According to Data.gov, at time of publication the Department of Labor has 34 datasets. That’s fine and quite useful. But the Department of Labor has an FTP site. There are more than 34 datasets there (just counted them again and there are 58). So you would think that there must be some fancy setup that means that some datasets are simply not polished sufficiently to qualify for listing at the data.gov site. You would be wrong.
One of the 34 datasets is simply a pointer to FTP site. There is no reason not to do this for the missing 20+ datasets. It is a bit of a mystery which datasets got into data.gov and which are left to sit on the ftp site, unadvertised. Multiply this by several orders of magnitude and you have a system which contributes to the feeling that the government is no longer under our control. It may pay lip service to openness and informing the public but the reality is that the system is broken.