Higgs Hunters Talk

Simulated data

  • johan_klein_entink by johan_klein_entink

    It seems that nine out of ten images I get to see, turn out to be 'Simulated data' after pressing the 'Ok' button. Does that mean that I am considered to be too stupid to look at real ones?

    Posted

  • davidbundy77 by davidbundy77

    they seem to be nearly all simulations today. Citzen science projects please note that volunteers quickly lose interest if you don't give them real data.

    Posted

  • Ptd by Ptd

    This is happening to me as well

    http://talk.higgshunters.org/#/boards/BHH0000001/discussions/DHH000018q

    Posted

  • patricle by patricle

    Same here! 90%+ is simulated data:

    "Thank you. Your classification has been recorded. This was simulated data that we show volunteers in order to calibrate the project."

    Posted

  • Mark_..._a_physicist_of_sorts by Mark_..._a_physicist_of_sorts

    Agreed, it is disappointing, but they need to improve the project so they can train the next generation of s/ware or wet-ware to do better than us. Hey ho!

    Posted

  • Ptd by Ptd

    I have found a response about this matter from the ST in another thread 'Simulated data?' http://talk.higgshunters.org/#/boards/BHH0000003

    Posted

  • JeanTate by JeanTate in response to Mark ... a physicist of sorts's comment.

    Which is fine ... as long as all sims are noted (immediately after classification is done), and the reasons why sims are included very carefully explained.

    If the (vast) majority of images are sims, this needs to be explained; zooites are not machines.

    If this project has a similar classifier/classifications distribution to most other Zooniverse ones, then the 'super-uses' are gold. Not only do they do thousands of classifications, but they are often the zooites who have the sharpest eyes (this has been investigated; the Supernova Project has a paper on this). De-motivate them by (potentially) making them do thousands of sims will drive almost all of them away, leaving this project with a huge number of classifications by a huge number of classifiers (who will average ~< 10). The opportunity cost will be huge.

    Posted

  • DZM by DZM admin

    I'll look into this tomorrow... Thanks for the heads-up, everyone!

    EDIT: There's a bug report in. The team is working on getting this fixed. Please bear with us... We really do want to be giving you as much real data as possible!

    Posted

  • DZM by DZM admin

    Okay, so it's looking like it has not actually all been simulated data... The "simulated data" message just got stuck on most/all of the objects.

    We're working on getting that fixed, but as far as I can tell, you have indeed been working with real data.

    Sorry for the error on our end!!

    [EDIT: This has since been revealed to be incorrect. More information in this thread. We should hopefully have another official update very soon.]

    Posted

  • DZM by DZM admin

    We've made some major fixes and posted a comprehensive update and explanation here.

    Posted