WEBVTT 00:00:00.000 --> 00:00:02.899 I want you to imagine, just for a second, trying 00:00:02.899 --> 00:00:04.860 to build a perfectly functioning, incredibly 00:00:04.860 --> 00:00:08.439 complex mechanical device. But, you know, here's 00:00:08.439 --> 00:00:11.419 the catch. Right. There is always a catch. Exactly. 00:00:11.820 --> 00:00:13.859 You have never actually seen a blueprint for 00:00:13.859 --> 00:00:16.120 this machine. You have never seen a photograph 00:00:16.120 --> 00:00:19.519 of it. You have never even held a single gear 00:00:19.519 --> 00:00:21.940 or a switch or a wire that belongs to it. Wow. 00:00:22.179 --> 00:00:24.579 OK. Yeah. The only thing you have, like your 00:00:24.579 --> 00:00:26.920 absolute only clue to how this whole machine 00:00:26.920 --> 00:00:29.820 operates, is the scrambled gibberish it spits 00:00:29.820 --> 00:00:32.920 out. It is... It's basically the cryptographic 00:00:32.920 --> 00:00:35.439 equivalent of trying to map out the entire floor 00:00:35.439 --> 00:00:38.460 plan of a massive pitch black building while 00:00:38.460 --> 00:00:40.640 you're standing blindfolded across the street, 00:00:40.859 --> 00:00:42.780 you know, simply by listening to the sound of 00:00:42.780 --> 00:00:44.780 someone else's footsteps echoing inside. That 00:00:44.780 --> 00:00:47.820 is such a wild image, but it's true. You are 00:00:47.820 --> 00:00:50.079 reverse engineering a ghost using nothing but 00:00:50.079 --> 00:00:52.659 abstract logic. And that is the staggering historical 00:00:52.659 --> 00:00:55.200 reality we are exploring today in this deep dive. 00:00:55.280 --> 00:00:59.350 We are looking at one of the most consequential 00:00:59.350 --> 00:01:02.350 intellectual triumphs of the 20th century, really. 00:01:02.630 --> 00:01:05.189 The Japanese type B cipher machine. Famously 00:01:05.189 --> 00:01:08.010 code named purple by the United States. Right, 00:01:08.049 --> 00:01:09.909 the purple machine. The Japanese government, 00:01:09.909 --> 00:01:11.689 I mean, they considered the purple cipher to 00:01:11.689 --> 00:01:15.450 be absolute mathematical perfection. They deployed 00:01:15.450 --> 00:01:18.129 it to protect their highest level diplomatic 00:01:18.129 --> 00:01:21.510 communications in those tense years leading up 00:01:21.510 --> 00:01:24.000 to and during. World War II. So they were putting 00:01:24.000 --> 00:01:26.200 all their eggs in this one basket. Completely. 00:01:26.359 --> 00:01:29.680 It represents this fascinating study of technological 00:01:29.680 --> 00:01:33.099 hubris, honestly, because its creators were just 00:01:33.099 --> 00:01:36.040 so convinced of its invulnerability that they 00:01:36.040 --> 00:01:38.659 basically trusted it with the survival of their 00:01:38.659 --> 00:01:42.459 entire empire. Which is heavy. So our mission 00:01:42.459 --> 00:01:44.819 for you today is to unpack this machine from 00:01:44.819 --> 00:01:47.599 the inside out. We are going to look at the mechanical 00:01:47.599 --> 00:01:50.310 architecture of Purple. explore how American 00:01:50.310 --> 00:01:53.349 code breakers managed to like blindly reconstruct 00:01:53.349 --> 00:01:55.950 it from thin air and trace how the intelligence 00:01:55.950 --> 00:01:58.409 they pulled from it completely altered the global 00:01:58.409 --> 00:02:00.489 battlefield. It really changed everything. It 00:02:00.489 --> 00:02:02.549 did. But, you know, before we get to the purple 00:02:02.549 --> 00:02:04.629 machine itself, the sources show we actually 00:02:04.629 --> 00:02:06.730 need to rewind a bit. We have to look at the 00:02:06.730 --> 00:02:09.169 predecessor that basically set Gopal up for failure. 00:02:09.430 --> 00:02:11.610 Yeah, you really cannot understand the fatal 00:02:11.610 --> 00:02:14.330 flaws of purple without understanding the paranoia 00:02:14.330 --> 00:02:16.650 that birthed it in the first place. We have to 00:02:16.650 --> 00:02:21.360 go back to, um... the 1922 Washington Naval Treaty. 00:02:21.680 --> 00:02:23.460 Right. The arms control conference after World 00:02:23.460 --> 00:02:26.099 War One. Exactly. So during those negotiations, 00:02:26.419 --> 00:02:29.099 the Japanese delegation was just completely stunned 00:02:29.099 --> 00:02:31.659 to discover that the American Black Chamber, 00:02:31.879 --> 00:02:34.219 which was the U .S. cryptography bureau at the 00:02:34.219 --> 00:02:36.620 time, had been intercepting and reading their 00:02:36.620 --> 00:02:38.919 secure diplomatic traffic. Just reading their 00:02:38.919 --> 00:02:40.599 mail right in front of them essentially. Yeah. 00:02:40.659 --> 00:02:43.379 And the Americans used that intercepted intelligence 00:02:43.379 --> 00:02:47.620 to force Japan into a really unfavorable treaty 00:02:47.620 --> 00:02:49.840 regarding the size of their naval fleet. Man, 00:02:50.000 --> 00:02:52.400 that level of exposure on the global stage, it 00:02:52.400 --> 00:02:55.360 must have created just massive institutional 00:02:55.360 --> 00:02:58.900 panic in Tokyo. Oh, it was a profound humiliation 00:02:58.900 --> 00:03:00.960 for the Japanese military establishment. I mean, 00:03:01.080 --> 00:03:03.500 the pressure to completely overthrow their cryptographic 00:03:03.500 --> 00:03:06.659 security was immense. So the Japanese Navy decided, 00:03:06.659 --> 00:03:08.340 you know, they had to take the lead on this. 00:03:08.580 --> 00:03:11.620 And by 1931, their engineers had developed a 00:03:11.620 --> 00:03:14.979 new electromechanical cipher machine called the 00:03:14.979 --> 00:03:18.960 Type 91. The United States would later code name 00:03:18.960 --> 00:03:22.699 this early machine Red. OK, so I am looking at 00:03:22.699 --> 00:03:24.939 the history of this Red machine. and I expected 00:03:24.939 --> 00:03:27.180 it to be the impenetrable fortress they wanted, 00:03:27.560 --> 00:03:30.159 but it actually had some really glaring operational 00:03:30.159 --> 00:03:32.699 issues, right? Oh yeah, big time. Beyond just 00:03:32.699 --> 00:03:35.020 being temperamental -like, apparently it required 00:03:35.020 --> 00:03:37.180 constant cleaning to prevent electrical shorts. 00:03:37.719 --> 00:03:40.479 The Japanese Navy eventually discovered a fundamental 00:03:40.479 --> 00:03:43.740 mathematical vulnerability. They realized the 00:03:43.740 --> 00:03:46.580 red cipher could... you know, theoretically be 00:03:46.580 --> 00:03:48.860 broken by a dedicated adversary. And see, it 00:03:48.860 --> 00:03:50.860 is worse than just a realized vulnerability. 00:03:51.319 --> 00:03:53.460 This is where bureaucratic dysfunction basically 00:03:53.460 --> 00:03:56.300 sabotages national security. How so? Well, the 00:03:56.300 --> 00:03:59.199 Japanese Navy knows the Red Machine has a cryptographic 00:03:59.199 --> 00:04:01.650 weakness. But instead of immediately pulling 00:04:01.650 --> 00:04:04.650 it from service, they actively obscure the severity 00:04:04.650 --> 00:04:06.830 of the flaw from the Japanese Ministry of Foreign 00:04:06.830 --> 00:04:09.449 Affairs. Wait, the same Ministry of Foreign Affairs 00:04:09.449 --> 00:04:11.930 that relies on this exact machine for all of 00:04:11.930 --> 00:04:13.509 their international diplomatic correspondence? 00:04:13.550 --> 00:04:15.490 That is the one. I am trying to wrap my head 00:04:15.490 --> 00:04:18.829 around that decision. It is like a corporation's 00:04:18.829 --> 00:04:21.310 IT department discovering that their company 00:04:21.310 --> 00:04:24.670 -wide password manager has this huge back door 00:04:24.670 --> 00:04:26.970 that hackers can exploit. Yeah. But instead of 00:04:26.970 --> 00:04:30.319 shutting it down, they just... vaguely advised 00:04:30.319 --> 00:04:33.439 the CEO to be careful while letting everyone 00:04:33.439 --> 00:04:37.279 keep using it just to save face. Yes, that institutional 00:04:37.279 --> 00:04:40.120 face -saving is the absolute core issue here. 00:04:40.579 --> 00:04:42.939 The lack of internal cooperation between the 00:04:42.939 --> 00:04:45.839 Japanese Army, the Navy, and the Foreign Office 00:04:45.839 --> 00:04:49.519 created these massive silos of information. Furthermore, 00:04:49.879 --> 00:04:52.000 the Navy placed just an unwarranted amount of 00:04:52.000 --> 00:04:54.439 faith in their own internal talent. Right, because 00:04:54.439 --> 00:04:56.399 they brought in their own guy. Exactly. They 00:04:56.399 --> 00:04:59.079 relied heavily on mathematicians like Tej Takagi. 00:05:00.139 --> 00:05:03.040 Now Takagi was undeniably a brilliant world -class 00:05:03.040 --> 00:05:05.040 mathematician in the realm of number theory. 00:05:05.720 --> 00:05:07.920 But... And there's always a but, he wasn't a 00:05:07.920 --> 00:05:09.920 code breaker. Right, he lacked any practical 00:05:09.920 --> 00:05:11.959 background in cryptanalysis. So he understood 00:05:11.959 --> 00:05:13.939 the pure mathematics of encryption, but he had 00:05:13.939 --> 00:05:16.259 no actual experience breaking codes. It's like 00:05:16.259 --> 00:05:18.759 he knew how to build a really complex lock, but 00:05:18.759 --> 00:05:20.699 he had never studied how a professional lock 00:05:20.699 --> 00:05:22.699 picker actually goes about breaking one open. 00:05:22.939 --> 00:05:25.540 That is a perfect way to put it. That offensive 00:05:25.540 --> 00:05:29.480 mindset was completely missing. So when the Japanese 00:05:29.480 --> 00:05:32.079 rolled out the next -generation machine for the 00:05:32.079 --> 00:05:35.439 Ministry of Foreign Affairs in 1937, which was 00:05:35.439 --> 00:05:39.139 the Type 97, or PURPLE, that interdepartmental 00:05:39.139 --> 00:05:41.939 silence meant that PURPLE inadvertently inherited 00:05:41.939 --> 00:05:44.459 this foundational DNA from the compromised RED 00:05:44.459 --> 00:05:46.920 machine. They just copied the homework without 00:05:46.920 --> 00:05:48.800 realizing the answers were wrong. Basically, 00:05:48.980 --> 00:05:51.579 yeah. The engineers carried over core design 00:05:51.579 --> 00:05:53.879 philosophies without realizing those philosophies 00:05:53.879 --> 00:05:56.160 were inherently flawed from the start. Okay, 00:05:56.199 --> 00:05:58.990 so... Because that flawed DNA was baked in, let's 00:05:58.990 --> 00:06:01.410 look at the actual architecture of purple. Because 00:06:01.410 --> 00:06:03.329 I originally assumed that, you know, since this 00:06:03.329 --> 00:06:05.589 was a World War II electromechanical cipher, 00:06:06.230 --> 00:06:09.189 they were using a system of spinning, wired rotors 00:06:09.189 --> 00:06:12.269 to scramble letters. Like the famous German Enigma 00:06:12.269 --> 00:06:15.350 machine. Exactly. But the sources show they went 00:06:15.350 --> 00:06:17.410 in a completely different mechanical direction. 00:06:17.449 --> 00:06:19.889 Yeah, they bypassed rotor technology entirely. 00:06:20.230 --> 00:06:22.569 Instead of rotors, the Japanese engineers utilized 00:06:22.569 --> 00:06:24.810 what are called stepping switches. Okay, stepping 00:06:24.810 --> 00:06:26.829 switches. What are those exactly? To understand 00:06:26.829 --> 00:06:28.310 these, you really have to look at the commercial 00:06:28.310 --> 00:06:31.250 technology of the 1930s. These were the exact 00:06:31.250 --> 00:06:34.910 same multi -layered mechanical switches utilized 00:06:34.910 --> 00:06:37.370 in automatic telephoto exchanges of the era. 00:06:37.889 --> 00:06:41.069 Oh, wow. So they essentially repurposed off -the 00:06:41.069 --> 00:06:43.800 -shelf commercial telecommunications hardware 00:06:43.800 --> 00:06:46.959 for a top secret military cipher. Pretty much. 00:06:47.040 --> 00:06:49.360 It was a very clever application of existing 00:06:49.360 --> 00:06:52.259 technology, to be fair. In the purple machine, 00:06:52.660 --> 00:06:55.819 these stepping switches featured layers of electrical 00:06:55.819 --> 00:06:58.139 contacts arranged in these semicircular banks. 00:06:58.300 --> 00:07:00.920 OK, I'm picturing that. And a mechanical wiper 00:07:00.920 --> 00:07:03.680 arm would sweep across these contacts, completing 00:07:03.680 --> 00:07:05.480 different electrical circuits to scramble the 00:07:05.480 --> 00:07:08.060 input. So an operator would type a plain letter 00:07:08.060 --> 00:07:10.240 on an electric typewriter. The signal would travel 00:07:10.240 --> 00:07:12.459 through a plug board, go into this chaotic maze 00:07:12.459 --> 00:07:15.480 of stepping switches, and emerge as a totally 00:07:15.480 --> 00:07:18.160 different enciphered letter on a second typewriter. 00:07:18.399 --> 00:07:20.939 Right. And this brings us to the specific design 00:07:20.939 --> 00:07:23.839 decision that completely compromised the entire 00:07:23.839 --> 00:07:26.649 system. Because Roman letters were the international 00:07:26.649 --> 00:07:28.670 standard for telegraph transmission at the time, 00:07:28.790 --> 00:07:31.250 right? Yes, exactly. So the Japanese text was 00:07:31.250 --> 00:07:34.449 first transliterated into the 26 -letter English 00:07:34.449 --> 00:07:37.339 alphabet. But when feeding those 26 letters into 00:07:37.339 --> 00:07:39.399 the stepping switches, the purple machine didn't 00:07:39.399 --> 00:07:41.980 scramble the whole alphabet together in one massive 00:07:41.980 --> 00:07:44.300 pool, did it? No, it didn't. It segregated them. 00:07:44.720 --> 00:07:47.519 The machine physically divided the 26 letters 00:07:47.519 --> 00:07:50.759 into two distinct groups. This is wild to me. 00:07:50.879 --> 00:07:53.220 It is. There was a group of six letters, which 00:07:53.220 --> 00:07:55.560 were typically vowels, and a group of 20 letters, 00:07:55.639 --> 00:07:57.939 which were usually consonants. The famous 6s 00:07:57.939 --> 00:08:01.000 and 20s. Right. And the 20 group was heavily 00:08:01.000 --> 00:08:03.540 protected. When a letter from the 20s group was 00:08:03.540 --> 00:08:05.870 typed, the electoral signal was routed through 00:08:05.870 --> 00:08:09.550 three successive highly complex stages of stepping 00:08:09.550 --> 00:08:11.629 switches. So lots of layers of scrambling for 00:08:11.629 --> 00:08:14.709 those 20. Tons. Some of these switches stepped 00:08:14.709 --> 00:08:17.410 rapidly with every keystroke, some stepped at 00:08:17.410 --> 00:08:20.910 a medium pace, and some moved very slowly. It 00:08:20.910 --> 00:08:23.730 created this wildly unpredictable rat's nest 00:08:23.730 --> 00:08:26.290 of permutations. But, and here's the kicker, 00:08:26.829 --> 00:08:29.810 the group of six letters was treated entirely 00:08:29.810 --> 00:08:31.810 differently. Completely different. A letter in 00:08:31.810 --> 00:08:34.509 the sixes group only passed through a single 00:08:34.509 --> 00:08:37.750 stepping switch stage, and that switch just stepped 00:08:37.750 --> 00:08:40.629 forward one predictable position for every character 00:08:40.629 --> 00:08:42.929 type. Just one click at a time. I mean from a 00:08:42.929 --> 00:08:45.529 security standpoint, if you split your alphabet 00:08:45.529 --> 00:08:47.909 and treat a fraction of it with vastly weaker 00:08:47.909 --> 00:08:50.450 encryption, aren't you basically handing the 00:08:50.450 --> 00:08:53.029 enemy a massive foothold? Oh, you absolutely 00:08:53.029 --> 00:08:55.389 are. You reduce the mathematical complexity of 00:08:55.389 --> 00:08:58.639 the cipher exponentially. because the cryptanalysts 00:08:58.639 --> 00:09:01.320 can isolate those six letters immediately. Like, 00:09:01.480 --> 00:09:04.000 why on earth would they do that? It drastically 00:09:04.000 --> 00:09:06.840 degrades the cryptographic integrity of the entire 00:09:06.840 --> 00:09:09.080 system. And the reason they implemented this 00:09:09.080 --> 00:09:11.340 split, it actually traces back to a holdover 00:09:11.340 --> 00:09:14.100 from the Red Machine. The prevailing historical 00:09:14.100 --> 00:09:16.519 consensus is that this architectural weakness 00:09:16.519 --> 00:09:19.179 began as a simple, mundane cost saving measure. 00:09:19.279 --> 00:09:20.759 Wait, I need to make sure I'm understanding this 00:09:20.759 --> 00:09:23.500 correctly. They compromised the security of their 00:09:23.500 --> 00:09:26.480 entire empire to save money. Believe it or not, 00:09:26.600 --> 00:09:29.559 yes. In the 1920s and 30s, international telegraph 00:09:29.559 --> 00:09:31.799 companies based their fees on the type of text 00:09:31.799 --> 00:09:34.419 being transmitted. If a telegram consisted of 00:09:34.419 --> 00:09:37.259 pronounceable syllables, it was often charged 00:09:37.259 --> 00:09:39.620 at a lower rate than a telegram consisting of 00:09:39.620 --> 00:09:42.980 just random, unpronounceable strings of consonants. 00:09:43.059 --> 00:09:45.379 Oh, you have got to be kidding me. No, I'm serious. 00:09:45.679 --> 00:09:48.220 By ensuring that vowels and consonants were encrypted 00:09:48.220 --> 00:09:50.580 somewhat separately, the designers hoped the 00:09:50.580 --> 00:09:52.759 resulting cipher text might occasionally group 00:09:52.759 --> 00:09:55.700 into pronounceable blocks, which could potentially 00:09:55.700 --> 00:09:58.309 qualify for lower transmission tariffs. That 00:09:58.309 --> 00:10:00.669 is unbelievable. So because that cost saving 00:10:00.669 --> 00:10:04.389 flow existed, the Americans had this tiny structural 00:10:04.389 --> 00:10:07.809 crack to pry open. Exactly. And in 1939, the 00:10:07.809 --> 00:10:10.090 United States Army Signal Intelligence Service, 00:10:10.269 --> 00:10:13.370 the SIS, decided to exploit it. They noticed 00:10:13.370 --> 00:10:16.690 this new, highly complex cipher system was replacing 00:10:16.690 --> 00:10:19.110 the old red traffic. Right, they saw the shift 00:10:19.110 --> 00:10:21.110 happening. And a legendary cryptologist named 00:10:21.110 --> 00:10:23.830 William F. Friedman was tasked with assembling 00:10:23.830 --> 00:10:26.110 a team to break it. And Friedman's team, I mean, 00:10:26.190 --> 00:10:28.210 they faced a monumental challenge. They were 00:10:28.210 --> 00:10:31.149 staring at reams of intercepted ciphertext with 00:10:31.149 --> 00:10:33.549 absolutely no knowledge of the physical machine 00:10:33.549 --> 00:10:35.490 that produced it. It's like a pile of gibberish. 00:10:35.830 --> 00:10:38.720 Right. But because of that 6 and 20 split we 00:10:38.720 --> 00:10:41.299 just talked about, statistical analysis quickly 00:10:41.299 --> 00:10:43.679 revealed a glaring pattern. When they counted 00:10:43.679 --> 00:10:45.980 the frequencies of the letters in the intercepted 00:10:45.980 --> 00:10:49.320 messages, they noticed that six specific letters 00:10:49.320 --> 00:10:51.399 behaved distinctly differently from the other 00:10:51.399 --> 00:10:54.539 20. The weaker encryption on the sixes basically 00:10:54.539 --> 00:10:57.669 betrayed their presence immediately. The SIS 00:10:57.669 --> 00:11:00.149 realized those six letters were essentially operating 00:11:00.149 --> 00:11:03.950 on a simple polyalphabetic cipher -like, just 00:11:03.950 --> 00:11:06.889 running through 25 fixed alphabets in a continuous 00:11:06.889 --> 00:11:10.129 loop. It was that simple. And by April 1939, 00:11:10.429 --> 00:11:12.950 the Americans had completely mathematically cracked 00:11:12.950 --> 00:11:15.690 the sixes. Which is a crucial first step, but 00:11:15.690 --> 00:11:18.009 it only allowed them to read about 20 percent 00:11:18.009 --> 00:11:20.509 of any given message, right? Exactly. The remaining 00:11:20.509 --> 00:11:23.389 20 letters, the consonants, were just an absolute 00:11:23.389 --> 00:11:25.570 brick wall. Those letters were being scrambled 00:11:25.570 --> 00:11:27.570 through the three stages of telephone switches, 00:11:27.570 --> 00:11:29.750 and the starting positions of those switches 00:11:29.750 --> 00:11:33.350 changed based on a daily key and a unique message 00:11:33.350 --> 00:11:36.200 indicator. So to penetrate that wall, the SIS 00:11:36.200 --> 00:11:38.799 team had to rely on cribs. Yes, cribs are essential 00:11:38.799 --> 00:11:41.440 here. For those who might not know, a crib is 00:11:41.440 --> 00:11:44.600 a cryptanalyst's best friend. It is basically 00:11:44.600 --> 00:11:47.940 a highly educated guess about the exact plain 00:11:47.940 --> 00:11:51.159 text that corresponds to a specific piece of 00:11:51.159 --> 00:11:53.840 scrambled ciphertext. And diplomatic communications 00:11:53.840 --> 00:11:57.220 are notoriously vulnerable to cribs. Because 00:11:57.220 --> 00:12:00.360 they're so formal. Precisely. The Japanese Foreign 00:12:00.360 --> 00:12:04.279 Office used this highly stylized, rigidly formal 00:12:04.279 --> 00:12:07.379 language. Messages would frequently begin with 00:12:07.379 --> 00:12:10.419 exact, predictable phrases like, I have the honor 00:12:10.419 --> 00:12:13.659 to inform your excellency. Ah, so if the Americans 00:12:13.659 --> 00:12:16.179 could align that known plain text with the scrambled 00:12:16.179 --> 00:12:18.679 ciphertext, they could start mapping the electrical 00:12:18.679 --> 00:12:21.340 pathways inside the machine. Yes. Furthermore, 00:12:21.419 --> 00:12:23.899 because Purple was deployed gradually, some of 00:12:23.899 --> 00:12:26.200 the smaller Japanese consulates around the world 00:12:26.200 --> 00:12:28.960 still only had the older Oh, meaning Tokyo would 00:12:28.960 --> 00:12:31.419 sometimes transmit the exact same diplatic circular 00:12:31.419 --> 00:12:34.440 twice. Yes, they did. Once in the new, supposedly 00:12:34.440 --> 00:12:37.639 unbreakable purple cipher, and once in the compromised 00:12:37.639 --> 00:12:39.879 red cipher that the Americans could already read 00:12:39.879 --> 00:12:43.320 fluently. It was a massive mistake. Having the 00:12:43.320 --> 00:12:45.320 translated red message sitting right next to 00:12:45.320 --> 00:12:48.000 the encrypted purple message gave Friedman's 00:12:48.000 --> 00:12:50.379 team the perfect Rosetta Stone. That's incredible. 00:12:50.440 --> 00:12:54.000 Even with those cribs, mapping the internal wiring 00:12:54.000 --> 00:12:56.899 of multiple cascading stepping switches without 00:12:56.899 --> 00:12:58.960 ever actually looking at the physical hardware. 00:12:59.480 --> 00:13:01.620 I mean, that sounds like an agonizing process. 00:13:01.620 --> 00:13:03.879 Oh, it was grueling. The pivotal breakthrough 00:13:03.879 --> 00:13:07.019 did not occur until September 1940, and it was 00:13:07.019 --> 00:13:09.759 driven by a brilliant cryptanalyst named Genevieve 00:13:09.759 --> 00:13:12.620 Gr - The sources really highlight Grachin's work 00:13:12.620 --> 00:13:15.700 is the absolute turning point here. She was staring 00:13:15.700 --> 00:13:18.240 at what the team called homologous messages. 00:13:18.659 --> 00:13:20.279 Right, homologous messages. These were messages 00:13:20.279 --> 00:13:22.139 that were sent on completely different days, 00:13:22.639 --> 00:13:25.399 but because of laziness or just operational error 00:13:25.399 --> 00:13:28.419 by the Japanese clerks, they utilized the exact 00:13:28.419 --> 00:13:30.360 same starting settings for the machine. Which 00:13:30.360 --> 00:13:33.840 is a huge security no -no. A massive one. Analyzing 00:13:33.840 --> 00:13:36.799 homologous messages allows cryptanalysts to compare 00:13:36.799 --> 00:13:40.240 the chaotic output of the 20s group across multiple 00:13:40.240 --> 00:13:44.669 texts. was essentially performing statistical 00:13:44.669 --> 00:13:47.250 gymnastics. She wasn't just looking at the letters 00:13:47.250 --> 00:13:49.990 themselves. She was looking at the physical distances 00:13:49.990 --> 00:13:52.490 between repeating cipher letters. So she's mapping 00:13:52.490 --> 00:13:55.549 the gaps. Exactly. And she discovered that the 00:13:55.549 --> 00:13:57.470 intervals between certain scrambled characters 00:13:57.470 --> 00:14:00.169 were multiples of specific numbers. Oh, wow. 00:14:00.309 --> 00:14:02.970 So by finding those repeating mathematical intervals, 00:14:03.350 --> 00:14:05.789 she proved that the seemingly random gibberish 00:14:05.789 --> 00:14:08.490 was actually generating a predictable loop. Yes. 00:14:08.610 --> 00:14:10.710 She found the mechanical period of the machine. 00:14:10.950 --> 00:14:13.980 That is amazing. She exposed the hidden mathematical 00:14:13.980 --> 00:14:15.919 cycles just from staring at the letters. She 00:14:15.919 --> 00:14:18.379 really did. Her discovery proved definitively 00:14:18.379 --> 00:14:20.460 that the chaotic permutations were the result 00:14:20.460 --> 00:14:23.279 of a specific number of switches moving at specific, 00:14:23.620 --> 00:14:26.500 measurable speeds. And shortly after Gratian's 00:14:26.500 --> 00:14:29.000 breakthrough, another team member, Navy Lieutenant 00:14:29.000 --> 00:14:32.039 Francis Raven, discovered the underlying protocol 00:14:32.039 --> 00:14:35.100 for the keys. He realized the Japanese were altering 00:14:35.100 --> 00:14:37.980 the machine's starting positions on a rigid 10 00:14:37.980 --> 00:14:41.639 -day cycle. with highly predictable minor daily 00:14:41.639 --> 00:14:44.639 shifts within that block. So the math is finally 00:14:44.639 --> 00:14:47.480 completely exposed, and with the mathematical 00:14:47.480 --> 00:14:49.600 rhythm of the 20 group figured out, the team 00:14:49.600 --> 00:14:52.500 realized the logic perfectly mirrored the operation 00:14:52.500 --> 00:14:55.039 of commercial automatic telephone exchanges. 00:14:55.279 --> 00:14:57.779 Which brings us to the hardware phase. Leo Rosen, 00:14:57.980 --> 00:15:00.159 who was an electrical engineer on the SIS team, 00:15:00.639 --> 00:15:03.279 quite literally went to a commercial telecommunications 00:15:03.279 --> 00:15:05.960 catalog, ordered standard uniselector switches 00:15:05.960 --> 00:15:08.200 like the exact kind used to route everyday phone 00:15:08.200 --> 00:15:10.179 calls, and just started wiring them together 00:15:10.179 --> 00:15:13.259 based on Grotchen's mathematical models. so audacious. 00:15:13.580 --> 00:15:16.600 By late 1940, through grueling methodical deduction, 00:15:17.000 --> 00:15:19.620 the SIS had constructed the Purple Analog. The 00:15:19.620 --> 00:15:22.379 Purple Analog. Yeah, it was this massive whirring 00:15:22.379 --> 00:15:25.320 assembly of commercial telephone parts just sitting 00:15:25.320 --> 00:15:28.159 in a laboratory in Washington, D .C. They could 00:15:28.159 --> 00:15:30.759 feed intercepted Japanese ciphertext into one 00:15:30.759 --> 00:15:33.720 end of this machine, and plain, readable Japanese 00:15:33.720 --> 00:15:36.019 would literally clatter out of the typewriter 00:15:36.019 --> 00:15:38.200 on the other end. Building the purple analog 00:15:38.200 --> 00:15:40.440 meant the United States could now read the minds 00:15:40.440 --> 00:15:43.519 of the Japanese government in real time. And 00:15:43.519 --> 00:15:46.980 this reality violently shifts our story out of 00:15:46.980 --> 00:15:49.899 the quiet laboratories of the SIS and onto the 00:15:49.899 --> 00:15:52.259 global battlefield. The shift is dramatic. The 00:15:52.259 --> 00:15:54.500 intelligence gathered from Purple and other intercepted 00:15:54.500 --> 00:15:56.899 Japanese communications was collectively codenamed 00:15:57.210 --> 00:16:00.350 Magic by the Allies, right? Right. Magic. And 00:16:00.350 --> 00:16:02.990 the geopolitical consequences of the magic intercepts 00:16:02.990 --> 00:16:05.909 were staggering. The most immediate and honestly 00:16:05.909 --> 00:16:08.049 tragic intersection of this intelligence with 00:16:08.049 --> 00:16:11.309 world events occurred in December 1941. The attack 00:16:11.309 --> 00:16:13.970 on Pearl Harbor. Yes. The Japanese government 00:16:13.970 --> 00:16:16.870 transmitted a 14 -part message to their embassy 00:16:16.870 --> 00:16:19.370 in Washington, officially terminating diplomatic 00:16:19.370 --> 00:16:21.870 negotiations with the United States, which effectively 00:16:21.870 --> 00:16:24.070 signaled war. And it was enciphered in purple. 00:16:24.330 --> 00:16:27.009 But because the purple analog was fully operational, 00:16:27.649 --> 00:16:29.889 U .S. cryptanalysts intercepted the transmission, 00:16:30.429 --> 00:16:33.470 decoded the 20s and the 6s, translated the Japanese, 00:16:34.070 --> 00:16:36.659 and had the finalized English text. in the hands 00:16:36.659 --> 00:16:40.519 of the U .S. government by 1 .00 p .m. on December 00:16:40.519 --> 00:16:43.059 7th. Which is incredible speed. They processed 00:16:43.059 --> 00:16:44.940 the message faster than the Japanese embassy 00:16:44.940 --> 00:16:46.740 scouts could even decrypt and type it up for 00:16:46.740 --> 00:16:49.440 delivery themselves. It's true. The tragedy, 00:16:49.440 --> 00:16:51.580 of course, is that while the cryptographic break 00:16:51.580 --> 00:16:54.940 was flawless, the speed of bureaucratic communication 00:16:54.940 --> 00:16:58.080 was not. By the time the final decrypted warning 00:16:58.080 --> 00:17:00.659 was distributed, the attack on Pearl Harbor was 00:17:00.659 --> 00:17:02.899 already underway. Just a heartbreaking delay. 00:17:03.149 --> 00:17:05.789 However, as the war expanded, the Purple Machine 00:17:05.789 --> 00:17:08.430 inadvertently transformed into one of the greatest 00:17:08.430 --> 00:17:10.910 intelligence assets the Allies possessed in the 00:17:10.910 --> 00:17:13.410 European theater. That is the big twist of this 00:17:13.410 --> 00:17:15.750 story. Right. The Japanese built a machine to 00:17:15.750 --> 00:17:18.390 protect their own secrets, and it ended up betraying 00:17:18.390 --> 00:17:21.309 the deepest secrets of their closest ally. We 00:17:21.309 --> 00:17:23.630 have to talk about General Hiroshi Oshima. Oh, 00:17:23.849 --> 00:17:26.740 General Oshima. He was the Japanese ambassador 00:17:26.740 --> 00:17:29.519 to Nazi Germany, and honestly, his role in the 00:17:29.519 --> 00:17:32.519 war effort cannot be overstated. He was deeply 00:17:32.519 --> 00:17:34.859 trusted by the Nazi high command. Which was rare, 00:17:34.960 --> 00:17:37.619 right? Extremely rare. Adolf Hitler, who was 00:17:37.619 --> 00:17:40.400 notoriously paranoid and rarely trusted anyone 00:17:40.400 --> 00:17:43.460 outside his inner circle, viewed Oshima as this 00:17:43.460 --> 00:17:46.440 ideological confidant. Hitler and his top generals 00:17:46.440 --> 00:17:49.220 would have incredibly candid detailed conversations 00:17:49.220 --> 00:17:51.740 with Oshima regarding German military strategy. 00:17:52.160 --> 00:17:54.359 And Oshima would take meticulous notes during 00:17:54.250 --> 00:17:57.390 these high level briefings, return to the Japanese 00:17:57.390 --> 00:18:00.890 embassy in Berlin and transmit incredibly detailed 00:18:00.890 --> 00:18:04.349 reports back to his superiors in Tokyo. All of 00:18:04.349 --> 00:18:07.210 it encrypted in the purple cipher. He was completely 00:18:07.210 --> 00:18:09.309 unaware that the Americans and the British were 00:18:09.309 --> 00:18:11.529 intercepting the transmissions, running them 00:18:11.529 --> 00:18:13.829 through the purple analog and reading every single 00:18:13.829 --> 00:18:16.799 word. Everything. Everything. General George 00:18:16.799 --> 00:18:18.880 Marshall, the US Army Chief of Staff, even stated 00:18:18.880 --> 00:18:21.259 that Ambassador Ishimu was their primary source 00:18:21.259 --> 00:18:24.099 of information regarding Hitler's strategic intentions. 00:18:24.980 --> 00:18:27.680 The level of detail Oshima provided was astonishing. 00:18:28.380 --> 00:18:30.640 I mean, he personally toured the Atlantic Wall, 00:18:31.220 --> 00:18:33.460 the massive coastal fortifications the Germans 00:18:33.460 --> 00:18:36.039 were constructing to repel an allied invasion 00:18:36.039 --> 00:18:38.839 of France. He sent comprehensive descriptions 00:18:38.839 --> 00:18:41.640 of Trump deployments, bunker locations and defensive 00:18:41.640 --> 00:18:44.720 strategies back to Tokyo. And the allies simply 00:18:44.720 --> 00:18:46.859 intercepted the transmissions and essentially 00:18:46.859 --> 00:18:49.299 received the German blueprints straight from 00:18:49.299 --> 00:18:51.140 the Japanese ambassador. Thanks for the tour. 00:18:51.319 --> 00:18:54.460 Right. Exactly. But Oshima also transmitted Hitler's 00:18:54.460 --> 00:18:56.720 confidential plans regarding the Soviet Union. 00:18:57.039 --> 00:19:00.359 And this introduces a critical geopolitical pivot 00:19:00.359 --> 00:19:02.779 point because the United States and Britain were 00:19:02.779 --> 00:19:05.519 not the only nations reading purple. Oh, this 00:19:05.519 --> 00:19:07.640 is fascinating. The Soviet Union broke it too. 00:19:07.759 --> 00:19:10.200 Yes. Through their own cryptanalysis efforts, 00:19:10.359 --> 00:19:12.359 the Soviets had also managed to break the purple 00:19:12.359 --> 00:19:15.559 system by late 1941. So Joseph Stalin is reading 00:19:15.559 --> 00:19:18.099 Oshima's mail as well. He is. And through these 00:19:18.099 --> 00:19:20.460 decrypted purple messages, which were corroborated 00:19:20.460 --> 00:19:22.579 by human intelligence from spies like Richard 00:19:22.579 --> 00:19:25.480 Sorge, the Soviets confirmed a vital piece of 00:19:25.480 --> 00:19:28.460 strategic information. Which was? Japan had definitively 00:19:28.460 --> 00:19:31.099 decided to target Southeast Asia and would not 00:19:31.099 --> 00:19:33.299 launch an attack against the Soviet Union's eastern 00:19:33.299 --> 00:19:35.940 border. OK. The impact of that single piece piece 00:19:35.940 --> 00:19:38.880 of intelligence on the Eastern Front is massive. 00:19:39.779 --> 00:19:42.299 Because Stalin knew with absolute certainty that 00:19:42.299 --> 00:19:44.160 his Eastern flank was secure from a Japanese 00:19:44.160 --> 00:19:47.160 invasion, he'd initiated a massive troop transfer. 00:19:47.380 --> 00:19:50.170 A huge gamble, but a calculated one. Right. He 00:19:50.170 --> 00:19:52.930 pulled his elite winter trained Siberian divisions 00:19:52.930 --> 00:19:55.829 away from the east and rushed them via the Trans 00:19:55.829 --> 00:19:58.609 -Siberian railway to defend Moscow. And those 00:19:58.609 --> 00:20:01.269 Siberian troops arrived just as the German panzer 00:20:01.269 --> 00:20:03.569 divisions were pushing toward the Soviet capital 00:20:03.569 --> 00:20:07.470 in December 1941. The reinforcements were decisive 00:20:07.470 --> 00:20:09.910 in halting the German advance and saving Moscow. 00:20:10.190 --> 00:20:12.630 It is entirely possible that the intelligence 00:20:12.630 --> 00:20:14.829 extracted from the Purple Cipher prevented the 00:20:14.829 --> 00:20:17.210 total collapse of the Soviet Union. That is just 00:20:17.210 --> 00:20:19.769 mind -blowing. But given the sheer volume of 00:20:19.769 --> 00:20:21.990 strategic leaks, it seems impossible that the 00:20:21.990 --> 00:20:24.069 Axis powers wouldn't realize their communications 00:20:24.069 --> 00:20:26.309 were compromised at some point. And this brings 00:20:26.309 --> 00:20:30.309 us to the most infuriating moment of the story, 00:20:30.410 --> 00:20:32.589 at least from the Japanese perspective, that 00:20:32.589 --> 00:20:34.450 Germans actually figured out the code was broken. 00:20:34.650 --> 00:20:38.490 They did. In April 1941, well before Pearl Harbor, 00:20:38.930 --> 00:20:41.289 a German diplomat stationed in Washington named 00:20:41.289 --> 00:20:44.690 Hans Thompson sent an urgent message to the German 00:20:44.690 --> 00:20:48.059 foreign minister in Berlin. Thompson explicitly 00:20:48.059 --> 00:20:50.480 warned that the Americans were actively reading 00:20:50.480 --> 00:20:52.980 the Japanese diplomatic ciphers. Where did Thompson 00:20:52.980 --> 00:20:55.140 even get that information? The leak was this 00:20:55.140 --> 00:20:58.180 remarkable chain of diplomatic whispers. The 00:20:58.180 --> 00:21:00.720 U .S. Undersecretary of State, Sumner Wells, 00:21:01.119 --> 00:21:03.940 vaguely hinted at the intercepts during a conversation 00:21:03.940 --> 00:21:07.240 with the Soviet ambassador to the U .S. The Soviet 00:21:07.240 --> 00:21:09.599 ambassador deduced the reality of the situation 00:21:09.599 --> 00:21:11.720 and subsequently leaked the information to the 00:21:11.720 --> 00:21:14.579 Germans. The German government deemed the intelligence 00:21:14.579 --> 00:21:17.019 reliable and immediately forwarded the stark 00:21:17.019 --> 00:21:19.859 warning to Tokyo. I mean, if your closest military 00:21:19.859 --> 00:21:23.160 ally urgently contacts you and explicitly states 00:21:23.160 --> 00:21:25.660 that your secure communications network is compromised 00:21:25.660 --> 00:21:28.400 and the enemy is reading your mail, the logical 00:21:28.400 --> 00:21:30.700 response is to immediately shut the system down. 00:21:31.019 --> 00:21:34.220 The logical response, yes. But the Japanese Foreign 00:21:34.220 --> 00:21:36.920 Ministry conducted a brief internal investigation 00:21:36.920 --> 00:21:39.339 and concluded that their cipher was perfectly 00:21:39.339 --> 00:21:42.019 safe. They just ignored it. They flatly rejected 00:21:42.019 --> 00:21:44.619 the German warning. They rationalized that the 00:21:44.619 --> 00:21:47.880 leak must have come from like a human spy operating 00:21:47.880 --> 00:21:50.539 within the embassy or that the Germans were simply 00:21:50.539 --> 00:21:53.500 mistaken. They refused to believe that an adversary 00:21:53.500 --> 00:21:56.019 could crack a machine as sophisticated as Purple 00:21:56.019 --> 00:21:59.400 without physically stealing one. Their blind 00:21:59.400 --> 00:22:02.019 faith in their own technological superiority 00:22:02.019 --> 00:22:04.819 completely overrode the empirical warning from 00:22:04.819 --> 00:22:07.730 their ally. Yeah, and because of that institutional 00:22:07.730 --> 00:22:10.250 arrogance, the Americans continued to read their 00:22:10.250 --> 00:22:12.529 highest level diplomatic secrets for the entirety 00:22:12.529 --> 00:22:15.089 of the war. The Japanese military establishment 00:22:15.089 --> 00:22:17.490 treated the technology as infallible. And there 00:22:17.490 --> 00:22:20.210 is a profound lingering irony to the physical 00:22:20.210 --> 00:22:22.710 legacy of this technology, too. In the final 00:22:22.710 --> 00:22:24.849 days of the war, as the empire was collapsing, 00:22:25.529 --> 00:22:27.809 Japanese embassy staff around the world were 00:22:27.809 --> 00:22:30.950 ordered to frantically destroy their purple machines 00:22:30.950 --> 00:22:32.769 to prevent them from falling into allied hands, 00:22:32.950 --> 00:22:35.180 right? Right. total destruction. They smashed 00:22:35.180 --> 00:22:38.259 them into unrecognizable pieces. The U .S. military 00:22:38.259 --> 00:22:41.440 only ever recovered fragmented burned components 00:22:41.440 --> 00:22:44.599 from the ruins of the embassy in Berlin. No fully 00:22:44.599 --> 00:22:47.359 intact original purple machine was ever found. 00:22:47.640 --> 00:22:50.160 The original machines were completely obliterated, 00:22:50.160 --> 00:22:54.099 yet the American built replicas. Purple analogs 00:22:54.099 --> 00:22:56.140 constructed out of standard telephone parts by 00:22:56.140 --> 00:22:58.140 engineers who had never even seen the original 00:22:58.140 --> 00:23:00.980 survived the war perfectly intact. It is a phenomenal 00:23:00.980 --> 00:23:03.119 narrative of cause and effect. A machine born 00:23:03.119 --> 00:23:05.779 out of inner service secrecy and paranoia between 00:23:05.779 --> 00:23:08.200 the Japanese Navy and foreign affairs. It was 00:23:08.200 --> 00:23:10.599 crippled from its inception by a budgetary decision 00:23:10.599 --> 00:23:13.759 to save a few pennies on telegraph fees by treating 00:23:13.759 --> 00:23:16.690 vowels differently. So preventable. And then 00:23:16.690 --> 00:23:19.470 it was painstakingly reverse engineered by brilliant 00:23:19.470 --> 00:23:22.049 crip analysts like Genevieve Grachin and William 00:23:22.049 --> 00:23:24.789 F. Friedman, who pulled mathematical cycles out 00:23:24.789 --> 00:23:27.910 of pure chaos. And ultimately, it served as the 00:23:27.910 --> 00:23:30.950 greatest unwitting spy of the 20th century, broadcasting 00:23:30.950 --> 00:23:33.910 the Axis powers deepest secrets right into the 00:23:33.910 --> 00:23:36.730 hands of the allies. It truly is a master class 00:23:36.730 --> 00:23:39.269 in the vulnerabilities of complex systems. But 00:23:39.269 --> 00:23:41.369 if we pull back from the history of World War 00:23:41.369 --> 00:23:43.509 II and look at our current technological landscape, 00:23:44.009 --> 00:23:46.250 there is a distinctly unsettling parallel here. 00:23:46.549 --> 00:23:50.190 How so? Today, our entire global infrastructure 00:23:50.190 --> 00:23:53.170 banking, military communications, personal data, 00:23:53.430 --> 00:23:56.230 it all relies on modern end -to -end encryption. 00:23:56.410 --> 00:23:59.990 We utilize algorithms like RSA, which rely on 00:23:59.990 --> 00:24:02.549 prime factorization. And much like the Japanese 00:24:02.549 --> 00:24:05.109 engineers who built purple, our modern cryptographers 00:24:05.109 --> 00:24:07.170 tell us that these mathematical fortresses are 00:24:07.170 --> 00:24:09.869 effectively unbreakable by any classical computer. 00:24:09.990 --> 00:24:11.990 Right. We've placed absolute trust in the math. 00:24:12.160 --> 00:24:15.579 We have. But the specter on the horizon is quantum 00:24:15.579 --> 00:24:19.099 computing. A fully functional quantum computer 00:24:19.099 --> 00:24:21.779 running Shor's algorithm could theoretically 00:24:21.779 --> 00:24:24.359 shatter our current encryption standards in moments, 00:24:24.680 --> 00:24:27.339 rendering our supposedly unbreakable systems 00:24:27.339 --> 00:24:29.759 entirely transparent. Just completely blow it 00:24:29.759 --> 00:24:31.960 wide open. Exactly. So it forces us to ask a 00:24:31.960 --> 00:24:33.980 very uncomfortable question. As we place our 00:24:33.980 --> 00:24:36.579 blind faith in the infallibility of modern cryptographic 00:24:36.579 --> 00:24:39.279 technology, are we simply setting ourselves up 00:24:39.279 --> 00:24:42.099 to be the victims of the exact same huge that 00:24:42.099 --> 00:24:43.640 doomed the purple machine.