EVE Information Portal

It's quite interesting, but some of the "correct" results don't make sense from either a description or sample image stand point.

For example
Baffled
edit: More Baffled

Could someone explain how that is the right answer? asymmetrical bloby/tubular compared to the "correct" symmetrical vainy/web.


Is that "correct" in the sense that an above average number of people on sisi clicked on it, so we're being penalised because people are just clicking through carelessly? a bug? naff data in the project? etc.


edit: couple of samples later, I get something almost the same as before "ok, it was this last time..." wrong.... erm

Despite not receiving a single point of the promised 2MM, I liked the project. Now I keep checkin' back to see if the test server is up, that I might see it again. I'll see if I can't find similar somewhere else online until it's here. Thanks to everyone involved.

o/

I tried to find the Discovery project app on Sisi but could not find it. Someone in the Help chat told me it was not unabled yet. Do you know when it's supposed to be up ?

Do rules for one classification group affect the others? For example:

If a sample shows signs of cytoplasmic vesicles which, by description can never appear within the nucleus that then means that it can never be any of the nucleus groups, right? I would assume that if that's true then nucleus selections should be ghosted out if you select it.

The fact that I may be unintentionally but consistently wrong made me think of something else: If there are certain players like myself with perhaps a misunderstanding of a fundamental rule like that, then surely we shouldn't be seeing the other player's best guesses. I'd much rather see the selections of the most accurate players (and maybe as a counterpoint, the least accurate players).

I feel like my accuracy should be rigorously tested with almost as many preknown test questions popping up as the real ones so as to constantly reweigh the likelihood or weight of my answers. I can foresee a lot of attempts at cookie clickering this thing and that's generally the way mechanical turkish type jobs weed out that sort of activity.

Coming back again to the other player guesses issue, I find myself fighting herd mentality (even though I might be wrong) but seeing other players' guesses definitely affects your own choices.

I guess what I'm trying to say is that if this goes live and our accuracy is checked as often as it is now and accuracy doesn't heavily influence the other player consensus as a factor then cookie clickers could start a horrible follow the leader trend and create absolute garbage data where we all just try to match whatever we think the other players would analyse the image as. It might not even be malicious.

I thought this was some missions users participated in and that was completed for the Injector project or something like that.


The Scope – Upwell announces winners of first Citadel blueprints
Though it is neither, of course.

Just thought this clarifies.


Edit:
o7: The EVE Online Show - Episode 14 at 3 mins 15 seconds to 13:42.
Project Discovery with 1 or 2 Researcher(s), one from Sweden.

I'm glad to see this feature released finally. I've been pretty unhappy with a lot of the changes your devs have been making to the game lately. This is one of the few good ideas you've had in recent months.

Hopefully some real benefit to science comes out of this. Thanks CCP.

I'd like to be able to make the interface bigger, I'm old and blind, thx.

The Folding Home option was on the PS3 when it first was released for sales. Too bad it's long since been removed. If I remember correctly, it allowed use of the console's CPU and other hardware for extra computing space during low use time periods. I just left the console powered up when I went to sleep or out and about so it was available. Can't however remember which project it was assisting with at that time, I've slept since then.

Once TQv3 server quits choking on ProjectDiscovery's usage I'll give it a try.

>Jeven

I got similar issue:

Result

But if we look at the description for "wrong" answer: Wrong answer

And compare to "correct" answer: Correct answer

Not sure if it's my eyes or what but the image to classify does not look evenly distributed in any way or form nor is there enough of green to really fit the nuclear speckles (all examples show far more staining than my image).

PS: I think I'll ask a few colleagues to check what they think as they are used to doing this stuff.

I have to say, Discovery project is another example, when community plays in inexpected way thrashing any rules purposed by CCP.

Current popular logic to solve samples is following:
1. green on blue? Select first choise in first row (nucleus)
2. green on red? Select first choise in second row (cytoplasm).
3. Submit.

And it works, because majority of people do this and they got concensuss and raising accuracy. Those who tries to think or (majority: OMG!) use color filters to find some holes (majority: what is it?) will get accuracy penalty because they don't follow rules of majority.

This is my first sample today: 100081122, and there is no "big ball that overlaps with blue marker and evenly stained" and there is no "green that covers cell, except nucleus", there are clear and visible clusters inside holes, but who cares? Nucleus and cytoplasm wins.

Here is screenrecord where I toggle filters on the sample: https://www.youtube.com/watch?v=qpHOz-86AQ8

Yesterday I stopped when hit 2 clear error answers (again nucleus and cytoplasm). So I have to either start following those rules too, or ... I don't know. Why should I try to do good job if popular error answers either way will win?

If it's not goal of the project to sort all samples in 4 categories (nothing, green on blue, green on red, green on both), then may be let's fix it now, before all of us start following new rules of majority to get high accuracy and isks?