Exploiting the Tiltman Break - Computerphile
ฝัง
- เผยแพร่เมื่อ 4 มิ.ย. 2024
- Professor Brailsford returns to the subject of why Colossus was built.
The professor's notes:
www.eprg.org/computerphile/lor...
Bletchley Park Playlist: • Bletchley Park (Coloss...
Professor Brailsford used the book "Colossus" by B. Jack Copeland and others (Oxford University Press, 2006). Also recommended are chapters 18 and 19 from : "The Bletchley Park Codebreakers" by R. Erskine and M. Smith (eds.) Biteback Publishing 2011
/ computerphile
/ computer_phile
This video was filmed and edited by Sean Riley.
Computer Science at the University of Nottingham: bit.ly/nottscomputer
Computerphile is a sister project to Brady Haran's Numberphile. More at www.bradyharan.com
10:43 i love that prof Brailsford has a stack of oldschool perforated printer paper for scratch paper
Professor Brailsford never "stutters." He is a great storyteller. Thank you!
Aka David Attenborough of computer science
@@ronniebasak96 Each to their own and well spoken, on both counts!
I looked at the thumbnail and immediately thought "That's tunny traffic, must be a Professor Brailsford video!"
Now there's traffic analysis for you!
Professor Brailsford is always great!
I barely understand the subject matter most of the time, but what I do understand is fascinating. And he delivers it so well!
True, he inspires me to make videos myself.
Let’s have a video on Prof. Brailsford exploiting the Amen Break next ;)
I worked on teletypes in the Army in the '80. "The quick brown fox jumps over the lazy dog" was a staple in debugging the machines. The newest one I worked on was a UGC-74. Thanks for the trip back in time!
Go with a modern version - "sphinx of black quartz, judge my vow"
You know things just got real when Prof. Brailsford says "Oh dear! Calamity."
Great video! A commentor on an unrelated video wanted to buy gear wheels with a simple integer ratio and wondered why he couldn't find 16 to 32 teeth, but rather 16 to 31. Someone else replied that it spreads the wear. Now I know that it's because they are relatively prime, giving the maximum number of turns before the same pair of teeth meet again. Of course integer ratios are needed in some apps, but it's, well, interesting anyway.
Great for gearboxes, but useless for clocks where you need exact ratios.
Sometimes those exact ratios are fairly large (three digit) prime numbers, such as 223, as in the case of the Antikythera mechanism.
Or Clocks with sidereal dials.
Fascinating
Mal-2 KSC thx for me mentioning that device. I’d never heard of it.
I could listen to this man for hours. More please! (:
Bletchley is absolutely worth the days time and I even got a yearly admission for the price of a day ticket. Very cool. Check the opening times first, as the National Museum of Computing (which you will want to see) is not open all day, just the Colossus exhibition. Even got a peek at the Bombe project.
This video finally explained to me why prime numbers are cared about in cryptography. I hadn't thought that all the code-wheels moved at the same time so didn't think about the dangers of non-primes presenting patterns.
During my national service, I used Telex connections which had a similar sort of encryption scheme as described early in the video for the connection only. It used a tape which one had to load monthly, which had seeds on it for a pseudo random generator in the encrypter. Each day, at a set time per line, one would call the operator at the other end and advance the machines together, passing a short message back and forth to verify the machines were in sync. They would then spit out the previous day's seed, which we had to destroy.
Secret, but not extremely sensitive information would be sent as clear text across these secure connections, while the real secrets were sent as seperately encrypted text.
This great video suddenly gave me to wonder about the people in Germany and elsewhere who were designing and building these systems. We’ve seen many videos about the amazing processes and people at Bletchley Park, but I’ve never seen anything about the also-rather-amazing work on the other side. We’ve seen lots of information on other German (and less so Japanese) science and tech, such as rockets and tanks, but nothing on the “codemakers”.
Gary Bickford Did you google it?
Part of 'Operation Paperclip' maybe. These were scientists, rocket designers and engineers that the Americans 'poached' from Germany after the war to work for America. They were given a free pass, no trials and citizenship...no questions asked.
Another excellent video featuring professor Brailsford. Thank you!
Another interesting video spiced by Prof. Brailsford enthusiastic personality. Wonderful and thanks for it.
What an epic story!! I love how this guy just remembered that long string just like that
I love Professor Brailsford. I could listen to him yarn all day.
Fascinating and engagingly told. Thank you so much!
Bill tut was indeed a genius. My head would've exploded trying to decrypt these messages.
Incredibly easy to understand explanation of a complex topic. Thank you!!!
Bet once the code was cracked the Axis side weren't so zmug.
I have a question for you - Is there an archive of all the Enigma messages that were decrypted at Bletchley Park? Are there messages that were not archived? E.g. the Naval messages with 8 rotors. I wonder if it is worth decrypting these messages. They might give us some fresh historical insights.
I wouldn't be surprised if all those documents were among those destroyed at Bletchley at the end of the war.
The Americans having most of the bomb power, wasn’t most of it decrypted in the states?
@@2adamast Not at all. The US had its own establishment of code-breakers, which was called Arlington Hall. It was run by the US Army's Signal Intelligence Service. Their efforts were directed mainly against Japanese codes. There was a lot of information exchange between BP and AH. Cryptographers from both places spent time visiting and working alongside cryptographers at the other in order to gain experience and share expertise. AH is still a US government facility, but I think all the cryptographic services are located elsewhere now.
@@JMDinOKC Knowing that "The main response to the Four-rotor Enigma was the US Navy bombe, ..." they build 125 of 180 total. And that "The fast drum rotated at 1,725 rpm, 34 times the speed of the early British bombes." the US had maybe 20 times the British computational power.
Add to that the super fast Three-rotor Enigma bombs of the US Army
@@2adamast Most of the US cryptographers working to crack the Japanese naval codes were based at Pearl Harbor. Arlington Hall mostly worked on Japanese Army codes, and, beginning even before WWII ended, on Soviet codes.
The concept of co-primes.., at last I understand something about Chemistry and Bonding/wave-packaging, that was not previously obvious, because it's the inclusion/exclusion quantum boundary of common denominator resonance modulation, (-> Spinfoam bubble relative size-positioning in Totality).
Brilliant stuff, thank you all.
David Wilkie you have lost me there - what is the connection with bonding?
Excellent and very informative video. Love the professor
He reminds me of my Uncle Phil, who fancied himself quite the raconteur; except that Professor Brailsford actually is one.
Love these decryption videos!
We only ever hear about these things through the stories of the allies reverse-engineering them. We never hear the stories of the germans designing them. We just have speculations about what the designers were/weren't thinking. And I think that's really sad. I like the excitement of figuring it out, but I also like confirmation.
You sir are a legend. Great lesson
Always love Prof Brailsford's videos on the WWII code breaking at BP. Can't wait for the next installment
Oh dear, what a wonderful video
Very nicely explained. Thank you very much.
A few minutes in and I just realized that the effectively 1 data line from USB 2.0 gets XORed.
I just watched a video (I think by Ben Eater) about USB keyboards (and how they compare with PS/2 keyboards). In that video he explained that the USB standard (or at least the USB 2.0 standard) has 2 data lines and no dedicated clock line. The two lines normally send opposite signals, this helps with data redundancy and creates destructive interference in the EM produced by the cables so as to eliminate most of the EM emissions. As such USB 2.0 effectively has only one data line (there are a couple special states).
This one data line doesn't just send it's data as 1's and 0's though, instead it's a 1 if the state changes and a 0 if it doesn't (or maybe the other way around). In any case what this means is that the previous input is effectively XORed with the current input to determine the actual value (apparently this is more reliable at higher speeds than just sending the data directly).
Having binge-watched a selection of these, I'm now on a cliff-hanger for the next video..l
Thanks! ;-)
SAME 👍
24:28 when he says "can you imagine setting up all of these teeth" and points at the cartoon fish👌
13:33 we do the same thing while learning fourier transform
finding period of signal produced by adding up two (or more) sine wave
We find the LCM of period of all sine wave.
He's done the cube!
Why do you look like Jeff sokol
I could watch Prof B all day.
Wonderful...
Thank you for share
One of the key insights in counterintelligence is that just because you can't imagine how to do something doesn't mean your enemy can't (the converse is also true, but much less damaging). The Axis suffered from a failure of imagination when it came to its assessment of Allied cryptographic capabilities, which is odd given that they did indeed manage to break some Allied codes (notably, and most scandalously, the British merchant shipping code and the US diplomatic code that the hapless military attaché in Egypt used, to Rommel's great advantage).
Lovely permutations.
Where could I get the booklet mentioned at around 2:12 ? I'd like to read it too. Thanks!
Hi, the booklet I used was by Frank Carter and was put out in the early 1990s. You might want to check on the BP web site to see if it's still available. Failing that a revised and updated version of that material is in Appendix 9 of the "Colossus" book edited by Jack Copeland (see Info header for this video)
MichaelKingsfordGray Must he listen very carefully?
Fascinating.
Thanks for the video
Fascinating
Very inspiring.
ii am by no means a statistician or maths person but these are great, entertaining videos..great work.
the answer to life is 42 not 41 :) . Great video , great passion from him explaining this
Excellent video as usual from Professor Brailsford. IMO the story of Tunny is more interesting than Enigma.
Dr. Tutte was Chairman of the Department of Combinatorics and Optimization at the Faculty of Mathematics, University of Waterloo. when I was an undergrad. I knew the name but it wasn't my specialization.
11:06 I miss that old computer paper. My dad worked in nuclear research and had access to the best computer equipment the taxpayer could afford. But, being a relatively lowly paid British civil servant scientist he would bring some home for me and my brother to draw on. We grew up with that stuff all over the place.
I remember the paper as well.
It was present in our household as well
I love his videos, but I believe this is the most revealing one
perfect video
lol that rubics cube in the background. "but you can't break that cube can ya!"
The best monolog ever.
11:55 I recognize this pattern, that's what we call the round-robin!
THANK YOU SO MUCH FOR KEEPING THE SOUND OF THE MARKER OFF THE AUDIO TRACK!
I've had to stop watching way to many of your videos that are really interesting because that sound just drives me crazy.
You need to fix your problems instead of avoiding them.
3:28 That sounds like a One Time Pad.
Ideally the key does not repeat and is used only once. But reality is not always ideal. VENONA was broken because the Russians reused their OTPs.
Yes, and it's a rather humongous omission to not point out that a truly randomly assembled key tape would be fundamentally different from the key stream the video goes on to discuss, which is a _generated_ key producing the next value from the previous one through an (as complicated as they can manage) rule. This very difference is what makes the latter crackable while the former _is not_. Of course, it also implies the former needs a copy of your completely random tape to decrypt, while the second only needs the starting settings...
@@michaelsommers2356 it's always the humans being lazy...
love it
it's the lock of a safe encrypting the space inside!
Give random number sequences random letter values then run the current data stream until real words are provided.
I've read and watched at least a dozen accounts of what Bill Tutte got up to, and each time have gone away with nothing more than "He must have been very clever."
But this time I actually get it !
So what is the Tiltman Break? 8:58 is mentioning some break, but was it explained in the video before that?
The break was that the same but just slightly different message was sent twice > this means you got a double length keystream, and by decoding the messages, you got the keystream back too, that was used to send it. You then had to use statistics and pattern recognition on it and you could theoretically get back the logic of the machine that generated the keystream. And this was what John Tiltman and the rest of the Bletchley cryptanalysts did.
See the previous "Fishy Ciphers" and "ZigZag decryption" videos which you'll find on the playlist
You are the best
Reminds me a lot of the mechanical calculator
What's that weird faint musical beeping sound that starts at 10:13 and continues to 10:28?
That is the sound a Samsung washing machine makes when it's done :D I was quite amused about that.
Too bad we will never hear the cryptography stories from the other side of the effort! Great piece, as always.
I thought for a moment someone was creepin in on you through the window but it was just a mirror
Why didn't they just have a uniform sync bit down each tape.
ahh form feed paper :) happy days
Why didn’t they just put a “1” start and stop bit, then they would have had a row of sprocket-feed holes on each tape? Then they could have fed both (tapes) through, much like in a “modern” dot-matrix printer with the sprocket-feed engaged.
1:10 Oh. Hey there Mr camera man!
Side note: Great topic, great video :)
i can only imagine the amount of eyerolls received by whomever came up with 'relative primes'
6:30 sorry but I don't understand why nobody got the idea of superposing the encrypted message tape and the key tape, then punching holes where the two were filled instead of syncing them side by side. This way 1 xor 1 = 0 (newly punched hole) and the rest works like XOR (1 xor 1 yields a filled hole, 0 xor 1 and 1 xor 0 also, and 0 xor 0 yields a hole which is what XOR would do).
Is this idea useless or has it been tried ? Why wouldn't it work ?
I don't know either - 35m film has holes on each side for synchronization, so it was possible.
Superimposing two tapes gives an OR, but you could use relative opacity to isolate XOR.
Anyway, in order to decrypt, you would still need to be able to synchronize the cipher with the key, whether aligned side by side or layered.
If by "pure luck" you mean "a well-developed debugging technique" then yes he found the diagonal pattern by "pure luck"
my thought as well when Brailsford said that. "Wait, that's how I fix most 'hard' bugs"
Or a case of 'fortune favours the prepared mind' .....
The pure luck wasn’t magically finding a pattern, it was that 574 is one less than 575 so he found an unrelated pattern on his first go. It just saved some work since as noted 575 would’ve shown patterns in stream 5.
He just means how the streams for the 23 and 41 cogs were one off for his 575 period. Not lucky that he found the pattern, just lucky that the two patterns coincided there. He would have had to check the fifth stream to fins the pattern otherwise (which he probably would have done eventually)
Did the Enigma support "ß" as a character, or did they just type "ss"?
It was just upper case letters, only 32 posible values.
It was A-Z only
I'm pretty sure they typed SZ.
Only 26 letters in the alphabet. Where did you get 32 from.
Though, looking at an image of an enigma machine on wikipedia, there are only the 26 standard letters on it.
Glad to see your Cube in a solved state in this video. I'd like to see you educate whilst solving your Cube.
a little bit too much was cut, what is zigzag decryption? also: 26 is not a prime
try this video: th-cam.com/video/yxx3Bkmv3ck/w-d-xo.html
You're right it's not prime, it's "relatively" prime compared to the other numbers.
Nice Robert's Radio in the background, I see :-)
What do you reckon it's tuned to?
@@twotone3070 The BBC Home Service. Probably listening to Workers' Playtime :)
Search for Curious Marc videos to see teletypes in action, among other things
Amazing, even by nowadays standards!
Hang on, so, how is this relevant to the Colossus, and what is the Tiltman Break anyway? As fascinating as was anyway, I have a feeling the wrong title and description has been put on this video.
The interesting bit is how these early cyphers totally relied on them being too difficult for humans to decrypt by hand. Using nothing more than Excel, figuring out repeating patterns in streams of bits takes no effort at all.
There is no sense at all in these ciphers being "early". People have used ciphers for thousands of years.
Yeah I know what you mean, I guess every cryptographic system relies on being too difficult to brute-force using current technology. I suppose back when the Romans were ROT-13ing things, the general population couldn't read so something a modern school kid can figure out must have been "too difficult".
At one point RSA and MD5 hashes were considered secure. Now we have computing power capable of rendering them useless.
I wonder if there'll ever be a time where the computation required to encrypt something securely outweighs the security it offers, but at the same time the technology to break the encryption will catch up.
"we have computers capable of rendering [RSA] useless"
Yeah, no.
They were no ordinary humans.
Excel is impressive. Another aspect to remember about encryption machines is that they must be easy to use and field-maintainable. Lorenz may be a for high level cyphers only and thus be an exception. But the Enigma was everywhere, on every boat, with every unit. It had to be easy to use by relatively uneducated people. Similarly, in WWI, soldiers in the trenches used pretty simple cyphers even though better ones were available. But nearly-illiterate farmers cannot do math with artillery shells exploding, bullets zinging by their heads, and while a gas mask.
Nice video
Imagine the random tape breaking during transmission . . . while a complacent operator had a quick smoke break . . .
Friggin amazing professor. I partially understand it. Lol
There is a logical error in the graphic at 6:44 line 5
But you cannot label all 41 teeth of that wheel with 25 letters. So how could an alphabetic indicator work? Did the operators use only 25 out of the 41 possible starting points?
This sounds like the birth of the automated "rolling code" encryption. Is this right?
23 * 25 = 575. Let's investigate.... Dusty went to Italy, xor with a plane ticket to Ztut, its a holiday in the Italian Alps!!
I've been inspired to watch The Imitation Game
You'll be disappointed at the inaccuracy.
@@twotone3070 This was a year ago but I did end up watching it. It was okay, the stupidest thing though was at the end when they implied he was "in love" with the machine or something ridiculous like that.
12:45 quick maffs
The Americans used a machine during the war that had a tape key. At a speed of 10 characters per second (CPS) - synchronizing is easy. Much above that and you can't do it electro-mechanically. To read a tape going 5,000 CPS photocells are going to be required. Back in the early 8080 days you could build a kit paper tape reader that used photocells. It could do 5,000 CPS if you could pull the tape that fast.
Collosus used an optical tape readers at 5k/s.
What happens next
3 times 2 is 6, instant maffs! 12:44
5:55
6:02 heh
Part of this explanation sounds a bit unlikely. 5-hole paper tape ALWAYS has a sixth sprocket hole. I've seen and used 5, 6, 7, and 8 bit paper tape and they ALL have sprocket timing holes. Now the colossus tapes look like they don't have timing holes, maybe that is what he his thinking of?
The Colossus tape mechanism didn't rely on the tape's sprocket wheels for transporting the tape. But it did use them - with a photoelectric sensor behind them - to provide the fundamental clock pulse for Colossus itself. In this way the Colossus computer was always in sync with the data it was reading
That has been true of optical paper tape readers in general. I had the privilege of playing with an obsolete Ferranti Mercury which used paper tape for all I/O. The reader used a rubber capstan to move the tape, and 6 photocells to read it, with the clock being provided from the sprocket hole.
dothemathright 1111 The teletype machine has no electronics - it is a mess of gears and relays. Magnetic logic in short.
Well, apparently cryptography is way more interesting than my c# high school class.
Oh, what a tangled web we weave when first we practice to deceive...