This topic shows up from time to time. My take on it is that the Public Mobile data meter is "legal tender" so to speak. Like the scale at the grocery checkout, it is subject to quality control. Personally, I use the data witness app for monitoring usage. From renewal to renewal, I find my tracking to be close enough to what is logged in my account. The differences may be due to round off. I am not the least bit concerned that the official data meter is cheating in any way.
On the topic of round off, I recall when I was a Wind customer the round off when roaming was so skewed in favor of the carrier. Whenever the flow of data stops for a short period of time, the meter would stop counting and assign a usage rounded up to the nearest MB. Over a 24 hour period, the numerous sync operations that use tiny amounts of data would lead to large amounts of chargeable data when the actual data usage without roundoff would be tiny. In that case, it was the coarseness of the roundoff that led to big discrepancies. I can't remember where I read about how the Public Mobile meter does roundoff, going by memory the roundoff was in kB, so the accumulated error should be reasonbly small.
