>Wikileaks has two audiences: the people that agree with its editorial voice and the people that are interested in the raw data.

This borderlines on being paternalistic in overlooking another audience -- arguably the most important one for an information clearing house -- those whose opinions can be swayed by new information if they trust the source of that information.

Engaging in editorial commentary that overshadows the content of the actual data undermines that trust in a very real way. When you're pushing an editorial agenda, people are going to dismiss the data because it's coming from a source they consider ultimately untrustworthy.

>Furthermore, by providing the raw data, scientific journalist give other organizations a chance to provide alternative context and/or analysis.

This is exactly why the clearing house and the editorial voice need to be separate entities. If the data is available, then anyone is free to (and will) perform analysis on it without tainting the data itself with the reputation of agenda of the organization releasing it.

The trustworthiness of data is a function of its provenance, not its accompanying analysis or commentary.

Conversely, the trustworthiness of information (i.e. analysis and commentary) is a function of the degree of bias of the source and not the data that informs it.

If the raw data is universally accessible then there should be no shortage of trustworthy information sources for the third audience you mention. Unbiased sources are ideal but that’s too much to hope for, so the next best thing for that audience is to consult a multitude of sources that are known to be examining the same data.

Finally, I do not think Wikileaks is ideal (I prefer something more like what Nate Silver does). But I take it as axiomatic that a free press — even one with an editorial voice — is a Good Thing. A journalistic organization that also releases its sources is a Better Thing and hence must also be a Good Thing.

>The trustworthiness of data is a function of its provenance, not its accompanying analysis or commentary.

From an idealistic standpoint, sure. From a real world standpoint, trustworthiness colors peoples' perception of the data itself. Including commentary with the release has lead to Wikileaks being viewed as untrustworthy in many people's eyes, resulting in them dismissing the analysis of data sourced from them outright.

Again, in an ideal world, the data would rule all, but that's simply not the way it actually plays out. It's not hard to find examples of exactly this playing out with Wikileaks, and it lessens the overall value of the data they do release in swaying public perception.

>(I prefer something more like what Nate Silver does).

Nate Silver does meta-analysis. He doesn't perform any data gathering himself. He represents the third-party editorial voice I'm advocating. In fact, Silver's model even further illustrates what I'm getting at because it accounts for and weights the reliability of the sources of the raw data.