Skype for Linux has been able to send SMS messages for 10 years or so. What's new about this?

In economics, it's rather difficult to run experiments. For example, we can't ask President Bush to overthrow Saddam Hussein's regime 100 more times so that we can get statistically significant data on its effects on unemployment! The next best thing is to look for "natural experiments" that generate data that is just as good.

In this case, the authors wanted to experiment with the supply of P2P "servers". If the number of people offering to share P2P songs increases, then they hoped to measure the corresponding decrease in music sales. Naturally, they didn't actually run this experiment -- the RIAA would have objected to the researchers engaging in piracy!

So, they did the next best thing -- a natural experiment. When German students have their vacation, they share more files. This makes it easier to download songs in the US. They found easier access to P2P file sharing in the US did not cause a large drop in music sales.

Of course, this natural experiment isn't as good as a real experiment:

• German vacations aren't randomly assigned -- but the authors argue that since German holidays vary in each state, that there is enough variation to get robust results.
• The "German vacation" effect on the ease of downloading songs might be too small to measure anything reliably. Germans P2P users supply only one sixth of all songs that US users download. This "experiment" might be akin to doing a drug trial with a very small dose of medicine. In this case, you need to collect a lot of data to reliably measure the effects.

There are many other concerns discussed in the article. But, this is probably the best evidence we are ever going to get.