This is misleading. A flaky machine will indeed see bit errors, and it will probably be visible as random crashes, but that's not even necessarily the case for the average machine. If you look at the quantitative study from Google, which the author links to:
http://www.cs.toronto.edu/~bianca/papers/sigmetrics09.pdf
... you can see that in terms of errors per DIMM:
"Across the entire fleet, 8.2% of all DIMMs are affected by correctable errors and an average DIMM experiences nearly 4000 correctable errors per year. These numbers vary greatly by platform. Around 20% of DIMMs in Platform A and B are affected by correctable errors per year, compared to less than 4% of DIMMs in Platform C and D. Only 0.05-0.08% of the DIMMs in Platform A and Platform E see an uncorrectable error per year compared to nearly 0.3% of the DIMMs in Platform C and Platform D. The mean number of correctable errors per DIMM are more comparable, ranging from 3351-4530 correctable errors per year."
So the mean rate of correctable errors is high, but the variance is also very high: depending on the manufacturer, 80 to 96% of DIMMs see out a whole year without a single correctable error. If the original statistics of 95% chance of error in 3 days were correct, a single-DIMM machine ought to have approximately 0% (astronomically close to 0%) of living out a whole year without errors - but we can see here that you seem to have between 80 and 96% of single-DIMM machines doing just that.
The moral here is to test your memory for a while - preferably a few days - before trusting the DIMMs. But once you know you have good DIMMs, it doesn't look like you need to be quite so paranoid about bit errors.