Feed Sign in with OpenID OpenID

Simon Willison’s Weblog

Help pick the best photos, but watch out, it's addictive! My favourite WildlifeNearYou feature yet—our new tool asks you to pick the best from two photos, then uses the results to rank all of the photos for each species. It’s surprisingly addictive—we had over 5,000 votes in the first two hours, peaking at 4 or 5 votes a second. The feature seems to be staying nice and speedy thanks to Redis under the hood. Photos in the top three for any given species display a medal on their photo page.

Posted 25th January 2010 at 12:36 am

Tagged , , , ,

11 comments

  1. That is fun! You need to work on the photo resizing / cropping, though. An example I saw was this Harris's Hawk, which was headless and sitting on top of half an unreadable sign in the voting. I picked the other one, naturally, but clicking through to the full photo I see that it was just a poor cropping.

    Anonymous - 26th January 2010 01:38 - #

  2. Yeah, we're reconsidering the decision to crop now. Originally the feature was mainly intended to help us pick the best square thumbnail to represent each species, hence the cropping. It turns out our users don't care that we're picking a thumbnail, they want to pick the best photo overall (and hence judge them on their original composition). We'll probably remove the cropping pretty soon.

    Simon Willison - 26th January 2010 08:22 - #

  3. Simon, why do you prefer Redis over others key-value DB? It is because performance, features, or something else? Do you write some wrapper for Redis and Django cache system?

    Michal Valoušek - 26th January 2010 14:52 - #

  4. Features and performance. No other key/value store that I know of supports all of the following:

    • List and set operations
    • Set intersections on the server
    • Retrieve a random key from a set
    • Atomic increment / decrement of counters

    Redis does all of the above, and more, at speeds of 80,000 read and write operations a second. I don't think it has any competitors for the kind of things I like doing with it.

    I actually still use memcached for regular caching, but only out of habit - someone has already written a Redis backend for the Django cache framework.

    The only downside is that it (currently) needs to keep its entire data set in memory. If I needed a fast key/value store that stored more data than I can fit in RAM, I'd probably use Tokyo Cabinet.

    Simon Willison - 26th January 2010 15:20 - #

  5. Thank you very much for your answer Simon.

    Michal Valoušek - 26th January 2010 16:19 - #

  6. Calling all cars, clainlg all cars, we're ready to make a deal.

    Azia - 8th October 2011 17:45 - #

  7. That's 2 clveer by half and 2x2 clever 4 me. Thanks!

    Stretch - 10th October 2011 03:40 - #

  8. I am happy to find so many useful information here in the post, we need develop more strategies in this regard, thanks for sharing. maxwigs.com

    LLwtsgoohe - 26th October 2011 03:25 - #

  9. I wanted to thank you for this great read!! I definitely enjoying every little bit of it I have you bookmarked to check out new stuff you post whitehaven beach

    flapwer - 27th October 2011 14:27 - #

  10. I will recommend my friends to read this. Bottes UGGI am quite sure they will UGG Pas Cher learn lots of new stuff here than anybody else!

    ugg pas cher - 28th October 2011 03:31 - #

  11. Thanks for making such a cool project. I've been checking the site for the Windows version, but I never left a comment about it. I know you are working hard and doing it for free so you shouldn't feel rushed or anything. Manual E-Book conversion

    Groundless - 1st November 2011 06:13 - #

Comments are closed.
A django site