Twitter Data Plan

Yesterday we had a request for a copy of all of our twitter ids.  Here is my proposed policy for Twitter Id Data. 



We have the most amazing database of Twitter Ids.  We have the national party's id, the state party's ids, the local a party's ids, and the politicians ids. 

If the Twitter Phone Trees take off, soon we will have numerous ids of private individuals.  How should we treat this data?

I propose the following policy. 

State  parties,  currently have password protected access to all of their Twitter ids on the map server.  It is split into two lists. Politicians in their state, and local parties in their state. 

I propose that every state should maintain 3 lists on Twitter.  A list of candidates in that state.  A list of local parties in that state, and a joint list.  Initially they can do this manually, 
now that I know what needs to be done, soon the map server will be able to manage this for them. If Twitter allows it. 

The national party should maintain a list of state parties.  

What drives this proposed policy?  A set of principals. 

1.  The data should be open to everyone.  "Data wants to be free".  Both Alex and I share this policy.  Sadly that does not appear to be the policy of other Green groups I work with. 

2. Control how the data and software is used.   The official policy is that we do not add Democrats to the map.  We do not want them creating their own map either.   Yesterday I tweeted to all the Greens in New York.  It worked brilliantly.  We do want people tweeting to all of the Greens in a state.  It brings them together  So make it easy to do, by creating a twitter list for each state.   And a twitter list for each caucus.

3. Structure the data.  Kind of a software developers perspective on data structures. Make it easy to find the lists, by hanging those lists off of the state twitter account  A google search takes you to the state twitter account, click into the list. And the national party @GreenPartyUS should host the list of state parties on their twitter account.   

3.  Organize things as a tree.  There should be no more than about 7 items in a category.  That is a basic principal in human factors.  Most states have less than 7 candidates, and less than 7 local parties.  So that works great.  Large states like California have way more, so they should organize their information as Northern California, Central Valley, Silicon Valley, Southern California.  The software allows them to do that.  CA has done the best of any state in using that feature, but finer grain could be applied.  Or maybe the current approach is better.  I am not sure.  Rules are made to be broken. 

4.  Avoid infinite lists.  Google, Twiiter and Facebook all give you these infinite lists.  A human factors disaster. Results should be structured. 

5.  Build the internet we want.  Decide how you want things structured and build them that way. 

6.  Minimize noise.  Releasing a very long list of Twitter ids is just asking for it to be accidentally misused, generating noise.  Release state lists, and that is how they will be used. 

7.  Of course private people's twitter ids should not be shared.