I don't have the time to research where I heard this, but I recall a UI focus group study that found pretty much equal user satisfaction between accurate linear

I don't have the time to research where I heard this, but I recall a UI focus group study that found pretty much equal user satisfaction between accurate linear progress bars and random progress bars, but universal dissatisfaction with progress bars that "reset". My own feelings mirror this finding.