The link provided by Gizmo strikes me as worthwhile here.
I am inclined to say that it is a little doom and gloom when discussing the various options to document first to invent. But then again, while the US's first to invent system works well theoretically to guarantee "opportunity," interference proceedings are rare and it is rare that "first to invent" really works in somebody's favor. Between 1991-2000, there were more than two million US patent applications filed, of which less than 2,000 led to interference proceedings, of which, traditionally, the "first to file" party tends to win about 75% of the time. And theoretically, even where a "first to invent" party comes out on top, the validity of their resulting patent can still be questioned in the event they move to enforce it.
Here is an article on the topic, which does a good job of documenting why "first to invent" should not provide a significant amount of security to inventors, and which at least begins to explain why those claiming a switch to "first to file" would hurt small inventors are arguably at least a little bit off base.
Back to the idearights.com article, while it mentions that the disclosure within a provisional app must be detailed and "enabling," it fails to mention that the disclosure contained within any "verifiable records" must still meet the same requirement in order to serve as proof of invention, and not just a "defensive disclosure" (which can still serve as blocking prior art for others).
Lastly, re:provisional applications. When inventor-written, they certainly do tend to provide a false sense of security. When attorney/agent written, not so much. However, if you are going to pay an attorney/agent to draft a solid provisional app, in many cases it is simply more cost effective over the long run, and potentially contains added benefit (earlier publication/issuance can mean earlier protection) to file a non-provisional app right out of the gate.