About this blog

  • My day job - I am founder of Battenhall, communications agency for the social media economy.


Subscribe to this blog

  • Subscribe to this site by email or RSS

    This is my personal blog and does not reflect the views of my company or clients.

Blog powered by Typepad


« Launching the Reading arm of Social Media Cafe next week | Main | Social Media Cafe comes to Thames Valley »

March 08, 2009


Take a look at @jobsworth and his recent post on the value of trusted recommendations.
I myself am looking for a way of ranking Twitter searches.
Your achievement today was asking a definite question and receiving (near) real-time responses - which were both personalised and relevant.
What I'd like to see is a reliable way to rank most relevant/interesting Tweets for a particular keyword combo.
Flickr has a good "Interesting" algorithm as, I would guess, does Digg and similar but what about Twitter rankings for Tweets gone by?
I replaced my homepage with search.twitter.com last week and am loving it. I'd like to replace Safari/Firefox search bar with Twitter next.
Good luck.

Cheers Conor. And I massively respect JP Rangaswami's writings, so will read that one as I prep :)

Hi Drew,

Neat experiment. I'm guessing that the results will depend on the goodwill of the network, and the authority of the respondents.

Goodwill exists in spades on Twitter, particularly amongst the social media/PR crowd, so I don't think that will be an issue.

But how do you assign authority to recommendations from a network? How did you rank the responses you received for lunch? Where did you end up going, and on what basis did you make the decision?

All the best,

Steve - good question on authority. So far my experience has been that you get a small number of quality responses, so you can make a judgement call. When Google throws up thousands of responses, the system determines the ranking of the results to that relevance prevails. When it's a small number of responses, like a few or a hand full, you get to make the choice.

For lunch today I did precisely that - made a call myself rather than go for 'authority' etc. I knew some of the places suggested, some were closer than others and some I knew would be unsuitable. So I chose The Swan. :)

Hi Drew, great idea for an experiment - looking forward to seeing how it goes.

I think we're seeing a jostling for position among a number of different ways of searching. I think it was Jason Calacanis who put it like this:

Short head searches e.g. 'Barack Obama' will be provided by human search like Mahalo where the results are pre-prepared on popular topics.

The 'fat middle' of searches will be social e.g. asking your Twitter friends for an answer to a question.

Then the long tail of very specific niche searches will remain algorithmic.

So I think that your experiment will work best for this middle range of searches, but it'll be interesting to see where the boundaries will lie. If you were to continually use Twitter for long-tail searching you'd create too much noise and you'd be taking more than giving.

The comments to this entry are closed.