Twitter Signal to Noise ratio about 10%?

From Brand Republic, reporting on a study by Peer Analytics

Tweets were studied for two weeks and categorized into six areas: news, spam, self-promotion, pointless babble, conversational and pass-along value.

More than 40% of tweets were found to be pointless babble, followed closely by conversational tweets at 37%.

Pass-along value, or re-tweets, accounted for 8.7% of the total, a distant third.

Self-promotion tweets, typical corporate tweets about products, services, or "Twitter only" promos were fourth on the list at 5.85%. Spam and news, rounded out the bottom at about 3% each.

So, news at 3%, and lets assume "the rest excluding spam and babble" - about 57% - has about 1/6th relevance (a figure derived by a quick count on Tweetdeck) gives about a 10% signal, the rest is pretty much noise. A lot if it is diverting noise of course, and herein lies the problem - poor filtering!

But given that many people seem to still lap it up, and I'd bet Facebook is far worse (seen this - buying virtual goods with slebs on them on Facebook?), is Social Media the new Opiate of the Masses?

Update - some feel that more of the stuff that I didn't count as relevant to me is useful in a sort of river of "small-cap news" way - I had another look and thought OK, I can see another c 10% that may be relevant to me (in the light of general background stuff not aimed at me) , for a S/N ratio of 20% - thoughts?

Update II - excellent set of discussions over Twitter and at Tuttle on Friday on this, 3 big takeaways:

- Some of the "noise" is social grooming and "stamp giving" (to quote the social transactional analysts) - but where to draw the line on "relevance"?
- We need "hard" metrics like these in order to build useful filtering systems, especially those that are in some way predictive and automateable
- There is an ideological battle at points like this between "hard" and "soft" social network scientists - I come down on the "hard" science side given my background, it is incomprehensible to me that people would not want to quantify these phenomena, bit it is clearly incomprehensible to others that one would want to

So, here is an argument that a lot of the claimed "its not noise" is actually voyeuristic stuff that we don't want to turn off now we can see it, and here is an argument that it is relevant.

Update III - story seems to have hit Twitter today (Monday) after some harrumphing blogs - and resultant twts - about how the "MSM" don't "get" Twitter. I see it more as evidence of resistance to the concept of "hard" metrics from mainly "soft" SocNet types.

UpdateIV - quite a nice corollary argument using a Shannon's Law type argument from Devin Coldeway over on TechCrunch:

What can be said in 140 characters is either trivial or abridged; in the first case it would be better not to say it at all, and in the second case it would be better to give it the space it deserves.

(Shannon's Law is another cornerstone of Information Theory, and essentially looks at how bandwidth limits the amount of information one can transmit vs noise)