Well, you certainly responded, not only in comments on the blog but also through emails and on Twitter. Your responses were very helpful and well-thought-out.
After reading your feedback and doing some research, I came to the conclusion that crowdsourcing can be a very effective tool. But only if it's used for a well-defined, focused portion of a project.
Crowdsourcing generally works best when you need a sampling of input from a large population. This can include activities such as requirements gathering, securing non-rights-protected content or a resource donation (such as computer bandwidth). Some mentioned software testing as a crowdsourcing activity. In this case, it's no different than what companies have always done when their products go "alpha" and "beta." People are simply slapping a new label on an old activity.
In any crowdsourcing scenario, the activities must be considered voluntary. There must be no compensation or contracts. And project participants must have a clear understanding that any contributions - tangible or intangible - are the property of the entity soliciting the input.
My rule regarding compensation for work could potentially be broken through a contest approach such as the Netflix Prize project, which focused on algorithms to enhance the company's ratings system. But activities like these would have to be tightly managed.
Are you or your company evaluating whether or not to foray into crowdsourcing? What types of projects will you use the activity for?