5 ms·

Even with English language skills, maybe just scanning the text and not fully comprehending? Slightly off-topic: I wonder how impartial a human has to be - wou

Even with English language skills, maybe just scanning the text and not fully comprehending? Slightly off-topic: I wonder how impartial a human has to be - would the outcome be different depending if the reviewer was a devoted catholic or an atheist?

bryanrasmussen 6y ago
when I say English language skills I don't necessarily mean they don't read English, but just that they don't read it well enough to have understood the distinction under whatever constraints they were working - if those constraints include scanning quickly then they did not have the skills to read that text quickly and understand it.
- mannykannot 6y ago
  True enough, but there really are at least two alternatives here that could be distinguished, except that we do not have sufficient information to make this distinction. In either of these two cases, the reviewer presumably did not understand the meaning of the statement, but in the first case this was because they did not know the language sufficiently well, while in the second, they lacked the critical reasoning skills to correctly discern its meaning. In the first case, but not the second, the reviewer's performance on this particular task would vary depending on the language it is expressed in. Of course, in neither case should the person be given the job of reviewing English-language tweets. There are additional possibilties, such as Twitter having created an environment in which it is much safer or easier to agree with the initial ruling, or where the reviewer is required to pick from a list of justifications for reversing the initial decisions, and none of the options fit (I'm sure you have seen questionnaires like that.) The one thing we can be sure of is that the Bishop should not have been banned.
  - richrichardsson 6y ago
    There is a possible 3rd option: they didn't have enough time to make a reasoned judgement. It's quite conceivable these human moderators are expected to get through <X> tasks per <time period>, and perhaps the values you plug in there are such that it only leaves a very short period of time for each one.

bryanrasmussen 6y ago

when I say English language skills I don't necessarily mean they don't read English, but just that they don't read it well enough to have understood the distinction under whatever constraints they were working - if those constraints include scanning quickly then they did not have the skills to read that text quickly and understand it.

mannykannot 6y ago
True enough, but there really are at least two alternatives here that could be distinguished, except that we do not have sufficient information to make this distinction. In either of these two cases, the reviewer presumably did not understand the meaning of the statement, but in the first case this was because they did not know the language sufficiently well, while in the second, they lacked the critical reasoning skills to correctly discern its meaning. In the first case, but not the second, the reviewer's performance on this particular task would vary depending on the language it is expressed in. Of course, in neither case should the person be given the job of reviewing English-language tweets. There are additional possibilties, such as Twitter having created an environment in which it is much safer or easier to agree with the initial ruling, or where the reviewer is required to pick from a list of justifications for reversing the initial decisions, and none of the options fit (I'm sure you have seen questionnaires like that.) The one thing we can be sure of is that the Bishop should not have been banned.
- richrichardsson 6y ago
  There is a possible 3rd option: they didn't have enough time to make a reasoned judgement. It's quite conceivable these human moderators are expected to get through <X> tasks per <time period>, and perhaps the values you plug in there are such that it only leaves a very short period of time for each one.