Yes, but people are saying that it's beyond impractical. It's beyond improbable. It is flat out mathematically impossible on the realm of A = !A impossible.
So if I randomly selected one of those possible answer than there is 0% chance of me getting the correct one? Not merely a incredibly imporbable one. Not merely a impractical one. A logically impossible chance of me selecting the correct answer at random?
The claim by people here is that it is mathematically impossible. So far, it doesn't seem that way to me. Which is strange, because everyone is so insistent that solving a one time pad generated message back to its plaintext would be equivalent to violating the law of noncontradiction.
Let's look at how one time pads/running key encryption work.
Imagine I have a message: Happy Days
I can translate these message to numbers using the following scheme (for the sake of simplicity, assume we can tell the difference between a 1 and a 2 and a 12. In real life this would all be binary anyway).
* 0 A 1 B 2 C 3 D 4 E 5 F 6 G 7 H 8 I 9 J 10 K 11 L 12 M 13 N 14 O 15 P 16 Q 17 R 18 S 19 T20 U 21 V 22 W 23 X 24 Y 25 Z 26 Space 27
Our message becomes 8-1-16-16-25-27-4-1-25-19
Now anyone knowing the above code can turn this message back into English and even if you don't know the code, you can figure it out because English has exploitable patterns. For a message of a large size, for example, the most common letter will be e.
So we want to protect this. To do this we use a running key - our one time pad. This is a message of randomly chosen letters the same length as out message. Let's go with: JSYJMS DGJ
In numbers this is 10-19-25-10-13-19-27-4-7-10
Now to encrypt our message using this one time mad, we add them together using modulo 27 arithmetic
08-01-16-16-25-27-04-01-25-19
+++++++++++++++++++++++++
10-19-25-10-13-19-27-04-07-10
==========================
18-20-14-26-11-19-04-05-05-02
To get our message back, we just do the opposite
18-20-14-26-11-19-04-05-05-02
-----------------------------------------
10-19-25-10-13-19-27-04-07-10
=========================
08-01-16-16-25-27-04-01-25-19
Happy Days
If you know the one time pad you can get the message back. That's a basic law of cryptography: if you put the correct key in, the correct message comes back. It has been broken. With most encryption schemes and messages of sufficient length (cardinality of English I think; It's been a while since I took this exam so I'm a little rusty), you know you have the correct key because the message suddenly starts reading as English.
This is not the case with a running key. By trying different random strings, you can get any message out. Some will be English, some won't be. One will be the true message but you have no way of telling which.
Say for example, I decide my one time pad was 12-26-00-26-09-18-19-20-17-9.
Using the process above I do the following
18-20-14-26-11-19-04-05-05-02
-----------------------------------------
12-26-00-26-09-18-19-20-17-9
=========================
06-21-14-27-02-01-12-12-15-20
This translates to Fun ballot.
As you see, I can have two running keys, both produce valid messages and I (Eve the attacker of course) have no way to tell which the real message is.
Now, there are some ways you can attack a onetime pad in SFF. Time travel is a big one (so don't go relying on your one time pad in the Xeelee Sequence). If you can record one copy of a message encrypted with the one time pad, jump back in time to after the one time pad was selected but before it was used to encrypt a message, act such that the message being sent is changed and record that second message, you can break the pad by comparing the two copies. You can do the same thing if you can view branching parallel universes and so on. Of course that's not very relevant to Exalted.
To break a one time pad in Exalted, we have to assume that the charm can somehow know the 'correct' message from among all the other equally valid messages.
So Flawless Diagnosis Technique is mathematically impossible now? Look, if I had a phone line to all possible futures, I could not decode a one time pad without somehow getting my hands on both components at some point. If I had the same phone line, I could diagnose the patient by simply killing him again and again in those possible futures until I got the correct information.
You could if you were calling from after the one time pad was created but before it was used to encrypt a message by examining two futures where the same pad was used to encryption different results (Lunars Attacked my Mansion vs Flood destroyed my crops say). Or, if you could do it by calling up the future when the use the pad a second time for some reason.