Isn't the purpose of a cipher to be able to encode and decode a message? This only goes one way (encoding), but it's impossible to decode the encoded message. Seems to defeat the purpose of a cipher...

Nav to the discourse page of your already solved katas, then click Show Kata Test Cases, and you will see how random tests are usually setup.

I see. Do you have a link I could look at? I understand the concept but I'm not sure how that would work or be implemented.

public class TransposeRandom {

  [Test]
  public void Test1() {
    string message = "PAmuapYIR cBANYpCm lETQoY";
    string output = TransposeEncryption.Encrypt(message, 4); // True
    Assert.AreEqual("parnmq apcylo mybpey uiact", output);
  }

Nope, this is not random test, it's still fixed test.

Random does not mean

getRandom()
{
    return 4 // chosen by fair dice
}

It means you need to generate random cases on the fly.

Added some random tests.

Needs Random Tests.

Difficulty is 5 kyu or 6 kyu.

Adding a sample test would make this easier to understand. Otherwise it's a good kata.

The spec for this kata is described terribly. I came up with 6 completely different solutions that can be interpreted as the answer, before finding the right interpretation.