media for teenagers. Their parliament recently 76 00:05:04,890 --> 00:05:07,610 approved a law which will limit access to people under the age 77 00:05:07,610 --> 00:05:11,050 of 15 starting at January 2027. 78 00:05:11,290 --> 00:05:14,570 All social media users in the country will have to verify 79 00:05:14,570 --> 00:05:17,795 their age when the law goes into effect. A process which we now 80 00:05:17,795 --> 00:05:21,635 know from both Australia and The United Kingdom works perfectly 81 00:05:21,635 --> 00:05:25,715 and without any hiccups or easy circumvention. A similar law has 82 00:05:25,715 --> 00:05:28,915 been proposed though not passed in Vietnam. As currently 83 00:05:28,915 --> 00:05:31,530 written, the draft law would apply to those under the age of 84 00:05:31,530 --> 00:05:35,050 16 who would only be able to log on under parental supervision. 85 00:05:35,050 --> 00:05:38,410 Vietnam would also require platforms to identify younger 86 00:05:38,410 --> 00:05:41,450 users and moderate the content served accordingly. 87 00:05:43,005 --> 00:05:47,965 Speaker 2: Let's dive into today's word. A French phrase 88 00:05:47,965 --> 00:05:52,685 originating from the expression sacred god used to express 89 00:05:52,685 --> 00:05:57,090 surprise or disbelief. Let's say it all together. 90 00:06:01,250 --> 00:06:06,050 Mike Rugnetta: Your taxes are paying for Internet no one is 91 00:06:06,050 --> 00:06:11,125 using in Alaska, if you're American at least. ProPublica 92 00:06:11,125 --> 00:06:14,405 reports that thousands of abandoned homes on the nearly 93 00:06:14,405 --> 00:06:19,125 empty Alaskan Adoc Island have an active obsolete Internet 94 00:06:19,125 --> 00:06:22,325 connection which the United States government subsidizes at 95 00:06:22,325 --> 00:06:28,450 the cost of $300,000 a year. Subsidies continue even after 96 00:06:28,450 --> 00:06:31,570 service providers are sent to prison for tax evasion or end up 97 00:06:31,570 --> 00:06:35,330 in court for fraud. What few residents remain on the island, 98 00:06:35,330 --> 00:06:39,365 which is 1,200 miles off the coast of Anchorage, use Starlink 99 00:06:39,365 --> 00:06:43,285 for internet access as it is both cheaper and more reliable. 100 00:06:43,285 --> 00:06:46,965 Amongst Starlink's customers, the local internet utility 101 00:06:46,965 --> 00:06:47,365 company. 102 00:06:53,260 --> 00:06:57,260 Doom is our Moby Dick, says the Washington Post, which this 103 00:06:57,260 --> 00:07:00,620 month named it one of the 25 most influential pieces of 104 00:07:00,620 --> 00:07:04,620 American culture. Placed between thriller and the matrix on the 105 00:07:04,620 --> 00:07:08,445 list, Wapo writes, Doom was foundational in digital 106 00:07:08,445 --> 00:07:12,445 entertainment. A three d world viewed in first person, self 107 00:07:12,445 --> 00:07:16,445 published with no gatekeepers and no retail store as it gave 108 00:07:16,445 --> 00:07:20,365 rise to user generated content years before anyone had a name 109 00:07:20,365 --> 00:07:24,280 for it. Programmer John Carmack handed the audience the tools to 110 00:07:24,280 --> 00:07:28,520 build their own corners of hell, end quote. This comes not long 111 00:07:28,520 --> 00:07:31,400 after Doom's soundtrack was inducted into the national 112 00:07:31,400 --> 00:07:32,760 recording registry. 113 00:07:32,760 --> 00:07:35,960 The shooter was also implicated in a trove of documents dumped 114 00:07:35,960 --> 00:07:39,055 by the White House this month. Documents which purport to 115 00:07:39,055 --> 00:07:42,655 reveal the dire state of the country's election security. 116 00:07:42,815 --> 00:07:46,815 Amongst the examples of how easy it is to rig an election, a 117 00:07:46,815 --> 00:07:50,495 tweet from Defcon twenty nineteen showing the screen of a 118 00:07:50,495 --> 00:07:57,590 vote tabulating computer running doom. BitChat, the Bluetooth 119 00:07:57,590 --> 00:08:01,350 based messaging app developed by former Twitter CEO Jack Dorsey 120 00:08:01,350 --> 00:08:05,670 has been targeted by the Indian government for its role in the 121 00:08:05,670 --> 00:08:10,445 ongoing Jantar Mantar student protests. The Indian Cybercrime 122 00:08:10,445 --> 00:08:14,605 Coordination Center has demanded Microsoft remove the open source 123 00:08:14,605 --> 00:08:18,365 publicly licensed software from GitHub as it allows protesters 124 00:08:18,365 --> 00:08:22,205 to circumvent local internet blackouts implemented by the 125 00:08:22,205 --> 00:08:22,605 government. 126 00:08:22,790 --> 00:08:26,310 Quote, the application enables anonymous communication without 127 00:08:26,310 --> 00:08:30,070 mandatory user registration, phone number verification, or 128 00:08:30,070 --> 00:08:33,270 centralized logging of communications. The I c four 129 00:08:33,270 --> 00:08:35,990 wrote, the technical architecture of the application 130 00:08:36,070 --> 00:08:39,085 significantly impedes lawful interception, interception, 131 00:08:39,245 --> 00:08:43,325 attribution, and investigation by law enforcement agencies, end 132 00:08:43,325 --> 00:08:48,285 quote. The protests began alongside the CJP or cockroach 133 00:08:48,285 --> 00:08:52,045 janta party, a satirical student led group formed online in 134 00:08:52,045 --> 00:08:55,620 response to comments made by an Indian judge describing the 135 00:08:55,620 --> 00:08:59,540 unemployed social media users and activists as quote, 136 00:08:59,620 --> 00:09:04,660 parasites of society. There are youngsters like cockroaches, he 137 00:09:04,660 --> 00:09:07,725 continued. Student groups have recently demanded transparency 138 00:09:07,725 --> 00:09:10,845 and legal consequences around a number of national education 139 00:09:10,845 --> 00:09:13,885 controversies, including the mismanagement of India's 140 00:09:13,885 --> 00:09:15,405 national examinations. 141 00:09:15,485 --> 00:09:19,325 The CJP began online, but its popularity quickly grew and 142 00:09:19,325 --> 00:09:22,560 leaked into meatspace. Its organizers now regularly 143 00:09:22,560 --> 00:09:25,840 communicate with the government and police. India's minister of 144 00:09:25,840 --> 00:09:29,040 education Darmindra Pradhan recently resigned following 145 00:09:29,040 --> 00:09:33,600 CJP's demands. To be eligible for membership, the CJP says one 146 00:09:33,600 --> 00:09:38,715 must be unemployed by force, by choice, or by principle, lazy 147 00:09:38,955 --> 00:09:43,435 referring only to physical activity, chronically online at 148 00:09:43,435 --> 00:09:46,795 least eleven hours daily including bathroom breaks, and 149 00:09:46,795 --> 00:09:51,050 have the ability to rant professionally. Content must be 150 00:09:51,050 --> 00:09:54,970 sharp, honest, and point at something that matters. 151 00:09:54,970 --> 00:09:59,530 You will find your destiny among the world's greatest warriors. 152 00:10:00,090 --> 00:10:07,925 The few are the proud. And finally, social security numbers 153 00:10:08,165 --> 00:10:12,965 in my physics research? It's more likely than you think. This 154 00:10:12,965 --> 00:10:16,005 May at the forty seventh I triple e symposium on security 155 00:10:16,005 --> 00:10:20,400 and privacy, five researchers revealed that a staggering 156 00:10:20,400 --> 00:10:24,160 percentage of papers uploaded to the open access research archive 157 00:10:24,160 --> 00:10:29,200 arxiv.org include what they call hidden secrets. 158 00:10:29,360 --> 00:10:32,400 Some of these secrets may be mundane, things like comments, 159 00:10:32,735 --> 00:10:36,575 template information, to do lists, and so on, but they also 160 00:10:36,575 --> 00:10:40,415 found social security numbers, credit card info, home and work 161 00:10:40,415 --> 00:10:46,015 addresses, gossip, and more. Archive is a massively popular 162 00:10:46,015 --> 00:10:50,510 platform for serving academic preprints, papers which have not 163 00:10:50,510 --> 00:10:54,670 yet been peer reviewed, and also industry white papers by 164 00:10:54,670 --> 00:10:59,070 companies like Microsoft, Nvidia, Meta, and OpenAI. For 165 00:10:59,070 --> 00:11:02,270 this paper, researchers focused on academic preprints 166 00:11:02,270 --> 00:11:05,915 specifically and found a particular vulnerability in 167 00:11:05,915 --> 00:11:10,235 LaTeX, a type setting system used to render among other 168 00:11:10,235 --> 00:11:17,520 things math equations. 93% of submissions to archive use LaTeX 169 00:11:17,520 --> 00:11:22,720 for which you are also required to upload the source files. 170 00:11:23,600 --> 00:11:28,720 These source files often contain comments and other metadata that 171 00:11:28,720 --> 00:11:31,200 authors never intended to publish. 172 00:11:31,585 --> 00:11:35,265 Now there are programmatic methods to sanitize the files, 173 00:11:35,265 --> 00:11:39,185 including methods recommended by archive itself, but many don't 174 00:11:39,185 --> 00:11:43,185 work well or well enough. And once your paper is uploaded, 175 00:11:43,920 --> 00:11:48,480 it's not so easy to take it down. We're gonna take a quick 176 00:11:48,480 --> 00:11:51,200 break, and then I'm gonna talk to security researchers Jan 177 00:11:51,200 --> 00:11:56,665 Penekamp and Johannes Lomola about the 2,700,000 research 178 00:11:56,665 --> 00:12:01,225 papers that they analyzed, the thousands of hidden secrets that 179 00:12:01,225 --> 00:12:05,465 they found amongst them, and why it's not so easy to get those 180 00:12:05,465 --> 00:12:10,025 hidden secrets off of the archive once they're there. 181 00:12:20,340 --> 00:12:25,525 Joining me are Jan Penekamp and Johannes Lomola. Jan Penekamp is 182 00:12:25,525 --> 00:12:29,205 a postdoctoral researcher in computer science at the chair of 183 00:12:29,205 --> 00:12:32,565 communication and distributed systems at RWTH Aachen 184 00:12:32,565 --> 00:12:33,365 University. 185 00:12:33,445 --> 00:12:37,445 Since 2023, he's been the head of Comsys' security and privacy 186 00:12:37,445 --> 00:12:40,690 group, which analyzes and evaluates the security of 187 00:12:40,690 --> 00:12:44,290 communication systems across the Internet of Things, healthcare, 188 00:12:44,610 --> 00:12:47,410 and more. Jan, thank you for joining us. 189 00:12:47,410 --> 00:12:48,610 Jan Pennekamp: Thanks for having me. 190 00:12:48,690 --> 00:12:51,170 Mike Rugnetta: Johannes Lohmoller is a researcher at the 191 00:12:51,170 --> 00:12:54,655 chair of communication and distributed systems at RWTH 192 00:12:54,655 --> 00:12:58,335 Aachen University. He's a member of Comsys' security and privacy 193 00:12:58,335 --> 00:13:01,215 group, where he researches privacy preserving methods, 194 00:13:01,215 --> 00:13:04,895 particularly in the domain of health care. Johannes, thanks 195 00:13:04,895 --> 00:13:05,455 for being here. 196 00:13:05,455 --> 00:13:06,495 ohannes Lohmoller: Thanks for inviting me. 197 00:13:07,690 --> 00:13:10,890 Mike Rugnetta: So, the two of you, along with three other 198 00:13:10,890 --> 00:13:14,490 coauthors, submitted a paper in April that you then presented at 199 00:13:14,490 --> 00:13:18,650 the twenty twenty six IEEE symposium on security and 200 00:13:18,650 --> 00:13:22,785 privacy, in May. The paper is titled Hidden Secrets in the 201 00:13:22,785 --> 00:13:27,185 Discovering, Analyzing, and Preventing Unintentional 202 00:13:27,185 --> 00:13:32,225 Information Disclosure in Source Files of Scientific Preprints. 203 00:13:33,070 --> 00:13:37,070 The paper details some, really interesting and slightly 204 00:13:37,070 --> 00:13:42,270 shocking, security lapses, I guess you could say, in the 205 00:13:42,270 --> 00:13:45,630 world of scientific publishing, and we're gonna get into, some 206 00:13:45,630 --> 00:13:48,965 of the dramatic details in a moment. But I wanna start at 207 00:13:48,965 --> 00:13:53,685 sort of first principles, if you don't mind. So for our audience 208 00:13:53,685 --> 00:13:57,525 who might not be entirely familiar, could you just explain 209 00:13:57,525 --> 00:13:59,925 what archive is? 210 00:14:00,245 --> 00:14:03,365 And this is spelled a r x I v. 211 00:14:03,990 --> 00:14:07,590 Jan Pennekamp: So archive is basically a way to circumvent 212 00:14:07,590 --> 00:14:11,350 the traditional publishing processes. So in traditional 213 00:14:11,350 --> 00:14:14,310 publishing, you have peer reviews and copy editing and so 214 00:14:14,310 --> 00:14:17,875 forth and so forth. And this is often a costly time consuming 215 00:14:17,875 --> 00:14:21,475 process while in contrast, archive which is a preprint 216 00:14:21,475 --> 00:14:24,915 repository basically is something where you can submit 217 00:14:24,995 --> 00:14:29,955 papers before peer review. So it also requires you to upload your 218 00:14:29,955 --> 00:14:34,020 latest sources which are then translated in PDFs. Have and not 219 00:14:34,020 --> 00:14:38,580 only do they require the latest sources for during submission, 220 00:14:38,580 --> 00:14:41,700 but they also make them publicly accessible for each and 221 00:14:41,700 --> 00:14:45,185 everybody basically to download without any registration, 222 00:14:45,185 --> 00:14:47,105 without any checking whatsoever. 223 00:14:47,105 --> 00:14:50,785 Yeah. So everybody can not only check out the PDF, but also the 224 00:14:50,785 --> 00:14:53,185 search files of submitted papers. 225 00:14:53,505 --> 00:14:57,985 Mike Rugnetta: And LaTeX is a big part of your findings in 226 00:14:57,985 --> 00:15:01,350 this paper. Johannes, can you tell us a little bit about what 227 00:15:01,510 --> 00:15:02,950 LaTeX is? 228 00:15:02,950 --> 00:15:05,350 ohannes Lohmoller: It's basically the typesetting system 229 00:15:05,350 --> 00:15:09,990 most of academia actually uses. So you don't just go to Word or 230 00:15:09,990 --> 00:15:14,195 just to write down your paper manuscript, but you typeset it 231 00:15:14,195 --> 00:15:18,435 in LaTeX, which enables, like, more fancy math. You can 232 00:15:18,435 --> 00:15:22,275 integrate other sources as PDFs, so you get vector graphics and 233 00:15:22,275 --> 00:15:25,395 all that kind of stuff. It's typically used for publishing 234 00:15:25,395 --> 00:15:25,955 those papers. 235 00:15:26,480 --> 00:15:29,840 Mike Rugnetta: And so what you found is not just that a 236 00:15:29,840 --> 00:15:34,640 majority of papers, but most papers that had submitted 237 00:15:34,640 --> 00:15:39,680 preprints to archive had unintentionally included what 238 00:15:39,680 --> 00:15:44,575 you call secret information in various portions of the the 239 00:15:44,575 --> 00:15:45,295 document? 240 00:15:45,375 --> 00:15:47,455 ohannes Lohmoller: I would prefer calling it hidden 241 00:15:47,455 --> 00:15:51,455 information in the first place, not necessarily secret because 242 00:15:51,455 --> 00:15:55,070 how we defined is is that it's information that's somehow 243 00:15:55,070 --> 00:15:58,350 contained in those source files that didn't end up in the 244 00:15:58,670 --> 00:15:59,870 produced PDF. 245 00:16:00,350 --> 00:16:02,910 Mike Rugnetta: What constitutes hidden information? What sorts 246 00:16:02,910 --> 00:16:08,670 of things are you finding that someone should not want publicly 247 00:16:08,670 --> 00:16:09,150 available? 248 00:16:09,845 --> 00:16:12,245 ohannes Lohmoller: I think what I, for instance, wouldn't like 249 00:16:12,245 --> 00:16:16,165 to see on the Internet on my papers is I mean, we often do 250 00:16:16,165 --> 00:16:20,805 these to do items and things that we still have to address in 251 00:16:20,805 --> 00:16:24,290 the papers and comments whether this is actually true or this 252 00:16:24,290 --> 00:16:28,530 and that is missing a citation or this would need further 253 00:16:28,530 --> 00:16:32,610 evaluation and analysis, conversations between authors in 254 00:16:32,610 --> 00:16:36,370 the documents. I think for me, it feels like it's personal 255 00:16:36,370 --> 00:16:37,090 information. 256 00:16:37,090 --> 00:16:37,410 Mike Rugnetta: Sure. 257 00:16:37,895 --> 00:16:40,295 ohannes Lohmoller: And then on the other end, we also found, 258 00:16:40,295 --> 00:16:43,975 like, clear personal information, like names of study 259 00:16:43,975 --> 00:16:48,215 participants that are not among the authors, but, well, like, of 260 00:16:48,215 --> 00:16:52,695 the research. We also found credentials in their credit card 261 00:16:52,695 --> 00:16:56,210 numbers, which was quite a surprise Wow. And things like 262 00:16:56,210 --> 00:16:56,690 that. 263 00:16:56,930 --> 00:16:59,810 Jan Pennekamp: And, obviously, this does not affect all of the 264 00:16:59,810 --> 00:17:04,610 2,700,000 papers that we analyzed. But still, if you also 265 00:17:04,610 --> 00:17:08,050 like 100 papers on there and you make, like, one mean remark 266 00:17:08,050 --> 00:17:12,565 about a colleague or, I don't know, basically reveal, for 267 00:17:12,565 --> 00:17:16,325 example, that you conduct scientific misconduct, those 268 00:17:16,325 --> 00:17:20,165 kind of things. This can also be very damaging to your career, 269 00:17:20,165 --> 00:17:23,010 reputation, and so forth. So, you know, it's just sufficient 270 00:17:23,010 --> 00:17:25,250 to have this one bit of information there. Yeah. 271 00:17:25,250 --> 00:17:28,450 Or alternatively, if you think about cyberstalking. Right? If 272 00:17:28,450 --> 00:17:31,010 you leave leave your address in one of the papers, it doesn't 273 00:17:31,010 --> 00:17:34,770 matter that it's that you have multiple papers on this website, 274 00:17:34,770 --> 00:17:39,795 for example. Yeah. We also found sort of some remarks on the 275 00:17:39,795 --> 00:17:43,475 Internet that indicate that also, like, these large LLM 276 00:17:43,475 --> 00:17:47,715 providers may be scraping the source files files this way. 277 00:17:47,715 --> 00:17:50,780 So they're not constraining themselves to the PDFs, but 278 00:17:50,780 --> 00:17:53,980 actually also in the training data, those all this information 279 00:17:53,980 --> 00:17:59,100 ends up there. And and and part of doing that, also the hidden 280 00:17:59,100 --> 00:18:02,060 information that Johannes was just talking about is also 281 00:18:02,060 --> 00:18:04,655 basically revealed or used by these companies. 282 00:18:04,655 --> 00:18:07,535 Mike Rugnetta: So there is there is evidence that say, a bot 283 00:18:07,535 --> 00:18:12,415 which is scouring the Internet for information related to an 284 00:18:12,415 --> 00:18:17,055 LLM, like, say, potentially an OpenAI or an Anthropic is 285 00:18:17,530 --> 00:18:21,770 gathering some of this hidden information. It whether or not 286 00:18:21,770 --> 00:18:24,570 it makes its way into a model that then people have access to, 287 00:18:24,570 --> 00:18:27,210 big question mark, but they've at least there's evidence that 288 00:18:27,210 --> 00:18:28,090 they've accessed it. 289 00:18:28,090 --> 00:18:30,010 Jan Pennekamp: Yes. To the best of our understanding. Yes. 290 00:18:30,010 --> 00:18:30,410 Interesting. 291 00:18:30,915 --> 00:18:33,635 Mike Rugnetta: I mean, does that mean that archive has to change 292 00:18:33,635 --> 00:18:34,755 the way that it works? 293 00:18:35,075 --> 00:18:37,795 Jan Pennekamp: I guess it's it's a matter of perspective again. 294 00:18:37,795 --> 00:18:41,475 Right? If you consider disinformation to not be that 295 00:18:41,475 --> 00:18:45,950 bad Sure. Then archive doesn't have to change anything because, 296 00:18:46,270 --> 00:18:49,390 I mean, it's it's just conversational content on there. 297 00:18:49,390 --> 00:18:49,870 Right? 298 00:18:50,030 --> 00:18:53,230 We also reached out to them and basically they told us, like, 299 00:18:53,230 --> 00:18:56,030 they do not consider this sort of information that we 300 00:18:56,030 --> 00:19:00,485 discovered to be sensitive so that they need to act upon it. 301 00:19:00,485 --> 00:19:03,445 So, basically, as they refrain from doing something, they told 302 00:19:03,445 --> 00:19:06,885 us, alright. The authors are being made aware that this 303 00:19:06,885 --> 00:19:10,980 information ends up there and we have clear reasons why we want 304 00:19:10,980 --> 00:19:15,540 to include this information in the dataset or the repository, 305 00:19:15,540 --> 00:19:18,660 basically. It's like if you go to Google and you don't like how 306 00:19:18,660 --> 00:19:20,100 they display the search results, 307 00:19:20,100 --> 00:19:20,740 Mike Rugnetta: well Sure. 308 00:19:20,900 --> 00:19:22,740 Jan Pennekamp: Tough luck. You don't have to use the service, 309 00:19:22,740 --> 00:19:25,540 but in the end, it's how they want to operate. Right? 310 00:19:26,135 --> 00:19:28,455 Mike Rugnetta: And do they feel this way even knowing that, say 311 00:19:28,615 --> 00:19:31,175 like, I'm looking at the list of some of the the things that you 312 00:19:31,175 --> 00:19:37,015 found, things like Slack tokens, API keys, Facebook OAuths, and 313 00:19:37,015 --> 00:19:39,175 they're like, yeah. That's fine. Like, if someone wanted to put 314 00:19:39,175 --> 00:19:39,975 that up there, sure. 315 00:19:40,260 --> 00:19:43,140 Jan Pennekamp: I mean, for those things, you can still revocate 316 00:19:43,140 --> 00:19:44,100 the credentials in 317 00:19:44,100 --> 00:19:44,420 ohannes Lohmoller: a way. 318 00:19:44,420 --> 00:19:45,220 Mike Rugnetta: True. Yeah. 319 00:19:45,220 --> 00:19:48,660 Jan Pennekamp: But, yeah, they they think it's the educator's 320 00:19:48,660 --> 00:19:53,725 author's responsibility. Yeah. Mo multiple people, after 321 00:19:53,725 --> 00:19:56,525 presenting the work or having it published multiple also reached 322 00:19:56,525 --> 00:20:00,765 out to us and they asked like, why does archive even surface 323 00:20:00,765 --> 00:20:04,125 content like these sources publicly? Like, they can just, 324 00:20:04,125 --> 00:20:08,400 you know, keep them internally, and not share them. So this is 325 00:20:08,400 --> 00:20:11,200 like the most common response that we're getting, I would say. 326 00:20:11,200 --> 00:20:13,920 Mike Rugnetta: They don't have to serve the metadata that 327 00:20:13,920 --> 00:20:18,080 reveals hold on. I'm looking at it here. Almost 4,000 United 328 00:20:18,080 --> 00:20:19,760 States Social Security numbers. 329 00:20:19,760 --> 00:20:22,355 ohannes Lohmoller: I mean, they argue it's for reproducibility, 330 00:20:22,355 --> 00:20:26,035 and I think they had some point several years ago where authors 331 00:20:26,035 --> 00:20:30,595 actually could upload ancillary data right into the submission. 332 00:20:30,595 --> 00:20:34,530 So that's an actually intended way of, like, providing further 333 00:20:34,530 --> 00:20:38,770 context or data and things like that next to the submission. 334 00:20:38,930 --> 00:20:42,130 Uh-huh. But I think there are other large research data 335 00:20:42,130 --> 00:20:45,650 repositories nowadays, like, say, Nodo, for instance, that 336 00:20:45,650 --> 00:20:49,615 kind of fulfill the service. So I don't think it's, like, 337 00:20:49,695 --> 00:20:52,575 necessary anymore to provide Lartic sources. 338 00:20:52,815 --> 00:20:54,895 Mike Rugnetta: You also mentioned that there is a way 339 00:20:54,895 --> 00:20:59,775 that this is a concern for the peer review process. Can you 340 00:20:59,775 --> 00:21:02,335 just briefly describe why that might be? 341 00:21:02,930 --> 00:21:07,250 Jan Pennekamp: In several disciplines, you do double blind 342 00:21:07,250 --> 00:21:10,370 review, which basically means that the reviewer does not know 343 00:21:10,370 --> 00:21:14,610 the author and the author does not know the reviewer. Now the 344 00:21:14,610 --> 00:21:18,385 reviewer does not get disclosed because you submit sources, but 345 00:21:18,385 --> 00:21:22,305 the author may get dis disclosed this way because you may not 346 00:21:22,305 --> 00:21:25,985 only have anonymous authors in there but other things as well. 347 00:21:26,305 --> 00:21:29,825 Similarly, if later on publishers collect the Lattice 348 00:21:29,970 --> 00:21:33,490 sources, they basically have also the same amount of access 349 00:21:33,810 --> 00:21:37,570 to the hidden information. Right? So while it is not 350 00:21:37,730 --> 00:21:40,770 publicly accessible, it certainly is for those 351 00:21:40,770 --> 00:21:43,595 publishing companies where you could say, right. 352 00:21:43,595 --> 00:21:46,235 I'm trusting them, but, you know, security cyber incidents 353 00:21:46,235 --> 00:21:49,435 happen across the board and depending on where where and how 354 00:21:49,435 --> 00:21:54,075 much they store, yeah, malicious actors could still get access to 355 00:21:54,075 --> 00:21:57,750 that after the fact. So it's it's also still crucial to 356 00:21:57,750 --> 00:21:59,510 consider that angle in a way. 357 00:21:59,510 --> 00:22:03,350 Mike Rugnetta: Right? Got it. And how is it that those things 358 00:22:03,350 --> 00:22:08,550 that aren't needed for the final rendering end up being left in? 359 00:22:08,550 --> 00:22:11,685 Is this just people don't know that they should remove them? Is 360 00:22:11,685 --> 00:22:15,365 this a problem with the the software itself when you submit 361 00:22:15,365 --> 00:22:16,165 the paper? 362 00:22:16,325 --> 00:22:19,685 Jan Pennekamp: I think it's a little mix of all all things you 363 00:22:19,685 --> 00:22:24,750 couldn't can think of. Yeah. So probably people are lazy. They 364 00:22:24,750 --> 00:22:27,710 don't really care. Sometimes they don't even know that the 365 00:22:27,710 --> 00:22:30,990 files end up being publicly accessible and they just upload 366 00:22:30,990 --> 00:22:31,470 that. 367 00:22:32,110 --> 00:22:36,430 Then archive actually lets you nowadays select whether files 368 00:22:36,430 --> 00:22:41,145 are required or not, but only on on root level basically. So they 369 00:22:41,145 --> 00:22:46,265 know about the issue. They sort of try to address it but I'm I'm 370 00:22:46,265 --> 00:22:50,825 not sure how many people would actually select or deselect 371 00:22:50,905 --> 00:22:51,465 stuff there. 372 00:22:51,690 --> 00:22:52,330 Mike Rugnetta: Sure. 373 00:22:52,650 --> 00:22:55,290 Jan Pennekamp: Yeah. And then for like if you have a big 374 00:22:55,290 --> 00:22:59,370 project with hundreds of files, you can also just get lost and 375 00:22:59,370 --> 00:23:04,245 not noticing that maybe you have 102 files in there. So it's 376 00:23:04,245 --> 00:23:08,325 sometimes also just a matter of scale why why things are popping 377 00:23:08,325 --> 00:23:08,965 up there. 378 00:23:08,965 --> 00:23:12,725 Mike Rugnetta: Yeah. It's such a particularly and I think 379 00:23:12,725 --> 00:23:18,085 uniquely specific example of all the ways that we leave data 380 00:23:18,085 --> 00:23:22,920 trails behind us when we use the Internet and various platforms 381 00:23:22,920 --> 00:23:26,360 on the Internet. And like, you know, not just the Internet, 382 00:23:26,360 --> 00:23:29,400 guess, also just work you do with computers, you know, you 383 00:23:29,400 --> 00:23:33,160 you end up leaving little bits of yourself and identifying 384 00:23:33,160 --> 00:23:36,915 pieces of information about your life all around. And I think 385 00:23:36,915 --> 00:23:41,155 like, there's all of these scenarios where we conceive of 386 00:23:41,155 --> 00:23:45,395 information as private, or we think of files as being clean, 387 00:23:45,395 --> 00:23:49,395 or not vectors for people learning sensitive things about 388 00:23:49,395 --> 00:23:53,290 us. For maybe a bunch of, like, arbitrary reasons, we think that 389 00:23:53,290 --> 00:23:56,250 things are safe, or maybe for no reason. 390 00:23:56,250 --> 00:23:59,690 We just have haven't had a reason to consider otherwise. 391 00:24:00,010 --> 00:24:01,530 But actually, like, those things are really public. 392 00:24:02,055 --> 00:24:04,215 Jan Pennekamp: Yeah. I believe it's like if you think about 393 00:24:04,215 --> 00:24:07,495 electric vehicles and kind of what information they record and 394 00:24:07,495 --> 00:24:10,855 sometimes depending on which third party apps you're using, 395 00:24:10,855 --> 00:24:14,535 maybe they also reveal stuff on the Internet. I mean, there are 396 00:24:14,535 --> 00:24:19,110 these these apps for fitness tracking, right, where basically 397 00:24:19,110 --> 00:24:22,230 your history is publicly available. So I think, you know, 398 00:24:22,230 --> 00:24:26,950 it it comes up every year and now. And so this was basically 399 00:24:27,590 --> 00:24:31,395 one of the occurrences where we as academia, people from 400 00:24:31,395 --> 00:24:35,155 academia basically spot, something sensitive in a way. 401 00:24:35,155 --> 00:24:37,955 Mike Rugnetta: Yeah. You talk about some of the ways that 402 00:24:37,955 --> 00:24:41,315 exist for dealing with this issue because this isn't a new 403 00:24:41,315 --> 00:24:46,900 problem. And in fact, archive basically recommends the use of 404 00:24:46,900 --> 00:24:53,060 a a Google tool, I think it is, to sanitize the things that are 405 00:24:53,060 --> 00:24:56,420 submitted. Could you talk a little bit about that tool and 406 00:24:56,420 --> 00:24:59,085 others like it and why they're they aren't sufficient for 407 00:24:59,085 --> 00:25:00,125 solving this problem? 408 00:25:00,525 --> 00:25:03,805 Jan Pennekamp: So Lattice is a very powerful language. So you 409 00:25:03,805 --> 00:25:09,485 have quite some diverse ways of inserting comments in there or 410 00:25:09,485 --> 00:25:14,630 basically adding stuff that you do not need, in the final PDF. 411 00:25:15,110 --> 00:25:20,870 And basically what we did is survey how prior solutions 412 00:25:20,870 --> 00:25:24,070 basically work and what they did and most of them use regular 413 00:25:24,070 --> 00:25:28,615 expressions to detect a comment and then act upon that. I would 414 00:25:28,615 --> 00:25:31,655 say like the Google tool is actually ArchiveLatigleena that 415 00:25:31,655 --> 00:25:35,495 is actually a quite mature tool but if you use some ways that 416 00:25:35,495 --> 00:25:39,175 are not supported by this tool then, well, you're all of luck 417 00:25:39,175 --> 00:25:44,480 and this information basically remains in there. And we argue 418 00:25:44,480 --> 00:25:47,680 that it's not sufficient to rely on any of those. 419 00:25:47,680 --> 00:25:50,560 And what's what strike us as interesting, none of them 420 00:25:50,560 --> 00:25:54,800 actually consider this metadata level, even though, you know, 421 00:25:54,800 --> 00:25:58,165 it's quite well known that metadata actually leaks or can 422 00:25:58,165 --> 00:25:59,605 leak sensitive information. 423 00:25:59,685 --> 00:26:02,085 ohannes Lohmoller: Now we were sitting on these 2,700,000 424 00:26:02,085 --> 00:26:05,605 submissions and had this archive Latest Kiena tool floating 425 00:26:05,605 --> 00:26:09,365 around. So we just, like, sew it on all of those submissions and 426 00:26:09,365 --> 00:26:13,260 kind of figured out what happens. And then we found out 427 00:26:13,260 --> 00:26:18,380 that, like, I don't know, nine above 90% have modified sources 428 00:26:18,380 --> 00:26:22,140 after applying the tool. So we're not quite sure whether the 429 00:26:22,140 --> 00:26:25,900 tool is more major nowadays or whether it was never applied to 430 00:26:25,900 --> 00:26:29,505 the sources, But we kind of had an empirical way of figuring 431 00:26:29,505 --> 00:26:34,065 out, like, how much additional content those tools would I 432 00:26:34,065 --> 00:26:35,745 could remove from the submission. 433 00:26:35,745 --> 00:26:37,985 Jan Pennekamp: It would not probably remove all the data 434 00:26:37,985 --> 00:26:41,745 that we discovered, but it would already remove quite a lot by 435 00:26:41,745 --> 00:26:42,145 margin. 436 00:26:42,840 --> 00:26:46,200 Mike Rugnetta: And did you you developed your own tool in order 437 00:26:46,200 --> 00:26:48,360 to sanitize submissions. Is that true? 438 00:26:48,920 --> 00:26:52,200 Jan Pennekamp: Yes. So we were fortunate enough that one of our 439 00:26:52,600 --> 00:26:56,435 co authors, a student that we supervised, that he came across 440 00:26:56,435 --> 00:27:00,115 a different way of basically working with these LATHE files. 441 00:27:00,115 --> 00:27:04,035 So instead of using regular expressions, he discovered 442 00:27:04,515 --> 00:27:09,530 TreeSitter which basically is an abstract grammar that sort of 443 00:27:09,530 --> 00:27:13,530 better reads or processes the LATH files. We're not saying 444 00:27:13,530 --> 00:27:18,330 that our tool is perfect in any way. It's just, let's say, the 445 00:27:18,330 --> 00:27:21,690 next step forward and showing, like, where there's potential 446 00:27:22,085 --> 00:27:25,605 and how to to make the best out of an unfortunate situation, I 447 00:27:25,605 --> 00:27:26,245 would say. 448 00:27:26,245 --> 00:27:30,965 Yeah. Also, sort of looking in the future to enable many many 449 00:27:30,965 --> 00:27:33,125 different authors and researchers, not only in 450 00:27:33,125 --> 00:27:36,965 computer science, but basically, yeah, across all the disciplines 451 00:27:37,340 --> 00:27:41,980 to apply that. And that's also why we sort of focused on really 452 00:27:41,980 --> 00:27:45,340 reliability. So we wanted to make sure if you apply our tool 453 00:27:45,340 --> 00:27:49,660 that then the information gets reliably removed and that it can 454 00:27:49,660 --> 00:27:55,055 be sure that sort of the content is retained as it was so that 455 00:27:55,055 --> 00:27:58,095 we're not over removing stuff. So, basically, altering the PDF 456 00:27:58,095 --> 00:28:03,535 because we found, several cases where the Google tool over 457 00:28:03,535 --> 00:28:04,495 removes content. 458 00:28:04,495 --> 00:28:07,950 So in a way, you use your tool and then your the content that 459 00:28:07,950 --> 00:28:10,510 you have in the paper is not identical anymore. And 460 00:28:10,510 --> 00:28:13,550 unfortunately, you may not notice unless you check really 461 00:28:13,550 --> 00:28:16,670 character basically in your paper. Now this is something 462 00:28:16,670 --> 00:28:19,705 that can be quite damaging. Right? If you have a proof on 463 00:28:19,705 --> 00:28:22,265 there and then for some reason there are like a couple of 464 00:28:22,265 --> 00:28:25,225 characters missing in there or a mass formula and you you don't 465 00:28:25,225 --> 00:28:27,785 know why that that doesn't work. 466 00:28:27,785 --> 00:28:30,665 It doesn't make sense. It's just not equal anymore. So, you know, 467 00:28:30,890 --> 00:28:34,810 it it also has an impact on sort of the scientific discourse and 468 00:28:34,810 --> 00:28:36,970 it's, yeah, in general. 469 00:28:36,970 --> 00:28:39,210 Mike Rugnetta: Yeah. Do you think the archive is gonna 470 00:28:39,210 --> 00:28:42,410 switch their recommendation from the Google tool to yours? 471 00:28:42,490 --> 00:28:45,605 Jan Pennekamp: I mean, we're not Google, so I I mean, I I don't 472 00:28:45,605 --> 00:28:49,205 know. Let's see. If it gains sufficient traction, maybe. 473 00:28:49,605 --> 00:28:52,005 Mike Rugnetta: So I'm curious what what do you think should 474 00:28:52,005 --> 00:28:55,845 happen next and what in theory actually happens next? You 475 00:28:55,845 --> 00:28:59,125 mentioned that, you know, archive has been pretty clear 476 00:28:59,125 --> 00:29:03,340 that, like, they they think this is the system working, except in 477 00:29:03,340 --> 00:29:07,020 extenuating circumstances where the information is very personal 478 00:29:07,020 --> 00:29:10,940 or, you know, very important that it not be made publicly. So 479 00:29:10,940 --> 00:29:14,705 it sounds like, you know, them changing the way that their 480 00:29:14,705 --> 00:29:19,665 service works is remote. Is there now, like, an education 481 00:29:19,665 --> 00:29:22,065 campaign that has to be undertaken in order to let 482 00:29:22,065 --> 00:29:24,385 authors know that when they're submitting things, this is 483 00:29:24,385 --> 00:29:25,745 something that they have to be thinking about? 484 00:29:26,410 --> 00:29:30,250 Jan Pennekamp: Yes. Certainly. So I think, like, policy wise or 485 00:29:30,250 --> 00:29:33,850 institutional guidelines on on how to work with archive is 486 00:29:33,850 --> 00:29:37,610 something that's essential. So nobody ever, like, really told 487 00:29:37,610 --> 00:29:40,010 me, like, how to work with archive. Right? 488 00:29:40,185 --> 00:29:43,145 So at some point, it just came up and you're like, alright. I'm 489 00:29:43,145 --> 00:29:46,905 just submitting my, first paper to archive. What can go wrong? 490 00:29:46,905 --> 00:29:50,025 Right? But now that we know, like, a lot of things can go 491 00:29:50,025 --> 00:29:52,905 wrong and, actually, you can't undo all of the things that can 492 00:29:52,905 --> 00:29:53,545 go wrong. 493 00:29:53,705 --> 00:29:57,700 So I I could imagine that funding agencies who basically 494 00:29:57,700 --> 00:30:02,500 mandate that papers are getting open access publishing also have 495 00:30:02,500 --> 00:30:05,940 sort of guidelines on on what they want to see up on archive 496 00:30:05,940 --> 00:30:09,060 and whatnot. Yeah. And similarly, the institutions 497 00:30:09,060 --> 00:30:13,945 themselves maybe want to be sure that authors do not basically 498 00:30:13,945 --> 00:30:17,145 reveal damaging information. Right? They also have a 499 00:30:17,785 --> 00:30:22,985 reputational desire to basically improve that situation. 500 00:30:22,985 --> 00:30:26,600 If if they were were to come up with that, I mean, most of the 501 00:30:26,600 --> 00:30:29,800 major universities have some kind of library service that 502 00:30:29,800 --> 00:30:33,720 handle all those things. So I think, certainly, the the staff 503 00:30:33,720 --> 00:30:38,280 and the way to implement those is there. It just hasn't been 504 00:30:38,280 --> 00:30:42,355 done yet. Right? So getting that on track, I think this is very 505 00:30:42,355 --> 00:30:42,995 crucial. 506 00:30:43,315 --> 00:30:47,635 And it is also something that would not only help, like, 507 00:30:47,635 --> 00:30:50,915 educated computer science or computer security researchers, 508 00:30:51,170 --> 00:30:54,050 but, basically, help people across all the disciplines 509 00:30:54,050 --> 00:30:55,970 because then you have guidelines, something to 510 00:30:55,970 --> 00:31:00,210 follow-up on. And even if it just says, you know, be careful. 511 00:31:00,210 --> 00:31:03,810 You like, use caution, maybe reach out to an expert, somebody 512 00:31:03,810 --> 00:31:07,115 who supports you on basically submitting that, that would be 513 00:31:07,115 --> 00:31:09,675 something to to really go a long way with this. 514 00:31:09,675 --> 00:31:11,515 ohannes Lohmoller: I think we should make this a shared 515 00:31:11,515 --> 00:31:15,755 responsibility between authors and publishers because even if 516 00:31:15,755 --> 00:31:18,875 we get the tooling and the awareness right, at least the 517 00:31:19,150 --> 00:31:24,030 time issues, so deadline, pressure, and so remains. And 518 00:31:24,190 --> 00:31:27,230 since we have seen at least some of those aspects could be easy 519 00:31:27,390 --> 00:31:33,685 easily fixable on archive sites, I wouldn't just, like, redeem or 520 00:31:33,685 --> 00:31:37,045 exempt the authors from their responsibility, but something 521 00:31:37,045 --> 00:31:39,765 could also be done at the publisher's side. 522 00:31:39,845 --> 00:31:42,725 Jan Pennekamp: In the end, it's a community effort to basically 523 00:31:42,725 --> 00:31:43,605 improve everything. 524 00:31:43,930 --> 00:31:48,570 Mike Rugnetta: Yeah. Jan, Johannes, thank you so much for 525 00:31:48,570 --> 00:31:50,890 joining us. I really appreciate you taking the time out of your 526 00:31:50,890 --> 00:31:52,970 important and busy schedules to chat with me. 527 00:31:52,970 --> 00:31:53,770 ohannes Lohmoller: Thanks for having us. 528 00:31:53,770 --> 00:31:55,050 Jan Pennekamp: It was a pleasure being here. 529 00:31:55,210 --> 00:31:57,850 Mike Rugnetta: What is the best place for folks to go to be able 530 00:31:57,850 --> 00:31:59,595 to keep tabs on what you're up to? 531 00:31:59,595 --> 00:32:02,635 Jan Pennekamp: This paper is already up on archive naturally, 532 00:32:02,635 --> 00:32:06,475 including sources. K. So you can also look around what we did in 533 00:32:06,475 --> 00:32:06,955 there. 534 00:32:07,115 --> 00:32:09,035 Mike Rugnetta: Including including, I assume, your credit 535 00:32:09,035 --> 00:32:10,235 card number and home address. 536 00:32:10,670 --> 00:32:12,990 Jan Pennekamp: Sure. You you can take a look if you find it and 537 00:32:12,990 --> 00:32:18,350 then let me know. I I will look at my bank statement to see. And 538 00:32:18,350 --> 00:32:23,150 we do also host a website, which is archive.com. It's 539 00:32:23,150 --> 00:32:28,995 .rwth-ahn.de, quite a long name, which basically gives you an 540 00:32:28,995 --> 00:32:32,275 executive summary and so any updates that probably would 541 00:32:32,275 --> 00:32:34,595 follow after this work, we would also host there. 542 00:32:34,595 --> 00:32:37,075 So this really gives you a high level overview about the key 543 00:32:37,075 --> 00:32:37,875 findings that we have. 544 00:32:39,560 --> 00:32:42,520 Mike Rugnetta: Thanks again to Jan and Johannes. If you would 545 00:32:42,520 --> 00:32:45,480 like to read their paper or see more work from them or their 546 00:32:45,480 --> 00:33:09,110 department, you can check the links in the show notes. That is 547 00:33:09,110 --> 00:33:11,270 the news we have for you this week. We are gonna be back here 548 00:33:11,270 --> 00:33:15,430 in the main feed on or around Thursday, August 6. In a couple 549 00:33:15,430 --> 00:33:18,150 short weeks, the Never Post Member Drive stream week will 550 00:33:18,150 --> 00:33:21,715 begin during which we will endeavor to add to the ranks of 551 00:33:21,715 --> 00:33:25,075 wonderful people who support this show, help keep it running, 552 00:33:25,075 --> 00:33:28,675 and maybe most importantly, it get better and weirder and more 553 00:33:28,675 --> 00:33:29,395 interesting. 554 00:33:29,555 --> 00:33:33,075 Do you wanna be one of those weird wonderful people and help 555 00:33:33,075 --> 00:33:39,340 us get off to a smashing start? Can you help us log on Twitch on 556 00:33:39,340 --> 00:33:44,940 Monday, 08/17/2026 at 11AM eastern and say with massive 557 00:33:44,940 --> 00:33:49,065 giggle stifling grins on our faces that we are already well 558 00:33:49,065 --> 00:33:54,985 on our way to our membership goals for the year. Would you do 559 00:33:54,985 --> 00:34:00,185 that for the low low price of $4? If so, please head to 560 00:34:00,185 --> 00:34:03,820 neverpo.st to become a member. You can get access to an ad free 561 00:34:03,820 --> 00:34:08,700 feed of the show, monthly at least members only uploads and 562 00:34:08,700 --> 00:34:09,260 more. 563 00:34:09,260 --> 00:34:13,660 Neverpo.st, but you knew that already. Never Post's producers 564 00:34:13,660 --> 00:34:16,220 are Audrey Evans, Georgia Hampton, and The Mysterious. 565 00:34:16,220 --> 00:34:19,075 Doctor first name, last name. Our senior producer is Hans 566 00:34:19,075 --> 00:34:22,835 Buto. Our executive producer is Jason Oberholzer, and the show's 567 00:34:22,835 --> 00:34:24,915 host, that's me, is Mike Rugnetta. 568 00:34:33,940 --> 00:34:38,340 If a shaft of light is broken into a thousand pieces and each 569 00:34:38,340 --> 00:34:42,260 piece into a thousand more, might not a stranger mistake any 570 00:34:42,260 --> 00:34:46,835 of the fragments for the original light? Am I not such a 571 00:34:46,835 --> 00:34:51,395 stranger? And if so, how can I speak of light or its origins? 572 00:34:51,715 --> 00:34:55,235 What I have taken for light may be something entirely different. 573 00:34:55,635 --> 00:34:59,715 Blood or mire or darkness itself. 574 00:35:00,890 --> 00:35:04,090 Excerpt of the interior prisoner, translation of an 575 00:35:04,090 --> 00:35:08,650 unsigned manuscript written at Salamanca around nineteen o two 576 00:35:08,650 --> 00:35:13,690 by Jeffrey O'Brien. Never Post is a production of Charts and 577 00:35:13,690 --> 00:35:16,845 Leisure and it's distributed by Radiotopia.