Why is Said.fm a good filter?
The internet with its profusely capacious choice reflects both the beauty and the frustration with the beast. How do you make your choices about what you will consume over the web? What to read, what to watch, what to listen to…
I read an interesting article on this recently by Doctorow who provocated that when you have such an abundance of choice how do you choose what goes in your playlist, which experts to follow on Twitter and what books to read. When the traditional boundaries of geography and income disappear in regards to what you can choose then it can become an overwhelming opportunity.
Here at Said.fm we’re building our start-up whilst being mindful of this internet problem of choice. There are audio programmes out there that are of poor quality (in terms of sound, editing or content) however, there are many programmes which are just downright awesome. So, we have chosen to put blood, sweat and tears into putting these downright awesome programmes in the limelight for you to listen to and discover. It’s not a search engine for podcasts, it’s not a directory, it’s not an aggregator but it’s a place where you can come and press the big green play button with re-assurance that a brilliant programme will be heard.
A trusted channel
But to give you an insight into why said.fm is a good filter or a trusted channel I would like to share how and why we cultivated an approach that is gaining momentum for us.
Machines and algorithms are good for processing data and streamling processes, however it’s people that make the best recommendations. If we’re honest (and why wouldn’t we be?) we only came to this realisation many months into starting Said.fm. Initially we were completely engrossed in the technical possibilities of the Said.fm experiment and worked ourselves into a dizzying frenzy of machine learning algorithms to make topic relevant recommendations.
Once we had the aha! moment we changed our approach and decided to widen the team by enlisting curators. Our curators help filter out the noise and make good audio recommendations (both channels and programmes). They are well placed in doing this as they are keen listeners, story tellers and great programme makers.
So, now we do have some machine learning algorithms at play but more appropriately to make semantic connections between curator’s recommendations.
Ultimately, we hope our recommendations, opinions and carefully nurtured technique will make hitting the play button and listening an easy choice.