As the number of available HITs claimed at the top of each MTurk page has been over 1 million for a couple weeks now, some people keep thinking there's not nearly as many HITs as there ought to be for that number. It's not like when there was a half-million-plus up from that one requester anyone could work on (the one who had that Thanksgiving mass rejection incident, don't recall the name at the moment), on top of the usual hundreds of thousands of receipt HITs and IIRC some constant huge piles of CrowdSource back then, that got us to a total in the high 900,000s back in fall 2014, as previously seen in ChicagoK's avatar.
Other people kept insisting they thought there was plenty of HITs currently listed to justify the total, on the basis of seeing the 200k+ of Crowdsurf and 200k+ of ScoutIt. But I thought they weren't properly considering the massive drop-off after those largest few groups. So I made a script tonight to allow me to quickly and accurately add up every single visible HIT.
With 2491 HIT groups visible in 'All HITs', sorted by 'HITs Available (most first)', the totals with 100 HIT groups per page as of a few minutes ago (late night on Sat Jun 4, 2016) are:
715,683 + 3711 + 525 + 400 + 347 + 300*2 + 216 + 200*3 + 137 + 100*12 + 91
= 723,510 total HITs
At the same time that the header claimed "1,095,904 HITs available now".
That's a difference of 372,394. This is similar in scope to the several hundred thousand I suspected based on rough estimates from the few largest HIT groups when it first went over a million a couple weeks ago.
This definitively proves that there are now hundreds of thousands of HITs being hidden from us, using the feature Amazon added for requesters a month or two ago that allows them to choose to make HIT groups completely invisible to workers who aren't qualified for them.
My theory is that Amazon added that feature specifically by request to convince some new client to come onboard with a huge amount of work for a private worker base.
This now-constantly-high volume of hidden HITs being counted in it makes the 'HITs available now' number completely meaningless to the general workers, and seeing it over a million - which was a point of interest because it had never before been seen that high when it first happened a couple weeks ago - is hollow now; the last real max was what ChicagoK's old avatar had. The number was never really important because a lot of it is always stuff most people can't or won't do, but at least we knew what went into it before.
Other people kept insisting they thought there was plenty of HITs currently listed to justify the total, on the basis of seeing the 200k+ of Crowdsurf and 200k+ of ScoutIt. But I thought they weren't properly considering the massive drop-off after those largest few groups. So I made a script tonight to allow me to quickly and accurately add up every single visible HIT.
With 2491 HIT groups visible in 'All HITs', sorted by 'HITs Available (most first)', the totals with 100 HIT groups per page as of a few minutes ago (late night on Sat Jun 4, 2016) are:
715,683 + 3711 + 525 + 400 + 347 + 300*2 + 216 + 200*3 + 137 + 100*12 + 91
= 723,510 total HITs
At the same time that the header claimed "1,095,904 HITs available now".
That's a difference of 372,394. This is similar in scope to the several hundred thousand I suspected based on rough estimates from the few largest HIT groups when it first went over a million a couple weeks ago.
This definitively proves that there are now hundreds of thousands of HITs being hidden from us, using the feature Amazon added for requesters a month or two ago that allows them to choose to make HIT groups completely invisible to workers who aren't qualified for them.
My theory is that Amazon added that feature specifically by request to convince some new client to come onboard with a huge amount of work for a private worker base.
This now-constantly-high volume of hidden HITs being counted in it makes the 'HITs available now' number completely meaningless to the general workers, and seeing it over a million - which was a point of interest because it had never before been seen that high when it first happened a couple weeks ago - is hollow now; the last real max was what ChicagoK's old avatar had. The number was never really important because a lot of it is always stuff most people can't or won't do, but at least we knew what went into it before.
Last edited: