PD Control-Samples that need to be reviewed

Communications Center General Discussion
1

As the title said - before we start a whole bunch of “wtf is this” threads, id say we collect some samples we think we need to be reviewed. Please try not to report a whole bunch of samples that confuse you, but rather those you can see plainly wrong.

Post the sample ID (Right bottom corner after submitting) and, if you want, a screenshot. Also reason what is wrong. Maybe we can get a central “collecting” thread.

Ill start:

200077055
Same object is twice in the sample, also folds perfectly. Only one of them is marked.

1 Like
2

The whole point of PD is that we reach a consensus through individual effort. Flagging samples as “wrong” defeats the purpose of the exercise. If enough people mark a sample “correctly” then the system’s definition of correct for that sample will shift to the consensus. That’s why you get rewards regardless of if you’re correct or not.

3

There are 2 types of sample, ones which have a consensus and ones which require a consensus.

The ones which require a consensus do not affect your accuracy score.

The one which have a consensus already affect your accuracy score and there are many of these which are either blatantly wrong or dubiously wrong.

The latter is what this forum post is regarding, not the samples which are still requiring consensus.

4

We are talking about the control samples, where the designers know the answer but want to test and assess us, whether we are actually able to find transits. Those which modify the accuracy. This has nothing to do with consensus.

5

and some of them are wrong. Im not talking about very difficult ones, but sometimes, you got the “Youre joking, right?” feeling. And i think we should collect them so they get reviewed :slight_smile:

6

Yeah, you are totally missing the point of the exercise.

7

What is so hard to understand that there are wrong test-samples? Do it yourself a while, and youll see.

8

I have, and the established process will sort them out.
Posting things like you suggest can work against the purpose of the established group-effort system.

Sometimes issues perceived within a sample can be with the observer themselves.

9

I think you dont quite understand how the system works :slight_smile:

10

Another one: 200077053. Also seems to miss one transit.

11

The point of the exercise it to help train people to spot transits. The WTF curated samples do not in any way help us to understand why we are wrong, thus don’t help us to improve our results.