Can there be suggested edit audits that should be approved?

Question

There has been some discussion over the years about the quality of suggested edit audits and the possibility of there being audits where the correct action is approve:

• Are there tests of suggested edits which should be approved?

• Are the suggested edit audits too easy?

• "Acceptable" Honeypot Suggested Edits

I could not find a proper feature-request on this topic though so I am creating one.

The problem I see is a robo-reviewer could simply indiscriminately reject all edits including good ones and they could still pass all the audits. I do not have much of an insight as to how common that problem is but I support the idea of having 'known good' audits that catch reviewers who are quick to reject good edits. Rejecting good edits can discourage editors from taking the time to propose them.

All queues except suggested edits have audits for both known good (Looks OK/Leave open/Reopen) and known bad posts (Share Feedback/Close/Delete/Leave closed) and suggested edits is the only queue in which I have never failed a single audit. Why should suggested edits reviewers not be held to the same standards as other reviewers?

My proposal is to implement some form of suggested edit audit which will fail reviewers who reject them to further educate reviewers and reduce the number of rejected helpful edits. Though possibly not reject and edit as a user who intended to replace the edit with an even better improvement should not be punished. What are some ways this could be implemented?

Answer 1

Current audits are auto-generated by adding random nonsensical text to various parts of the post. A potential implementation of a known-good audit could be auto-generated by finding a positively scored post (more likely to have 'correct' formatting) and displaying a 'revision' that changes a fictional version of the post with randomly generated bad changes back into the current state of the post, similar to (but not the same as) the reverse of an existing known-bad audit.

For example, the system could find a post then display a suggested edits audit which 'revises' a non-existent and broken version of the question, such as by removing code fences or replacing 'I am' with 'im' or similar, back into the actual state of the question.

Example audit applied to a hypothetical example question:

Displayed 'old' version:

im trying to concatenate strings in C. im expecting result to be Hello World but my code is giving a segmentation fault instead. Could u please help?

#include <stdio.h> #include <string.h> int main(void) { char *str1 = "Hello "; char *str2 = "World"; puts(strcat(str1, str2)); return 0; }

Displayed 'new/revised' version (actual current version of the hypothetical question):

I am trying to concatenate strings in C. I am expecting the result to be Hello World but my code is giving a segmentation fault instead. Could you please help?

#include <stdio.h>
#include <string.h>
int main(void) {
   char *str1 = "Hello ";
   char *str2 = "World";
   puts(strcat(str1, str2));
   return 0;
}