OK, I’ve been talking a lot about rewarding people in EVE Metrics 2 for uploads. Specifically our new uploader credit system.
We’ve decided to remove all the EVE Metrics 1 statistics. Partly because there’s a fair bit of stale data in there, and we want to start afresh with our new crediting mechanism, which I will now describe in some detail.
However, we’re not detailing exactly what the equation is. We don’t want people to cheat the system for higher credits. The basic premise is this: you should be rewarded more if your upload is more useful. You’ll always get credited for an upload. But you’ll get more credit for an upload with more information (more orders changed) and for item/region combinations that haven’t been updated in a while. Here’s a graph to illustrate the point.

Graph of time since last upload against changed data metric against credit for upload. Dots indicate where one of the uploads in the EM2 test DB would fall.
We want to encourage people to not only update items that change often regularly (by rewarding high data changed metric uploads) but also to let us know what’s happening with items we’ve not seen in a while. Each upload’s credit scales from 1 to 100. When we wrote EM2, we added tracking of individual uploads as distinct objects, so we can do some even nicer things as far as tracking unmonitored items goes. We already use this data to generate market upload suggestions, and we’ll be refining that algorithm in the future.
This of course means better feedback for users on where they’re uploading to a lot and where other people upload most. The top uploaders page will be making a return, with a display of Top Region for each user as well as credits, uploads, and average credit per upload (the average usefulness of an user’s uploads). While we’re on the topic, though, have a screenshot from 2.1. This is the new ‘My Metrics’ page.
The eagle-eyed amongst you may spot the ‘account upgrades’ mention. This isn’t planned for 2.1, or even 2.2. You might also spot that we now track server status information and API status on the site. We actually capture player counts and so on, adding more data to our repertoire. The main function here though is to have API pages degrade gracefully when the API’s down, and we threw in the full monty server monitoring package as I had the code knocking around from an old project.
I think that’s enough for one blog. Oh, and for the Ruby enthusiasts in the audience: We’re on 1.9.1 now! The production site will be migrating to 1.9.1 with the release of 2.1, but our staging servers are already running it successfully with only minor fixes to a few libraries. Reve is amongst these, and Lisa Seeyle has happily pulled in our fork and merged it into Reve 0.0.120. Get it while it’s hot on Github (Or gem install lisa-reve if you already have github gems enabled)!
To install my reve from github you first should add github’s gem repository to your list using:
gem sources -a http://gems.github.com
Sounds very good!
Any plan for release date?
We’d love to release on Thursday night but it all depends on how well the API stuff cooperates and what state the tests are at then. Right now we have nearly 100% of our API code done, and the test suite is passing. It will probably involve an evening of downtime, however, and if we don’t deploy by Thursday night it won’t be till next week as I’ll be away from an internet connection from Friday to Tuesday.