Study Shows Carriers Often Over- and Undercharge for Data
Try this: Does a certain coffee vender that just so happens to cover the nation in its drink bliss at an average of 4 - 5 dollars a cup "overcharge" it's customers for the drink experience?
This is going to happen in every industry. My Tshirt doesnt have the exact thread count. My car tires do not have the perfect amount of rubber. My CPU has a ~-5 to 5% overclock ability. The Phonescoop articles have a chance to error on the words used on the site (granted, they are fixed quickly and I dont pay to read the news)
Well folks, there you have it. If you want perfection, we need Phonescoop to make us a cell network. Im in!
T Bone said:
It's not the same no, he's not saying that it is exactly the same, he's saying that when you're counting massive amounts of 'stuff' that it is unreasonable to expect that the count will be 100% accurate and that a certain margin of error is expected...and he's right about that...
That's how I read it. Though, I agree that the analogy is still poor because when it comes to computers doing what they should be doing (computating) these errors should not exist. It simply means that the software (or hardware's firmware) has bugs and it needs to be fixed.
T Bone said:
Well, we are talking about hundreds, maybe even thousands of Terrabytes of data,,,,even a computer can't count that much data perfectly....a margin of error around 5% is reasonable...although I don't want to one of those who gets overcharged.....
Actually, no. In the computer world a 5% margin of error is just terrible... it falls under Class 0. The most common acceptable class is Class 4 which yields out a 0.001% margin of error.
Well, even a 0.001% error can lead to quite a few people getting overcharged. I didn't look very long, so I only grabbed a number from 2008, but back then AT&T had just shy of 100M subscribers. Even at only a 0.001% error rate, that's still 100,000 people with billing errors. Most of those are likely to be negligible, but I'm sure quite a few of those get quite messed up. Of course, that's assuming that only one computer handled all 100M subscriptions, and assuming they have no QA double-check before sending out bills. Neither of which are likely to be true.
1. 0.001% of 100M is 1000 (not 100,000).
2. I was inferring on the data error, not the number of subscribers impacted.
This forum is closed.