Musing about tweets as recommendations

If you read my last post, Thinking further about syndication,  then you would be aware that I’m devoting a lot of cycles to alerts, status messages, mini news items and tweets. I thought it was time I expanded a little on one small segment, tweets as recommendations.

When is a tweet a recommendation? Not when someone says “I’ve just blogged this” or “I’ve just posted this” or even “I’ve just launched this”. Not when someone says “And here’s me on <pick some media>”. That’s advertising, no getting away from it. No pigment, oil, wax or emollient can hide the porcine nature of such tweets.

I’m not saying that’s wrong. There’s nothing wrong with any of us advertising something we’ve done, so long as we don’t try and pretend it is something else. Such tweets (and for that matter, such alerts and status messages in general) are not recommendations.

So what makes a tweet a recommendation?

I think it’s the following:

• The tweet must be unbiased, it must be about an entity that is in no way connected financially with the tweeter
• The tweet must be clear in signalling something sufficiently positive about the entity
• The tweet must be actionable, it must link to somewhere that the tweetee can do something about the recommendation
• Most importantly, a tweet becomes a recommendation when the tweetee decides it is one

I think that’s a key point. When you tweet about something, you don’t decide that it’s a recommendation, it is the person who follows you who makes that decision. It is the “follower” who imbues the “tweeter” with the “recommender” tag about a particular class of social object.

Hugh Macleod has been hammering on around this space with his insights, initially on personal microbrands and then on social objects. Doc Searls has been working in a similar space with the work he’s been leading on VRM. Chris Locke was all over this in the lead-up to Cluetrain, in Cluetrain, and ever since.

The authenticity of recommendation is not conferred by the recommender, but by the recipient.

There’s a lot more we can do with tweets as recommendations. Staying with the Cluetrain theme, straying over to the Weinberger side of the argument, it is the very miscellaneity of the tweets that makes such things valuable. But when everything is miscellaneous, it’s time for the taggers to ride in.

So will we see a time when a tweet gets tagged by someone other than the tweeter? Is that how a tweet becomes a recommendation, when a second party adds something like #recommend #books to someone’s tweet? Maybe we’re almost there already.

We can also make the tweets carry more context. As I’ve said before, I just love the last.fm to foxytunes to twittertunes to twitter sequence; you’re listening to something, the title and artist/s get scrobbled off and packed to go somewhere, expanded into a full mashed-up result covering the song lyrics, artist web site, google returns, last.fm entry, iTunes link, wikipedia entry, the whole shooting match. Then it gets compressed again, snurled or tinyurled into submission, repackaged for suitability in a 140 character text medium.

We can have ratings associated with such tweets; we can use collaborative filtering techniques as well.

There’s a lot that can be done. I’ve used the music analogy, but this covers many classes of “social object”. Books, films, restaurants, airlines, hotels, anything.

And sure, people will try and game this. But you know what, it’s not that easy. What if we take such tweets and move them into a tweetypedia? A wiki place where customer-driven recommendations go? Tagged with value like Flickr, editable with care like Wikipedia. Then, if we take the Clay Shirky maxim of keeping repair costs as low as, if not lower than, damage costs, the tweetypedia can become a valuable common good.

Just musing.