Closed mschwamb closed 9 years ago
Do we have a feel for what % of classifications are duplicates over say the past month? There were some front-end issues that we cleaned up awhile back that shouldn't affect anything anymore. Ouroboros still might very rarely serve a dupe, but that should be in the < 1% range.
I can confirm this person classified the image twice separated by 6 months. I'll send you the entries in the database to you directly in email. (I believe this is different from the other duplicate behavior I've seen previously). Duplicate classification was on June 14th.
If changes have been made to fix the other duplicate issue since end of April then, perhaps that issue is solved if not, the issue is still happening. I want point out the development team can also get those numbers too since you have access to the live database and daily copies, but I can get you this estimate you ask for over the weekend.
I agree with you that if the site has heavy traffic and queuing of the classifications is happening, a duplicate classification will happen rarely, but from my understanding of Ouroboros, the backend shouldn't send a duplicate to someone who classified the same image 6 months before Redis should already have that info and know it's already been classified by this person.
Statistics sent to @chrissnyder's zooniverse email.
Adding a formal ticket for the duplicate classifications bug. Appears Ouroboros is stilling sending cutouts to people who have already classified a cutout.
I spotted this today - a volunteer commenting on an imaged he classified in December before the switch to targeted regions and then got again today http://talk.planetfour.org/#/subjects/APF0001hgc