I completely agree with this, when I was voting I saw lots of pleasant 
looking countryside scenes and rated most of them all quite highly, it 
wasn't immediately obvious to me how one scene was more 'scenic' than 
another. The idea of comparing two pictures seems to make more sense to me.

Francis Davey wrote:
> 2009/4/4 Simon Roe <[email protected]>:
>   
>> Personally I'm not interested in playing, and the idea I'm doing it
>> for some unknown project is actually more off-putting than anything
>> else for me.
>>
>>     
>
> Can I make a strong suggestion on this? When I had a go I was
> presented with a picture of somewhere really quite nice and a scale
> 1-10. Now that's really hard since I have nothing to compare it with
> and in fact all the pics I saw were of really quite nice places, which
> may skew the score a bit.
>
> I strongly suggest you use a kittenwar like model. Show two photos and
> ask which is the more scenic. It feels less like work (rating a number
> is hard) and everyone will agree on the semantics (if not on their
> taste). There are all sorts of psychological problems with value
> scales (eg people not liking to use certain of the values).
>
> That way you can rank all your pics much more finely and rather more
> reliably than with a pick score.
>
> Finding out how other people rated it is good. Finding that I have
> similar tastes to someone else is also good (you might like.....)
> using factor analysis or some such scheme. That rewards me by showing
> me places that people like me liked and that I might like to visit. Of
> course it can't do this until I've put some data in, so I am more
> inclined to play because I have to to get the reward.
>
> Hope that makes sense.
>
> --
> Francis Davey
>
> _______________________________________________
> Mailing list [email protected]
> Archive, settings, or unsubscribe:
> https://secure.mysociety.org/admin/lists/mailman/listinfo/developers-public
>   


_______________________________________________
Mailing list [email protected]
Archive, settings, or unsubscribe:
https://secure.mysociety.org/admin/lists/mailman/listinfo/developers-public

Reply via email to