Why is "bad data" such a common occurrence?

Surely the exchange knows exactly what price the stock was traded at, and the system is automated from then on, right? How do mistakes get into the data so frequently?

One would think that since all of this is computerized (online broker order to exchange for execution to data provider) then there would be reliable transfer of information without bad data. But not all trades occur that way so here's my guess at one reason why.

A cross trade is when a broker executes a matching buy and a sell order for the same security between two of his clients. He then reports the completed trades to the exchange (manual entry). It's quite possible that this could be one source of data entry error.

When it comes to a secondary source like Yahoo, they sometimes just have bad data. I have seen instances where a stock trading around say $150 suddenly trades near $60 for a few days or more and then reverts back to its normal $150 data. That I would blame on the service (programmers?).