January 29, 2008

Change in How We Calculate Statistics

1m read
Several times in the last year, we noticed very large spikes in the number of downloads on this website. We have often traced this phenomena to a single IP address (which usually means a single computer or a small group of computers in close vicinity) that appeared to be downloading the same recording over and over and over again, thousands to ten-thousands of times in a single day. However, no two spikes appear to have been from the same IP address. We haven't been able to trace the source of the problem, whether it's from buggy download managers, individuals trying to interfere with our statistics, or other phenomena. There have also been smaller spikes that we never noticed but were of a similar nature. Whatever the source of the problem may be, we deem these events to be erroneous at best, and malicious at worst.

The good news is that we have finally implemented a solution. From now on, our website will only report the number of downloads per media file, per IP address, per day. Thus, the most that any faulty computer program or malicious visitor could skew our statistics is by downloading every single media file on our website once per day. While this could still visibly affect our statistics, we believe this will yield numbers that are much closer to the intended meaning of what we report (which is the number of different people downloading each media file.)

This change is retroactive back through December 7th, 2007, which is when we launched the last major revision of this website. We will also apply this strategy to the numbers prior to this date, if we can find a reasonable way to do so.

Happy downloading!