BLS numbers are routinely revised in retrospect to be worse a quarter or two later and with a friday release so that no one notices. As in, more than one in two BLS unemployment reports is retroactively revised, and I don't remember a single "it's actually better than we previously reported" revision.
tptacek (IIRC quoting someone else) described it well: ZeroHedge is the conspiracy theory site without the conspiracies. Ignore the tone, take the analysis with a lot of salt, but do look at the data - they bring to the front a lot of data that no mainstream publication will show.
There are two things wrong with the common view of conspiracies:
1. That it is impossible or at least, extremely unlikely, for a group of people to meet in private and make plans to further their own agenda at the expense of others -- that's all a "conspiracy" is. It happens thousands of times per day at any scale of operations.
2. That you actually need conspirators for a conspiracy-equivalent effect to take hold. Do you remember the option backdating scandal? Turns out tens of companies did the same thing. Hundreds involved was aware, no one talked, until an outsider forced the SEC to respond. This doesn't need to be centrally orchestrated. Whether or not you consider this "a conspiracy" or not, doesn't matter. It was effectively a conspiracy against shareholders. Similarly, there's the still unfolding emission scandal which is (effectively) yet another large scale conspiracy, whether or not Mitsubishi and Volkswagen engineers met in a shady bar to co-ordinate their cheating or not.