Wielding An Ushahidi-Shaped Hammer
(courtesy of George Chamales)
If someone sets up an Ushahidi Deployment on the Internet and no one’s around - does anyone care?
Start with the Questionnaire
- go over it on the phone with them
- that way they don’t need to be bothered to read it
- pitch it to the customer as the “rapid deployment” questionnaire
Hitting the Sweet Spot for a Deployment
- right number of reporters
- right number of reports
- right fidelity (information, location, timing) of reports
- right number of people using / responding the information
- right number of people publicising the deployment
Changing instances mid-way is hard and annoying
- all the categories got messed up
- had to add images, categories, news, verification types by hand
- is going to be even more fun to migrate from 1.1 to 2.0 over
- it’s even worse if you’re also switching domain names
Have your verifiers and team outside of the disaster area!
Faisal (pakreport.org)- “one TV can start advertising us. got a few connections. but we need to setup logistics before that. power breakdowns are big problem for us right now”
Measuring the success of your deployment by the number of reports you’ve collected is dangerous. Sure it’s tempting and by far the easiest way to do it, but it’s not the size of your reports, it’s how you use them.
Stuff you must do
- understand the importance / difficulty of what you’re about to do
- domain name
- email groups
- keep everyone in one big group to start with, specialize later
- x specializations: dev, media, management
- deployment blog
- team roster: name, location, telephone, email, responsibilities (spreadsheet)
- skype channel
- twitter handle
- other stuff: facebook account, google voice
- backups of code and database
- x if possible, replicate server
- staging server and production server and local dev server
- sysadmin commitment to staging changes before deploying -- nothing straight to production
- use ssh keys when possible
- use unique logins when possible
- use non-shared hosting to ensure availability of sudo access and cron, etc.
- have regular briefings
- x public conference call for status reports
- x blog posts - weekly?
- set up standard ppt template for briefings / presentations to other people
- what about building it on the fly?
- having not only context, but sources
- should have a description of each category
Who are you serving?
- responders: collecting, ordering, and disseminating high-value, categorized information?
- citizens: collecting and ordering requests for assistance?
Who are you communicating with?
- what’s the easiest way to track all the various passwords you’re generating?
- what are the most effective means to get the word out
- how do you make a deployment in a crisis last and grow into a long-term rallying point?
- check whether your api key is in use
- see your reports on usage
- Collect Aid Requests: need monitoring and coordination by aid agencies
- Build Situational Awareness: need people submitting info + vetted aid groups to ask questions to them
- Share Aid Status: need info from aid agencies, aid agencies / people looking