zooniverse / wildcam-gorongosa

http://www.wildcamgorongosa.org
Apache License 2.0
2 stars 6 forks source link

Multiple users experiencing repeats in WildCam #192

Open DarrenMcRoy opened 8 years ago

DarrenMcRoy commented 8 years ago

Several users have complained that they are seeing the same images repeated on WildCam Gorongosa. It's been mentioned since September; here's a thread about it, with several people confirming that it's still an issue.

Initially created in https://github.com/zooniverse/Panoptes-Front-End/issues/1878 -- to bring over @camallen comment:

"Basically if users have done all the 224010 then the api will show it to them again. it will mark the response with already_seen / retired but not sure the UI is displaying these marks and letting the users move on. Can we get a science team member to reduce the data to check for dups?"

I don't think that all of these users have done 224,010 classifications. Besides being utterly unfeasible (and probably provable if we look at their stats), that wouldn't explain why they're only getting duplicates some of the time (up to 1/3, they say).

They also say that they're being served the exact same image, not a data duplicate, and they can tell because, quote: "first, I recognized the pic and, second, I went to "Talk" and I already had comments there." Which wouldn't happen with a data duplicate.

camallen commented 8 years ago

They also say that they're being served the exact same image, not a data duplicate, and they can tell because, quote: "first, I recognized the pic and, second, I went to "Talk" and I already had comments there." Which wouldn't happen with a data duplicate.

The api could serve them the same image. I'm not talking about duplicate subjects for the same image, instead getting the same subject id from the api would result in their talk comments being for the same subject.

@DarrenMcRoy can you get a list of usernames experiencing dups. Also can we chase up with this "Can we get a science team member to reduce the data to check for dups?" science team?

DarrenMcRoy commented 8 years ago

Users who have reported experiencing duplicates are:

@stonepenny @Taubenus @bmst @banjozmom @Olomana

And going a bit further back:

@LeeReiswig @msyfoopoo99 @LucyLewis

As far as the team goes, I'll DM Bridget and link her to this issue.

DarrenMcRoy commented 8 years ago

@camallen Users have started offering several specific subjects that they have seen repeated, including:

830348 - 892913 - 869415 - 840073 - 691167 - 838464 - 874949

conneelyb commented 8 years ago

Hi @DarrenMcRoy - sorry for the delay and not sure if you still need my help with this one... can you clarify exactly what you need from me? I'm happy to help if I can.

aliburchard commented 8 years ago

@conneelyb sounds like Darren was reporting that your volunteers were experience duplicate subjects, which is bad because it's not a good user experience and wastes volunteer effort. Can you check in on the talk boards to see if this is still an issue? If so, a good place to start would be to investigate the subject IDs that Darren mentioned above. You might have duplicates in your dataset, or it might be an error on our part.

conneelyb commented 8 years ago

Thanks guys, I agree that this is an issue that we should try to resolve soon. I have looked through the csv export of images that I sent your team and searched for the photos that correspond to each of the subject IDs you listed. I don't see any duplicates for those specific photos on our end. I also checked the talk forum and there have been reports of duplicates very recently: https://www.zooniverse.org/projects/zooniverse/wildcam-gorongosa/talk/79/7125?comment=48697&page=3

This leads me to think that it may be a glitch in how the images are being served up to users. If you haven't already, it may be a good idea to check on whether the system allows images with the same subject ID to be served to the same user multiple times.

Let me know if there is anything else that you need help with.

Cheers, Bridget