I am pretotyping a new web site, Who disagrees with me?, in order to find out if the basic idea behind the service is viable or if there is no interest at all.
Who disagrees with me subscription
A filter bubble is a controversial concept developed by Internet activist Eli Pariser in his book by the same name to describe a phenomenon in which Internet search queries selectively guess what information a user would like to see based on the user's past search history and, as a result, searches tend to play back information which agrees with the user's past viewpoint. According to Pariser, users get less exposure to conflicting viewpoints and are isolated intellectually in their own informational bubble. -- http://en.wikipedia.org/wiki/Filter_bubbleI have refined my RSS reader, Twitter and Google+ accounts so much that I never encounter something that challenge my views. For example, I get PHP-related content all day; but I am learning much from reading 7 languages in 7 weeks and being exposed to Ruby or Prolog. I suggest to view the short TED talk (9') from Pariser for a longer explanation of the phenomenon and its issues.
While this filtering is definitely useful for singling out an appropriate slice of the Internet cake, it menaces growth as I always threading on the same subjects. Most of the new content I see comes from personal blog and accounts, which are by definition free to link to links they find interesting.
When building a web service an engineer typically starts from designign a data or object model, or from the algorithm that will find challenging links by starting from a set of references. And it must be an adaptive algorithm, that improves as the user gives out more information on his preferred technology, economic policy, or music.
But that's not what to start from: coding the perfect algorithm won't suffice if there is no one interested in using it. Pretotyping consists in narrowing down the scope of a product (like in an MVP) and even faking the existence of the product in order to check the market viability. Will people be interested in reading a challenging newsfeed, or they will prefer conservative filters and I will waste my time buying a domain name and coding?
I want to validate my idea by spending the minimum amount of time and money.
In my case
Instead of a continuous feed of information like for Facebook or Twitter, I limited the pretotype of the service to a single, weekly e-mail containing links. I also limited the category to a single choice: preferred technology.
I started preparing a subscription form with Google Docs, but I switched to MailChimp, a newsletter service, due to the support for verification emails.
I'm using a mixture of two pretotyping strategies described in the book: the Mechanical Turk and the Re-label; they are viable thanks to the very limited scope of this trial. The Mechanical Turk means I will write the weekly emails by hand, not select links with an automated algorithm. Re-label means I'm using an existing service (MailChimp) to take care of the subcription part.
What I'm saving here? The time to build a complex algorithm and a website with a 2nd level domain and an authentication system to store user preferences.
My metrics will be the initial subscription level compared to the number of view and links this post get, and the returning users of the service (people who do not unsubscribe in a month, for example.) MailChimp is adequate for measuring both, while Google Analytics will count the unique views this post get.
What will happen?
The two extremes of the spectrum are:
- no one submits the form. The service is a crappy idea, because everyone wants to read interesting articles and not from other fields. In this case I have saved my time and I have discovered the idea was not really valid with an expense of just 4 Pomodoros (2 hours, with the time spent writing this post included) and no monetary investment. If it fails, it's normal: most of new ideas fail.
- I get a good amount of subscription, which do not quit from the newsletter for more than a month (let's say 4 issues). In this case I can move to coding and build a real prototype.
Who disagrees with me subscription