Actually, I would say that is incorrect. The randomness means that you will see a lot of "spread" around the expected target time to generate a block, but over the long term, you should DEFINITELY see the expected "a block each X hours/days" ratio emerge from the data. The bitcoin calculator website will tell give you this data based on your khash/sec. I'm sure you know these things, but I'm stating them for the benefit of the new poster who I think might be misled by your statement.

The difficulty level has just increased again substantially, so it has become even harder for a user with relatively modest computational resources to personally verify their client is interacting with the generation process across the network correctly. I feel this is a bit of challenge for both building user confidence and trust and also the debugging process. Recently for instance from version 0.3.6 to 0.3.8 coin generation was broken in 64-bit linux clients, but this bug took awhile to be detected despite users running that client version, because the bug symptom of "not generating blocks" was more or less identical for the user to the expected behavior of generating blocks very slowly due to the high difficulty!

Anyway, even when everything is working correctly there is no guarantee your behavior will match statistical expections apart from over the very long term. At current difficulty, a standard desktop's expected time to generate a coin block may be a few weeks even with 24/7 operation.