A Twitter App I want

Whose tweets do the people whose tweets I value, value?

That is, this application would help me find the people whose tweets I consistently value, and then find out whose tweets they consistently value, and it would do it in a reasonably anonymous way. (One can immediately imagine extending this to news groups and then books, music….)

[This RFC is the result of brief musings with John Papa who is not responsible for my corruption of it, but who is a great person to bounce ideas off of]

The Client

Here’s how it works… a client (Windows, Mac, Phone, iPhone, etc.) is created that lets me rank tweets on a scale of –3 to +3, with an unranked tweet scoring 0.

 

 

These ratings are sent to a web service where they are tied to my username but it is not possible for anyone but me to find out my rankings. (I might use my own rankings for culling my lists, discovering that there are people I follow whose tweets I don’t value very much).

Finding People To Follow

The next step is that I can ask the system for the top N valued by my top X.  The restriction is that both N and X must be a multiple of 5. That is, I can ask the system for the top 10 valued by my top 20.  The system will then examine the rankings of my top 20 and combine them to create a list of the 10 names most highly valued by those 20.

It will then return something like this:

Your top twenty ranked the following people in their top ten. They are not in order,  but they are shown as the top 5 and the second 5.

 

Friendly Name @Name # Tweets last 3 months Halves
George @GeorgeW 238 1
JohnA @JohnAdams 12 1
Tom @TJefferson 40 1
JQ @JohnQ 10 1
AJ @AndrewJ 2 1
JMonroe @JamesM 38 2
MVB @MartinVanB 32 2
JT @JohnTaylor 45 2
Millie @MillardF 101 2
Bucky @JamesB 23 2

 

This indicates that overall your top twenty found all of these ten to be in their most highly ranked group, but the group in the first half were ranked more highly than the group in the second half. If you ask for 10 that may not matter much, but if you ask for 50, having 10 groups of 5 can be a good compromise between the false precision of an ordered list and a totally random grouping.

If no one builds this by the time I finish ViKi perhaps this will make a good 2nd project.

Share

About Jesse Liberty

Jesse Liberty is an independent consultant and programmer with three decades of experience writing and delivering software projects. He is the author of 2 dozen books and multiple Pluralsight courses, and has been a Senior Technical Evangelist for Microsoft, a Distinguished Software Engineer for AT&T, a VP for Information Services for Citibank and a Software Architect for PBS. He is a Xamarin Certified Mobile Developer and a Xamarin MVP, Microsoft MVP and Telerik MVP.
This entry was posted in z Silverlight Archives and tagged , . Bookmark the permalink.

Leave a Reply

Your email address will not be published.