Because if there are more than one module for the same purpose, then we can use that as an input to select a module.
Really? Here are a couple of reasons why it's a bad selector. Assume there are CPAN modules G and B doing the same thing, G in a good way, B in a bad way.
- If B has been there longer than G, the download count will be higher. If people use download count to determine what to download, people will download B, increasing its count.
- The author of B may do a lot of propaganda on Perlmonks and elsewhere, while the author of G doesn't give a rats ass how many people use his/her module.
- If a popular module P has a dependency on B, it will increase the download count on B.
- Redhat knows that G is good and B is bad. So it includes G in its distribution. Other vendors follow suit. Hence, large numbers of users of G actually never download G, not contributing to its download count.
- People that actually decide for themselves download B and G anyway.