Codebusters C

[dragonfruit35]

Looking for opinions here, I'm curious what everyone else thinks of the rules. I've always wished they'd lean more towards the cryptography/math intensive problems and shy away from the monoalphabetic tasks (I see them as more repetitive, less thought provoking). Each year so far the rules have moved away from what I'd like though, so I'd love to hear other thoughts on it.

I agree - I wish there were less of an emphasis on the monoalphabetics. All it does is encourage more memorization when this event could have a much greater focus on being able to effectively encrypt/decrypt a variety of cipher types relatively quickly using advanced techniques, such as matrices and modular arithmetic (a good example of this is RSA or the Hill Cipher).

Are you guys referring to Affine/Caesar/etc or aristos/patristos? If you're talking about aristos/patristos, I have to disagree. If anything, it's the math ciphers that get memorization-heavy, as there are limits on the variations of questions that might appear, meaning that you're basically just memorizing the methods for solving the possible problems that could appear and plugging in numbers. On the other hand, aristos and patristos are always different and you can get better through practice, but you can never be certain that you'll get them correct. On math ciphers, if you know how to do them, you can be reasonably sure that you can get them correct, and if the event is largely math ciphers it's just "who can do math fastest", which I don't feel should be the entire point of the event (plus, this will lead to closer scores, meaning that results could frequently come down to the timed question). They should definitely be a part of it, and having a large variety of ciphers is nice, but the hard work that people put into this event really comes through on aristo/patristo questions.

[iwonder]

The distinction to me is that with the aristocrat style ciphers, you can get better with practice, but there's not a large amount of study that could be done to get better. With mathematic style ciphers, you're right, it's less practice (more study), but I would argue that if all you see is the same half dozen formulae used over and over, it's a poorly written question. The mathematics of cryptography is a very deep and quite intense subject (just look up proofs for the Chinese remainder theorem) and I think that type of mathematics is more suited to a Science Olympiad event than the much less technically rigumorous aristocrats.

Certainly though, this year, in what I believe is the effort to shy away from the rigour, the rules put a few more limits on what can be asked from the mathematics side of the picture. It'll be interesting to see what the tests look like this time around.

[Longivitis]

I have to agree with dragonfruit here because in theory if all teams memorized how to do all the math-based ciphers and can do them quickly enough, then the only thing that will set them apart would be the timed question and the aristos/patristos which can make or break a couple of teams. I'm leaning more with a 50/50 of aristos and math-ciphers so teams that focus on one type are not limited as much.

Of course, learning how to solve basic aristos and patristos can be difficult, last year I was the one on my team capable of doing them and I had my other team members do the math-based ones, but I believe with more practice on them people will get better at them, but it's not really something you can just study for hours and get good at it because each one is different.

I say the ideal composition of that one member can do the hard math ciphers and okay at the others, one that is okay at both, and one that is good at aristos/patristos so that the workload is ideally spread around because with tests averaging around 20 questions and taking maybe 5-10 minutes to solve one it is better to split on the easy ones and work together on the last few hard ones like no hints patristos.

[will0416]

Gotta agree with dragonfruit too. Although it's definitely true that the monosub variants are less "sciency", the ways in which you can formulate a question for them are definitely more abstract than how the rules restrict questions on ciphers like Hill. If there were an imbalance favoring these "mathy" ciphers that are definitely more reliant on memorization, it would just become a matter of who can plug and chug the fastest, since most (if not all) competitors are using the same methods.

Also there's a lot more to monosubs than a lot of people care to look into

[pepperonipi]

Gotta agree with dragonfruit too. Although it's definitely true that the monosub variants are less "sciency", the ways in which you can formulate a question for them are definitely more abstract than how the rules restrict questions on ciphers like Hill. If there were an imbalance favoring these "mathy" ciphers that are definitely more reliant on memorization, it would just become a matter of who can plug and chug the fastest, since most (if not all) competitors are using the same methods.

Also there's a lot more to monosubs than a lot of people care to look into

Good point - I rescind my argument.

Also, shouldn't this thread be in the Lab Events section of the forums, or did this event become a study event for the 2020 season?

[iwonder]

Thanks for all the insight guys, I appreciate it. I agree that the rules place fewer restrictions on the aristocrat style ciphers, and I like the point that they could be more creatively asked. Maybe next time we can get a ruleset that opens the floor to math related cipher questions that y'all don't feel are just plug and chug. Modern cryptography is a really interesting field that encompasses a huge subset of 'pure math' that a lot of engineers don't get exposed to. I'd be excited to see some of it in SciOly.

[gz839918]

I'm about to both disagree with all of you here in some part or another, so I'm sorry if I sound a little touchy since I don't mean to be overly contentious here

No modern scientist truly has a need for cracking monoalphabetics without a computer, but aristocrats, patristocrats, and xenocrypts are hardly empty of scientific rigor. Proficiency at monoalphabetics doesn't necessarily have to come from just repetition and blind effort; by analyzing the patterns of language, you can predict what letters and words will appear in what places. Awareness of language patterns, in turn, requires science to learn, and by learning it helps science.

An example: if in a patristocrat you see ?he?he? where question marks represent unknown letters, this is more likely to decrypt to "then he" and not "he then". Why? This is because (most) people are more likely to use conjunctive adverbs like then and however before nouns and not after nouns. This factoid may seem obscure, but forensic linguists have actually used this knowledge to exonerate a man with a murder charge posthumously after the forensics team suspected that police faked his confession of guilt because the confession statement contained too many instances of I then yet much fewer of then I. Aside from forensics, linguistic knowledge is finding a niche in research on Natural Language Processing (the stuff that makes online translators work) as well as spell checking and text prediction. So yes, codebusters is useful, if only by teaching you language facts you'd be unlikely to encounter otherwise. It's not the finding of the message in a monoalphabetic that is valuable, but rather, it is what could be learned in the process. In that way, being a better scientist could help with any events, and all events are scientifically useful, but to varying degrees.

On a different note,

Maybe next time we can get a ruleset that opens the floor to math related cipher questions that y'all don't feel are just plug and chug. Modern cryptography is a really interesting field that encompasses a huge subset of 'pure math' that a lot of engineers don't get exposed to. I'd be excited to see some of it in SciOly.

I would be excited for an event that favored cryptographic theory over mere practice, and that would address the ongoing shortage of computer science events. However, that would mostly require students with strong math skills. RSA is hard for many college students to understand; if we mixed in the Chinese Remainder Theorem, Fermat's Little Theorem, Euler totient functions, and many more, I'm not sure how Div C students would handle it (or if it would even be possible to move the event into Div B). If it were balanced by having both content knowledge (like understanding the intuition of why RSA works) and math ability, I would gladly support the idea for such an event.

[iwonder]

Chinese Remainder Theorem, Fermat's Little Theorem, Euler totient functions

Man those proofs still give me nightmares...

[Blank25]

I've never done Codebusters before and I'm wondering how you are able to decide which cipher to use. Is it given usually? Or do you just have to somehow figure it out? And is this event really hard?

[Name]

I've never done Codebusters before and I'm wondering how you are able to decide which cipher to use. Is it given usually? Or do you just have to somehow figure it out? And is this event really hard?

Ciphers are given. Look at toebes for how each question works because the majority of tests are written with toebes. Hard is very subjective. Would xenocrypts be hard if you don't know spanish? Probably. Would someone good at spanish find xenocrypts hard? Probably not. For the most part if you're good at the timed cipher, you will probably do somewhat well. Are aristocrats hard? Not really, but it takes a significant amount of effort to make even small improvements in your time.

This is just my opinion, but I found code very easy to learn, but impossible to master.