It’s a topic that obviously goes well beyond tone alone, certainly, but humour me for a moment. Automated or not (a whole other debate), what do we/ should we be toning and what shouldn’t we? What’s meaningful? Where do we cut it off? For example, arguably, one could, I suppose:
- apply an overall leaning or sentiment to a blogger (I’ve seen this done for political bloggers)
- tone for only the original post
- tone for each comment individually that responds to each post
- apply one aggregate tone that accounts for an overall sentinment of all of the comments (a colleague coined the phrase ‘conversation tone’…similarity to conversation index intended…which I think has some merit)
Again, before I get an onslaught of comments and e-mails, I recognize the discussion about measuring blogs is far more complex (traffic, relationship indices, polls linking to biz tangiles, social network analysis etc.) that strictly tone.
PRincipal of PRooph 
For me it often helps to tie proposed measurements with outcome: “What outcome do you want to predict/explain from your measurements?”
Tone of voice is relevant to measure because one may be able to make the argument from “tone of voice” can be transformed into positive/negative promoters which again can be correlated with relative growth.
Similar there is good research to document the relationships between share-of-voice and share-of-impact (influence adjusted share-of-voice) and outcome (sales), for example in the pharmaceutical sector.
I am not aware of any research that suggests that a “conversation index” can predict/explain/correlate with any kind of outcome and I’m having trouble making the logical case that it should.
So if a proposed measure doesn’t explain (or help to explain) part of an outcome, what would be the justification for it?
I don’t claim for one second that a conversation index or a conversation tone is in any way a measure of outcome. (Absolutely agree with you that outcomes are critical) It’s output only. While outputs are bottom of the measurement barrel, they are a small part of a much larger equation.