from:    Hamish Allan <hamish@gmail.com>
to:      office@power2010.org.uk
date:    Thu, Feb 18, 2010 at 3:41 PM
subject: Re: Time for a strong finish

Hi Pam,

I find your request to go back and keep voting very odd. If all your users voted for every proposal, it would do you no good: you'd have an equal number of votes for each proposal, so the voting would not help you choose five of them. So it's just as important for users NOT to vote for some things, as it is for them to vote for others.

Do you believe you have many users who've voted for some things, and who intend to vote for others but didn't get round to doing it? That may be actually the case, purely because of the design of your website. Each time you vote for something, it reloads the main page, and each time it reloads the main page, the order of proposals is randomised. Because there's no visual indication of which proposals you've voted for already, it's really difficult to keep track of which things you've voted for and which you haven't. In fact, it's difficult to imagine how you could design a system that makes it *more* difficult to keep track of it (except perhaps if it didn't even inform you that you've already voted for something when you try to vote for it a second time).

Basically, the way you've designed your site means you can't really derive that much useful information from it. This email you've sent out -- requesting your users to go back and vote more -- is a tacit admission of this. You can't distinguish between a vote which has not been made because the user got fed up with your website, and a vote which has not been made because the user doesn't care about the issue.

Why not just have all the proposals in a list, with two buttons next to each -- "I care about this" and "I don't care about this" -- which visually indicate any votes already taken? Better yet, allow people to fully order the proposals according to their importance.

It's a great shame because I think your heart is in the right place, but your voting system is just plain unusable, which leaves your results statistically invalid.

Best wishes,
Hamish