1 00:00:00,920 --> 00:00:05,040 S1: Unsupervised Learning is a podcast about trends and ideas in cybersecurity, 2 00:00:05,160 --> 00:00:10,000 S1: national security, AI, technology and society, and how best to 3 00:00:10,039 --> 00:00:17,640 S1: upgrade ourselves to be ready for what's coming. All right. 4 00:00:17,640 --> 00:00:21,400 S1: Welcome to unsupervised Learning. This is Daniel. All right. Episode 465. 5 00:00:21,400 --> 00:00:23,960 S1: What do we have? So I spent like 20 hours 6 00:00:23,960 --> 00:00:29,560 S1: over the past few weeks updating my homepage about page projects, Telos, 7 00:00:30,080 --> 00:00:34,440 S1: a whole bunch of different stuff. Uh, predictions page, just 8 00:00:34,440 --> 00:00:36,559 S1: a lot of new content on the website, which is 9 00:00:36,560 --> 00:00:42,480 S1: a completely new looking website. Basically, the website was all 10 00:00:42,479 --> 00:00:45,919 S1: on beehive, and now it's back in its natural home 11 00:00:46,400 --> 00:00:53,760 S1: at regular. Daniel Mysa.com and the newsletters on newsletter.com. So yeah, 12 00:00:53,800 --> 00:00:58,520 S1: go check that out. Highly recommend checking out two podcast episodes. 13 00:00:58,520 --> 00:01:03,840 S1: I've been absolutely loving Dan Carlin on Alexander the Great. I'm. Yeah. 14 00:01:03,840 --> 00:01:05,800 S1: I just love it. I'm so happy that he's back. 15 00:01:05,800 --> 00:01:07,600 S1: I thought he actually got out of the game or 16 00:01:07,600 --> 00:01:10,399 S1: retired or something, but this thing has from like June 17 00:01:10,440 --> 00:01:15,000 S1: of 2024, so it's good news that he's back. I 18 00:01:15,040 --> 00:01:18,640 S1: absolutely love getting history in this sort of way. He's 19 00:01:18,640 --> 00:01:22,600 S1: describing history. It's also mostly biography based. I just love 20 00:01:22,600 --> 00:01:25,360 S1: the whole thing. It's great. And I just listened to 21 00:01:25,360 --> 00:01:29,720 S1: the acquired episode on TSMC and it made me buy 22 00:01:29,720 --> 00:01:33,039 S1: more of their stock. I tend to buy stock from 23 00:01:33,240 --> 00:01:40,679 S1: companies that they talk about, um, namely Costco and now TSMC. But, uh, yeah, really, 24 00:01:40,680 --> 00:01:44,920 S1: really good episode. And my buddy Joseph Thacker has gone independent. 25 00:01:44,920 --> 00:01:47,720 S1: So happy for him about this. He's now a full 26 00:01:47,760 --> 00:01:51,920 S1: time bug bounty hacker and entrepreneur, and I really can't 27 00:01:51,920 --> 00:01:56,800 S1: wait to see what he does this year in 25 security. 28 00:01:56,800 --> 00:02:00,160 S1: So new research shows that the average employee creates a 29 00:02:00,160 --> 00:02:03,800 S1: new SAS account every two Two weeks, which is creating 30 00:02:03,800 --> 00:02:06,920 S1: massive security blind spots for companies. If I click on 31 00:02:06,920 --> 00:02:12,320 S1: this link, um, yeah, it's pretty bad. It's pretty bad. Uh, 32 00:02:12,360 --> 00:02:16,519 S1: definitely recommend checking this out if you are into app security, 33 00:02:16,560 --> 00:02:20,160 S1: enterprise security, that kind of stuff. Uh, a lot of 34 00:02:20,160 --> 00:02:23,720 S1: this data comes from the new, um, Verizon Dbir report 35 00:02:24,080 --> 00:02:28,080 S1: found web applications were involved in 50% of security incidents, 36 00:02:28,080 --> 00:02:33,080 S1: and 80% of breaches now involve compromised SaaS credentials. And 37 00:02:33,080 --> 00:02:36,280 S1: that's according to CrowdStrike. Google just shared how they handle 38 00:02:36,280 --> 00:02:39,480 S1: threat detection at a massive scale. They're doing some wild 39 00:02:39,480 --> 00:02:45,200 S1: stuff around automation and response time. So standard response time 40 00:02:45,200 --> 00:02:48,040 S1: is like days or weeks. And they're now doing it 41 00:02:48,040 --> 00:02:53,440 S1: in just hours. And they're automating automating 97% of their 42 00:02:53,440 --> 00:02:56,839 S1: detection work. So you got to imagine how much that 43 00:02:56,840 --> 00:02:59,919 S1: actually is at Google. Microsoft's AI Red team published their 44 00:02:59,919 --> 00:03:03,920 S1: findings from attacking over 100 AI I products. Sharing key 45 00:03:03,960 --> 00:03:08,240 S1: lessons for identifying risks and vulnerabilities. Most interesting insight is 46 00:03:08,240 --> 00:03:12,399 S1: that basic techniques like prompt injection usually work better than 47 00:03:12,400 --> 00:03:16,320 S1: complex attacks. It makes sense. I mean, it is the 48 00:03:16,320 --> 00:03:19,480 S1: main attack. In my opinion. Prompt injection is the thing 49 00:03:19,480 --> 00:03:23,559 S1: that kind of makes this different than other types of security, 50 00:03:23,800 --> 00:03:27,400 S1: because otherwise we're mostly talking about API security. When we're 51 00:03:27,400 --> 00:03:31,720 S1: talking about AI security, it's threat modeling the entire data 52 00:03:31,720 --> 00:03:35,560 S1: flow from the ingress point all the way to the 53 00:03:35,560 --> 00:03:38,680 S1: different places it could be stored for, like second order attacks. 54 00:03:38,880 --> 00:03:41,720 S1: And all of that is kind of similar. The thing 55 00:03:41,720 --> 00:03:45,080 S1: that makes it different is like English being the language 56 00:03:45,200 --> 00:03:48,360 S1: of attack and the fact that you could trick a 57 00:03:48,400 --> 00:03:54,240 S1: Semi-intelligent processor, right? So parsers have always been vulnerable, but 58 00:03:54,240 --> 00:03:58,520 S1: it's even more vulnerable now because you can trick it, right? 59 00:03:58,720 --> 00:04:00,680 S1: It's a different kind of tricking. You could trick it 60 00:04:00,680 --> 00:04:04,350 S1: using logic and stories and stuff like that. You could 61 00:04:04,350 --> 00:04:07,270 S1: always trick passers, and that's kind of the game in security. 62 00:04:07,630 --> 00:04:11,230 S1: But the types of attacks that you can do now are, 63 00:04:11,230 --> 00:04:15,350 S1: are broader and more powerful with prompt injection. So I 64 00:04:15,350 --> 00:04:19,070 S1: think that's the main avenue which which seems like it's 65 00:04:19,070 --> 00:04:21,870 S1: what they found as well. Basic techniques like prompt injection 66 00:04:21,870 --> 00:04:25,230 S1: work even better than complex attacks. Not quite the same point, 67 00:04:25,230 --> 00:04:28,430 S1: but related. And DJI just announced that they're removing all 68 00:04:28,430 --> 00:04:31,950 S1: geofencing restrictions from their drones in the US, which means 69 00:04:31,950 --> 00:04:37,070 S1: you can technically now fly them anywhere, including airports, wildfires. 70 00:04:37,070 --> 00:04:40,390 S1: Which one took down one of those big, uh, water 71 00:04:40,390 --> 00:04:44,710 S1: planes and the white House? This can't be right. Like, 72 00:04:44,710 --> 00:04:47,590 S1: how is this not going to get them TikTok? And 73 00:04:47,589 --> 00:04:49,710 S1: by TikTok, I don't mean turned off for one day 74 00:04:49,710 --> 00:04:53,470 S1: and turn back on. I mean, serious trouble. The new administration, 75 00:04:53,470 --> 00:04:58,910 S1: along with Larry Ellison, Sam Altman and Masayoshi Son, launched 76 00:04:58,910 --> 00:05:01,950 S1: a new AI project called Stargate, basically a new company 77 00:05:01,950 --> 00:05:06,589 S1: and a $500 Billion dollar investment. That's half $1 trillion 78 00:05:06,950 --> 00:05:09,390 S1: designed to make sure the US wins the AI war 79 00:05:10,230 --> 00:05:13,870 S1: versus China. And the Pentagon is now using AI from 80 00:05:13,870 --> 00:05:18,070 S1: companies like OpenAI and anthropic to help identify and assess 81 00:05:18,070 --> 00:05:22,349 S1: threats faster. But supposedly they're being careful to keep humans 82 00:05:22,350 --> 00:05:25,909 S1: in control of actual weapons for now. And Rachel Toback 83 00:05:26,750 --> 00:05:29,870 S1: posted that it was her pleasure serving on the Cisa 84 00:05:30,310 --> 00:05:33,710 S1: Technical Advisory Council, which has now officially been shut down. 85 00:05:33,830 --> 00:05:37,630 S1: Trump dissolves the advisory committees, which is exactly what she 86 00:05:37,630 --> 00:05:41,510 S1: was on, and that's now been shut down. And AI powered. 87 00:05:41,510 --> 00:05:47,070 S1: Brad Pitt scam costs a woman £830,000, which I don't 88 00:05:47,110 --> 00:05:49,510 S1: know how much that is in real money, but I 89 00:05:49,510 --> 00:05:53,589 S1: think pretty close to $1 million might be over $1 million. 90 00:05:53,589 --> 00:05:56,790 S1: It's something around there, and it was basically someone claiming 91 00:05:56,830 --> 00:05:59,550 S1: to be Brad Pitt in a hospital, and you could 92 00:05:59,550 --> 00:06:02,589 S1: see the AI. It wasn't super convincing, but things are 93 00:06:02,589 --> 00:06:05,470 S1: a lot more convincing when you are lonely. Trump killed 94 00:06:05,870 --> 00:06:10,390 S1: Biden's main AI safety executive order from 2023 that required 95 00:06:10,390 --> 00:06:14,350 S1: companies like OpenAI to share safety test results with the government. 96 00:06:14,430 --> 00:06:17,950 S1: And then he launched Stargate, which is about moving fast 97 00:06:17,950 --> 00:06:21,870 S1: to beat China. So essentially that order was about moving 98 00:06:21,870 --> 00:06:25,750 S1: slower and more carefully. And Stargate is about moving as 99 00:06:25,750 --> 00:06:29,190 S1: fast as possible. And I'm pretty happy with this, actually, because, 100 00:06:29,390 --> 00:06:32,390 S1: I mean, I'm worried about AI moving fast no matter what, 101 00:06:33,029 --> 00:06:37,390 S1: but I'm more worried about AI in China moving fast 102 00:06:37,390 --> 00:06:40,150 S1: while AI in the US moves slow. So that's where 103 00:06:40,150 --> 00:06:44,310 S1: the priority is for me. NATO has launched Operation Baltic Sentry, 104 00:06:44,350 --> 00:06:48,910 S1: putting 20 autonomous boats in the Baltic Sea to protect 105 00:06:48,910 --> 00:06:52,430 S1: undersea cables from Russian sabotage. I think this is excellent. 106 00:06:53,110 --> 00:06:59,790 S1: But 2020 drones in the Baltic Sea. How much area 107 00:06:59,790 --> 00:07:02,430 S1: do you have to cover? I imagine you can cover 108 00:07:02,430 --> 00:07:05,630 S1: a lot because you can use radar above, Of a 109 00:07:05,670 --> 00:07:08,630 S1: sonar below. I imagine they have lots of different ways 110 00:07:08,630 --> 00:07:12,030 S1: to cover a lot of ground. I'm just saying it's 111 00:07:12,030 --> 00:07:17,350 S1: a small number of boats and very large body of water. Palmer. 112 00:07:17,350 --> 00:07:22,430 S1: Lucky's defense tech company. Andrew. Andrew. I'm going to call it. 113 00:07:22,430 --> 00:07:27,150 S1: Andrew is building a massive 5,000,000 square foot weapons factory 114 00:07:27,150 --> 00:07:31,990 S1: in Columbus, Ohio called Arsenal one, dropping close to $1 115 00:07:31,990 --> 00:07:34,390 S1: billion of their own money on this thing. I and 116 00:07:34,390 --> 00:07:37,270 S1: tech companies are using your data to charge you more. 117 00:07:37,390 --> 00:07:41,230 S1: FTC just released data showing companies are using your location, demographics, 118 00:07:41,230 --> 00:07:43,990 S1: and even mouse movements to charge you different prices for 119 00:07:43,990 --> 00:07:47,030 S1: the same products. Trump just announced Stargate. We already talked 120 00:07:47,030 --> 00:07:50,190 S1: about that one. Perplexity just launched an API that lets 121 00:07:50,190 --> 00:07:54,030 S1: developers build their real time AI search capabilities into their 122 00:07:54,030 --> 00:07:56,830 S1: own apps. I'm already using this, so I'm not sure 123 00:07:56,830 --> 00:08:00,870 S1: exactly if this is sonar that I was already using before, 124 00:08:00,870 --> 00:08:04,350 S1: and they just now launched it. Not sure about that, 125 00:08:04,350 --> 00:08:07,710 S1: but it is pretty cool because you're getting live data, right? 126 00:08:07,950 --> 00:08:11,150 S1: It's not just an API call to a static database, 127 00:08:11,990 --> 00:08:14,590 S1: or it's not just an API call to an LLM. 128 00:08:15,110 --> 00:08:20,710 S1: It's the combination of a search engine with an LLM, 129 00:08:20,710 --> 00:08:24,430 S1: with live data queries on top of it. So quite 130 00:08:25,070 --> 00:08:30,990 S1: quite useful. Transformer two introduces Self-adapting language models. Sakana I 131 00:08:31,030 --> 00:08:33,869 S1: just released a new approach that lets language models dynamically 132 00:08:33,870 --> 00:08:36,030 S1: adjust their weights in real time based on the tasks 133 00:08:36,030 --> 00:08:39,550 S1: that they're working on. TSMC has officially started making four 134 00:08:39,550 --> 00:08:43,750 S1: nanometer chips at their new Arizona plant, which is a 135 00:08:43,750 --> 00:08:47,870 S1: massive win for the US semiconductor manufacturing. I wonder how 136 00:08:47,870 --> 00:08:51,910 S1: many people are thinking that Arizona is like plan A 137 00:08:51,910 --> 00:08:55,429 S1: if China takes Taiwan? I love the fact that we 138 00:08:55,470 --> 00:08:57,949 S1: are in this position. I mean, it would be horrific 139 00:08:57,950 --> 00:09:01,190 S1: if this happened because it would still massively disrupt for 140 00:09:01,190 --> 00:09:05,110 S1: years and years. It would massively disrupt chip production and 141 00:09:05,110 --> 00:09:10,710 S1: basically technology production worldwide. If we were to lose the 142 00:09:10,710 --> 00:09:15,590 S1: TSMC plant in Taiwan, like, it's not a easy hot swap. 143 00:09:16,230 --> 00:09:19,990 S1: First of all, four nanometer is like old tech, and 144 00:09:20,150 --> 00:09:22,550 S1: TSMC has been talking for a long time about how 145 00:09:23,350 --> 00:09:26,110 S1: the staff in Arizona is nowhere near as good as 146 00:09:26,110 --> 00:09:29,709 S1: the staff in in Taiwan. The tech isn't as good. 147 00:09:29,710 --> 00:09:34,550 S1: It's not as as modern of a facility. It's like 148 00:09:34,590 --> 00:09:40,309 S1: way behind. But but if China basically takes this piece 149 00:09:40,309 --> 00:09:43,949 S1: off the board for the West, which is what Taiwan 150 00:09:43,950 --> 00:09:46,590 S1: is and what the US is and what you know, 151 00:09:46,630 --> 00:09:48,590 S1: the West is, if they were to take that off 152 00:09:48,590 --> 00:09:50,350 S1: the board, it would be nice to be able to 153 00:09:50,350 --> 00:09:52,150 S1: put a piece back on the board and do it 154 00:09:52,150 --> 00:09:57,309 S1: inside of Arizona, inside of the US. That would be wonderful. And, 155 00:09:57,309 --> 00:10:01,990 S1: you know, they would pump whatever administrations in power would pump, 156 00:10:02,230 --> 00:10:05,390 S1: you know, a couple of trillion dollars into that, probably 157 00:10:05,750 --> 00:10:11,350 S1: like immediately build two more plants, fully move tons of 158 00:10:11,350 --> 00:10:15,750 S1: people from TSMC over to Arizona or wherever, and get 159 00:10:15,750 --> 00:10:19,230 S1: this thing going properly. And speaking of TSMC, they had 160 00:10:19,230 --> 00:10:23,670 S1: to temporarily halt chip production after a 6.4 magnitude earthquake 161 00:10:23,670 --> 00:10:27,150 S1: hit the southern part of the island. And basically whatever 162 00:10:27,190 --> 00:10:30,470 S1: was being made at the time, right, right at this tiny, 163 00:10:30,470 --> 00:10:34,630 S1: tiny little scale. By the way, only five carbon atoms 164 00:10:34,870 --> 00:10:38,910 S1: fit inside of one nanometer. That is the thing I 165 00:10:38,950 --> 00:10:43,829 S1: learned from that acquired episode on TSMC. But yeah, basically, 166 00:10:43,830 --> 00:10:46,829 S1: while these things were being made, while a bunch of 167 00:10:46,830 --> 00:10:53,429 S1: chips were in production, this 6.4 earthquake hit. And earthquakes 168 00:10:53,429 --> 00:10:56,790 S1: shake things. That's what they do. So they basically have 169 00:10:56,790 --> 00:10:59,390 S1: to say, okay, anything being made at the time, most 170 00:10:59,390 --> 00:11:01,390 S1: of it is probably garbage. And they kind of throw 171 00:11:01,390 --> 00:11:05,390 S1: that out, which is probably very expensive. And then of 172 00:11:05,390 --> 00:11:07,150 S1: course you have to make sure everything is fine before 173 00:11:07,150 --> 00:11:10,300 S1: you start everything back up again. Zuckerberg just announced they're 174 00:11:10,300 --> 00:11:14,339 S1: letting go of around 3600 people by February 10th, and 175 00:11:14,340 --> 00:11:16,900 S1: they're planning to refill them with new hires. My read 176 00:11:16,900 --> 00:11:20,099 S1: on this is that it's a constant cleansing and replacing 177 00:11:20,100 --> 00:11:24,780 S1: of the workforce with what I call Alaskan boat crews. 178 00:11:25,460 --> 00:11:28,699 S1: Fishing boat crews. Right. And I think after doing this 179 00:11:28,700 --> 00:11:31,820 S1: for like a year, doing this a few times, the 180 00:11:31,820 --> 00:11:37,220 S1: culture will be basically like a killer culture. It'll be like, look, 181 00:11:37,220 --> 00:11:40,100 S1: we are here to work. We don't tolerate people who 182 00:11:40,140 --> 00:11:42,740 S1: don't work. We will report you. You will get put 183 00:11:42,740 --> 00:11:45,620 S1: on the naughty list and you'll get taken out. So 184 00:11:45,740 --> 00:11:49,939 S1: what it does is it just encourages like this super aggressive, 185 00:11:49,940 --> 00:11:54,500 S1: super high performance culture. And my prediction is that the 186 00:11:54,500 --> 00:11:57,140 S1: stock is going to go up quite a bit. But 187 00:11:57,179 --> 00:11:59,700 S1: don't take my word for that. Just uh, that's a 188 00:11:59,700 --> 00:12:02,260 S1: random prediction I like to put into the world so 189 00:12:02,260 --> 00:12:04,860 S1: that I can capture it and tell me when I 190 00:12:04,860 --> 00:12:07,380 S1: get it wrong. Okay. Meta just released a new AI 191 00:12:07,620 --> 00:12:12,300 S1: translation system called seamless that can translate speech between 36 192 00:12:12,300 --> 00:12:17,100 S1: languages while preserving the speaker's voice and emotional tone. Getting 193 00:12:17,100 --> 00:12:20,980 S1: closer to the universal translator. Scientists have created a new 194 00:12:20,980 --> 00:12:26,660 S1: laser measurement technique that can measure distances of over 100km 195 00:12:26,660 --> 00:12:31,980 S1: with nanometer level precision, which is insane. It's basically the 196 00:12:31,980 --> 00:12:40,380 S1: distance between two cities with the precision of 1/1000. One one. Oh, goodness. 197 00:12:40,660 --> 00:12:45,140 S1: 1/1000 the width of a human hair. I didn't realize 198 00:12:45,140 --> 00:12:49,780 S1: that was a nanometer. 1/1000. Holy crap. That's that's small. 199 00:12:49,980 --> 00:12:57,700 S1: 1/1000 of a human hair can fit five carbon atoms. Interesting. 200 00:12:57,980 --> 00:13:00,860 S1: RSS is making a comeback because it lets you get 201 00:13:00,860 --> 00:13:04,860 S1: all the good stuff from social media without the algorithmic manipulation. 202 00:13:05,540 --> 00:13:07,660 S1: And it got a link to mine. It's just Daniel 203 00:13:08,940 --> 00:13:14,940 S1: RSS or. Feed RSS. There's lots of different ones. If 204 00:13:14,940 --> 00:13:17,459 S1: you just do slash feed or slash RSS. It will 205 00:13:17,460 --> 00:13:20,980 S1: take you there. GitHub actions fall short for complex projects. 206 00:13:21,420 --> 00:13:24,980 S1: So somebody shares frustrating experience with GitHub actions breaking down 207 00:13:24,980 --> 00:13:30,660 S1: a larger, more complex environments like Monorepos with multiple teams. Humans. 208 00:13:30,660 --> 00:13:34,780 S1: Japan's elderly women choose prisons over loneliness. So a lot 209 00:13:34,820 --> 00:13:38,380 S1: of women are choosing to do petty crimes, so they'll 210 00:13:38,380 --> 00:13:41,980 S1: get thrown in jail where they get meals, health care 211 00:13:41,980 --> 00:13:47,220 S1: and community. This is extremely depressing to me. How are 212 00:13:47,220 --> 00:13:50,540 S1: we at this point that you've got these poor old 213 00:13:50,540 --> 00:13:53,620 S1: ladies who don't want to commit crimes? And it's not 214 00:13:53,620 --> 00:13:56,380 S1: really about the crimes. It's the fact that you would 215 00:13:56,380 --> 00:13:59,819 S1: rather go to jail because you're that lonely. What are 216 00:13:59,820 --> 00:14:04,060 S1: we doing? What are we doing wrong? Human connection is everything. U.S. 217 00:14:04,100 --> 00:14:06,900 S1: worker job satisfaction hits a ten year low. A new 218 00:14:06,900 --> 00:14:09,620 S1: Gallup poll shows American workers are more checked out than 219 00:14:09,620 --> 00:14:15,460 S1: they've been in a decade. Decade? What? 2015. Only 31% 220 00:14:15,460 --> 00:14:18,420 S1: saying they're engaged at work. You know, you know my spiel. 221 00:14:18,420 --> 00:14:20,660 S1: I'm not going to go into it. Sweden is reversing 222 00:14:20,660 --> 00:14:26,260 S1: its 2009 all digital education initiative by bringing back printed textbooks. 223 00:14:26,260 --> 00:14:30,940 S1: A comprehensive study in Texas from 2012 to 2018 found 224 00:14:30,940 --> 00:14:34,980 S1: that undocumented immigrants commit violent and drug crimes at less 225 00:14:34,980 --> 00:14:38,500 S1: than half the rate of native born citizens and property 226 00:14:38,500 --> 00:14:43,740 S1: crimes at just 25%. The rate of non know at 227 00:14:43,740 --> 00:14:49,060 S1: the rate of native born citizens. So dangerous crime basically 228 00:14:49,700 --> 00:14:55,500 S1: half and property crime only 25% as often. FTC just 229 00:14:55,500 --> 00:14:58,780 S1: released a report showing UnitedHealth and other major health care 230 00:14:58,820 --> 00:15:03,380 S1: companies are marking up cancer drugs by over 1,000%. Medicare 231 00:15:03,420 --> 00:15:06,460 S1: just added 15 more drugs to their price negotiation list, 232 00:15:06,460 --> 00:15:11,340 S1: including Ozempic and Wegovy. Scientists discovered Greenland sharks are the 233 00:15:11,340 --> 00:15:17,220 S1: longest living vertebrates on Earth. Some potentially being as old 234 00:15:17,220 --> 00:15:22,220 S1: as like 600 years old. Ideas Stoicism's gift. The greatest 235 00:15:22,220 --> 00:15:26,060 S1: gift that stoicism has given me personally is the ability 236 00:15:26,060 --> 00:15:28,900 S1: to enjoy something I still have, as if I no 237 00:15:28,900 --> 00:15:32,540 S1: longer have it. It is the ultimate frame. The greatest 238 00:15:32,540 --> 00:15:35,020 S1: gift that stoicism has given me is the ability to 239 00:15:35,020 --> 00:15:37,860 S1: enjoy something I still have, as if I no longer 240 00:15:37,860 --> 00:15:43,580 S1: have it. Discovery. SSH. SSH into throwaway Docker containers. Basically, 241 00:15:43,620 --> 00:15:45,900 S1: it's an open source tool lets you instantly spin up 242 00:15:45,900 --> 00:15:50,620 S1: a disposable Docker containers via SSH for quick testing and development. 243 00:15:50,740 --> 00:15:53,620 S1: And it got a bunch of my buddy Joseph Thatcher's 244 00:15:53,620 --> 00:15:58,540 S1: favorite lists. And, uh, UL was lucky enough to make 245 00:15:58,540 --> 00:16:00,460 S1: it on there a couple of times and the list 246 00:16:00,460 --> 00:16:03,340 S1: was quite long, so worth taking a look at. In fact, 247 00:16:03,340 --> 00:16:06,380 S1: I'm going to click over there. So best AI app cursor. 248 00:16:06,780 --> 00:16:12,500 S1: Best model cloud sonnet. Best red team company Hayes Labs, right? 249 00:16:12,540 --> 00:16:15,300 S1: So these are some of the things that he put 250 00:16:15,300 --> 00:16:18,100 S1: in their command line tool. Yeah. Fabric. That's one of 251 00:16:18,100 --> 00:16:23,380 S1: our mentions. Google bug bounty I jailbreaking resource. So lots 252 00:16:23,380 --> 00:16:25,140 S1: of cool ones over here. What I wish I knew 253 00:16:25,140 --> 00:16:29,340 S1: before quitting my job. So Michael Douglas shares his raw 254 00:16:29,340 --> 00:16:32,060 S1: experience of how quitting his job to work on his 255 00:16:32,100 --> 00:16:35,140 S1: own turned out way harder than expected. I post this 256 00:16:35,140 --> 00:16:38,740 S1: because I'm always talking about the counter narrative of how, oh, 257 00:16:38,780 --> 00:16:41,060 S1: you got to go independent. You got to do it 258 00:16:41,060 --> 00:16:43,620 S1: sooner than later. I just want to make it clear 259 00:16:43,620 --> 00:16:46,180 S1: that there is a counter narrative here. You do have 260 00:16:46,180 --> 00:16:50,540 S1: to be careful. You've got to get oriented and you 261 00:16:50,540 --> 00:16:52,940 S1: do have to be careful, is basically what I'm saying. 262 00:16:52,940 --> 00:16:56,980 S1: Michael Box shares a really practical framework called hypothesis sheets, 263 00:16:56,980 --> 00:17:01,460 S1: for validating B2B startup ideas before committing to them and 264 00:17:01,460 --> 00:17:04,980 S1: the recommendation of the week. Try something different with your 265 00:17:04,980 --> 00:17:08,300 S1: meditation for the next few weeks. Make a list of 266 00:17:08,300 --> 00:17:10,820 S1: the relationships and other good things you have in your life, 267 00:17:11,380 --> 00:17:14,659 S1: which are things like your husband, your wife, your particular 268 00:17:15,260 --> 00:17:18,340 S1: particular kid. You have a close friend. The fact that 269 00:17:18,340 --> 00:17:21,900 S1: you aren't hungry or cold at the moment. And now 270 00:17:21,900 --> 00:17:25,860 S1: imagine that that thing or that person is gone. But 271 00:17:25,859 --> 00:17:28,700 S1: as per your meditation, really, really imagine it. Like put 272 00:17:28,700 --> 00:17:31,300 S1: yourself in that mental mode if that thing is really 273 00:17:31,300 --> 00:17:34,540 S1: not in your life anymore. Imagine what you would do next. 274 00:17:34,940 --> 00:17:37,860 S1: What does a day look like? Imagine watching TV. Imagine 275 00:17:37,859 --> 00:17:41,780 S1: brushing your teeth. But without that person on the planet. 276 00:17:41,859 --> 00:17:44,500 S1: Then wake up and realize that they are still here. 277 00:17:44,700 --> 00:17:47,500 S1: This is a really, really powerful thing. And lots of 278 00:17:47,500 --> 00:17:52,179 S1: different stoic teachers taught this, and I soaked it in 279 00:17:52,180 --> 00:17:57,219 S1: from Marcus Aurelius. But when I was doing my stoicism stuff, 280 00:17:57,300 --> 00:18:00,660 S1: I must have been like 2018, 2019. I probably read 281 00:18:00,700 --> 00:18:04,940 S1: like 5 or 10 books around stoicism, including like summaries 282 00:18:04,940 --> 00:18:07,780 S1: and a whole bunch of source material. And I just 283 00:18:07,780 --> 00:18:10,780 S1: got this lesson drilled in over and over is like, 284 00:18:11,609 --> 00:18:14,169 S1: Realize what you have. I also learned this in the 285 00:18:14,170 --> 00:18:19,129 S1: army after being very, very cold and really appreciating things 286 00:18:19,130 --> 00:18:23,530 S1: like a warm bed and, uh, and Burger King. Honestly, 287 00:18:24,050 --> 00:18:26,530 S1: kind of a kind of a stupid example, but I 288 00:18:26,530 --> 00:18:29,050 S1: wasn't even in combat. I mean, I was just out 289 00:18:29,050 --> 00:18:33,210 S1: in the in the nasty stuff in, like, Kentucky freezing 290 00:18:33,369 --> 00:18:37,290 S1: my balls off. And it was bad. It was really 291 00:18:37,290 --> 00:18:41,210 S1: bad compared to being back at the base. And, you know, 292 00:18:41,250 --> 00:18:45,650 S1: in normal life, like most people lead. So just imagining, like, 293 00:18:45,650 --> 00:18:47,850 S1: right now I won't go too far into this, but 294 00:18:47,890 --> 00:18:51,850 S1: like right now it annoys me that there are several 295 00:18:51,850 --> 00:18:55,010 S1: things in my life right now in my environment and 296 00:18:55,010 --> 00:18:57,129 S1: just around me. The fact that I have this microphone, 297 00:18:57,130 --> 00:19:00,210 S1: the fact that I'm looking at a computer screen. It 298 00:19:00,210 --> 00:19:03,370 S1: annoys me that there's a list of hundreds of things 299 00:19:03,890 --> 00:19:06,010 S1: that if I didn't have them, I would wish I 300 00:19:06,010 --> 00:19:08,770 S1: had them. I have pictures on my wall here in 301 00:19:08,770 --> 00:19:13,490 S1: my house of the view of Earth from Mars. It's 302 00:19:13,490 --> 00:19:16,650 S1: a tiny little speck, and I like to walk by. 303 00:19:16,690 --> 00:19:18,170 S1: The reason I put him up is because I like 304 00:19:18,170 --> 00:19:20,850 S1: to walk by and look at this picture and be like. 305 00:19:21,530 --> 00:19:25,810 S1: Or a picture of the sun from Mars. It especially 306 00:19:25,810 --> 00:19:28,570 S1: though looking at the Earth like the pale blue dot picture. 307 00:19:28,690 --> 00:19:30,929 S1: I've got one of those as well. I love looking 308 00:19:30,930 --> 00:19:34,050 S1: at it, imagining that we've been jettisoned off the planet 309 00:19:34,050 --> 00:19:37,050 S1: like I'm in a prison colony. I've been kicked out 310 00:19:37,050 --> 00:19:39,729 S1: of Earth, and I'm looking at it from Mars. And 311 00:19:39,730 --> 00:19:42,770 S1: I'm just wishing. I'm just wishing I could go to 312 00:19:42,770 --> 00:19:45,850 S1: that one Mexican restaurant. I could hang out with my friends. 313 00:19:45,850 --> 00:19:48,770 S1: I could get a Starbucks. There's no more Starbucks. I'm 314 00:19:48,770 --> 00:19:52,010 S1: in prison on Mars. I will never have another Starbucks again. 315 00:19:52,010 --> 00:19:54,609 S1: And I will tell all my friends every time I 316 00:19:54,650 --> 00:19:58,649 S1: meet them in the prison yard on Mars how cool 317 00:19:58,690 --> 00:20:01,610 S1: Starbucks was. Am I actually a huge fan of Starbucks? 318 00:20:01,650 --> 00:20:04,369 S1: Not really. I do like the vibe. I like how 319 00:20:04,369 --> 00:20:07,010 S1: it smells in there. I like, you know, it's got 320 00:20:07,010 --> 00:20:09,209 S1: a vibe to it. I like to go in there 321 00:20:09,210 --> 00:20:12,330 S1: and work. The point is, whatever you don't have is 322 00:20:12,330 --> 00:20:15,010 S1: the thing you want the most. And there's a there's 323 00:20:15,010 --> 00:20:17,050 S1: a list of 100 of those things that I have 324 00:20:17,050 --> 00:20:20,450 S1: right now that are just amazing. And the most important 325 00:20:20,450 --> 00:20:23,850 S1: ones are actually relationships. And I think everyone's figured this 326 00:20:23,850 --> 00:20:27,690 S1: out by now. But stoicism reminds me to take inventory 327 00:20:27,890 --> 00:20:31,330 S1: and to imagine yourself without those things. And I recommend 328 00:20:31,330 --> 00:20:34,370 S1: you you try this practice as well. And the aphorism 329 00:20:34,410 --> 00:20:37,650 S1: of the week A rational person can find peace by 330 00:20:37,650 --> 00:20:42,330 S1: cultivating indifference to things outside of their control. A rational 331 00:20:42,330 --> 00:20:47,290 S1: person can find peace by cultivating indifference to things outside 332 00:20:47,290 --> 00:20:52,170 S1: of their control. Naval Ravikant. Unsupervised learning is produced on 333 00:20:52,170 --> 00:20:56,570 S1: Hindenburg Pro using an SM seven B microphone. A video 334 00:20:56,570 --> 00:20:59,490 S1: version of the podcast is available on the Unsupervised Learning 335 00:20:59,490 --> 00:21:02,690 S1: YouTube channel, and the text version with full links and 336 00:21:02,690 --> 00:21:07,929 S1: notes is available at Daniel Miessler newsletter. We'll see you 337 00:21:07,930 --> 00:21:08,530 S1: next time.