Post in Gathering input from one million people

Gathering input from one million people From: Miles Fidelman Date: Feb 07 22:29 UTC Short link

Steven Clift wrote:
> I started the building consensus online discussion with a couple thousand
people online.
>
> I am trying to imagine what an online input feature on the next
WhiteHouse.Gov or you name the world leader might look like.
>
> Assuming the topic is set and people are reacting to a set of questions with
minimal framing text:
>
> 1. How would you gather input from one million people online?
>
> 2. How would you report the results?
>
>
I keep looking at various rating and reputation systems: review on
Amazon, digg, slashdot, helium.com, etc.

MoveOn has a sort of interesting model for soliciting input on what
issues to pursue next, and what positions to take.

Google pagerank is a way of ranking web pages by how many people link to
them.  Then there are the various pingback ratings for blogs.

Wikipedia is not a bad model either

The basic assumptions are:
- nobody can read everything
- it's important that everything is read by a statistically meaningful
subset of everyone
- that as things move toward the top, more people see them.

> 3. Would you support two-way interactivity among participants? If yes, how?
>
Depends on how you do it.  Attaching comments to a posting (a la blog
comments, or Amazon ratings) is pretty straightforward.  Real-time
conferencing across lots of people is not as straightforward.
> 4. How would you ensure that the highest quality contributions are noticed?
>
Filtering, and bubbling.
> 5. How would you apply demographics to your analysis and deal with
representativeness?
>
Keep track of profile information, which might have to be different for
different issues (e.g., parent/non-parent when discussing school issues).

Allow filtering of responses based on profile information.  Perhaps make
sure that postings are routed to a statistically relevant sample for
ratings.

Of course this begs a few questions:
- how to decide what profile information is collected for a particular
discussion item
- how is the information collected and validated?
- how to decide how it's applied
> 6. How would you analyze open ended responses?
>
>
Well... the above model doesn't have any other kind.  It's all
open-ended, and ranking of open-ended.

If the goal is to converge on a decision, the decision is the collection
of postings that bubble to the top.
    

Miles