20
$\begingroup$

I don't intend this to be directed at the Mod and User in question, but this instance has me questioning what kinds of edits to posts are acceptable here.

See this recent edit to a post of mine. It is:

  • trivial to the point of uselessness, there is zero practical benefit to the reader,
  • requires an additional package unrelated to the others to be installed before the code can be used/run.

By "trivial" I don't mean specifically "small"; what I do mean is "inconsequential" or "superficial", where there is no improvement per se following the edit.

In the linked instance, the edit involved changing from using a standard base R data frame to the modern reimplementation of data frames from the tibble package.

Also, the intent comes across as ideological. By "ideological" what I meant specifically was the ongoing debates about base R versus tidy vs data table. I don't think this is specific to R, so that part of the question pertains to how should we treat edits wherein code is edited to change it from one paradigm to another?

I know that I as the author of the post have an overrule, which I exercised, but as this was approved by a Mod I wondered if it was me that was out of step here?

$\endgroup$
12
  • 10
    $\begingroup$ You don't seem to be out of line. In general, trivial edits shouldn't be approved, & the standard can get stricter as posts are older. But also, sometimes things slip though. $\endgroup$ Jun 16, 2020 at 15:38
  • 5
    $\begingroup$ What you did was definitely not out of line. Edits to older posts (usually more than a few days) and small edits to answers usually shouldn't be approved unless they obviously correct errors. $\endgroup$
    – whuber Mod
    Jun 16, 2020 at 15:51
  • 1
    $\begingroup$ Sadly so many other packages live in the tidyverse that many users (like me) find they have it installed already without ever wanting it directly. $\endgroup$
    – mdewey
    Jun 16, 2020 at 16:02
  • 1
    $\begingroup$ @mdewey Not sure why "sadly"? I don't actually care one way or another about this ideology; I find tidyverse has it's uses and use it when it does, and don't when it's easier to use something else. I've developed packages that use either base R or tidyverse code, so it's not like I'm anti-tidyverse. $\endgroup$ Jun 16, 2020 at 16:18
  • 4
    $\begingroup$ I think this question mixes small issues on two quite different levels. The first level is general, about small edits and how far they are distracting, In my view, poor presentation is itself distracting to many, regardless of whether the question is clear overall. In that sense, minor edits can help every reader. $\endgroup$
    – Nick Cox
    Jun 16, 2020 at 18:18
  • 1
    $\begingroup$ If I understand correctly, the second level is to do with how, and how far, R code should be explained -- in certain specific senses. I think there is also an assumption here that the R issues called "ideological" are self-evident to readers on Meta. I don't think they are likely to be self-evident to all of those who mostly or entirely use quite different software. $\endgroup$
    – Nick Cox
    Jun 16, 2020 at 18:20
  • $\begingroup$ Previous threads on editing include stats.meta.stackexchange.com/questions/2810/… stats.meta.stackexchange.com/questions/5816/… So far as I am aware R ideology is a new issue on Meta. $\endgroup$
    – Nick Cox
    Jun 16, 2020 at 18:25
  • 3
    $\begingroup$ @NickCox The size of an edit is not really in question (by me anyway); I would hope that a good but small edit that improved the post would be acceptable. What I meant by "trivial" was that it added nothing (& likely cost more to run the code, and also was incomplete - if you have a tibble you shouldn't be using with() etc), didn't change the practical behaviour, etc, etc. I'd be just as miffed if someone came in and edited the entire post to use tidyverse tools, because what's the point, aim? It wouldn't change anything other than make it more difficult for non-tidyverse-users to understand $\endgroup$ Jun 16, 2020 at 18:30
  • 2
    $\begingroup$ @NickCox This kind of thing ("ideology" based edits) occurs on Stack Overflow from time to time, especially on older posts. I almost always reverse the edits there for the same reasons. Invariably it's done by low-rep users and not by the people who actually develop the tools of interest. $\endgroup$ Jun 16, 2020 at 18:34
  • 2
    $\begingroup$ @Gavin Simpson Fine by me, but detail crucial to understanding your question is appearing in the comments, and on Meta as well as on the main site we then encourage posters to edit the question itself when that happens. If you don't have a question about small edits in general the only hint in the title about your focus is "ideological", which itself is hardly a neutral word. I would be happy if this question were focused on edits to or about R code, on which I should remain silent, except that i don't have enough space in this margin to express thoughts about the tidyverse adequately. $\endgroup$
    – Nick Cox
    Jun 16, 2020 at 18:38
  • 2
    $\begingroup$ @NickCox I see what you're getting at; I'll get my editing pen out and add some info from the comments to the question. $\endgroup$ Jun 16, 2020 at 19:29
  • 6
    $\begingroup$ I personally have an aversion to the tidyverse, which I explain here. This particular edit only made your post harder to read - before, any user of base R could understand it, now it requires (some small) knowledge of the tidyverse. As such, it is a clear disimprovement, and I would see rejecting the edit as a no-brainer. I honestly don't see any ideology at work here. (Then again, it's probably me that is the ideologue, and I just don't see it.) $\endgroup$ Jun 17, 2020 at 13:15

3 Answers 3

13
$\begingroup$

The code in the question was this:

df = data_frame(group = as.factor(rep(1:5, each=100)),
                x = rep(seq(-3,3, length.out =100), 5),
                y=as.numeric(dnorm(x, mean=means[group]) > 0.4*runif(10)))

In your answer you adapted it to

df <- data.frame(group = as.factor(rep(1:5, each=100)),
                 x = rep(seq(-3,3, length.out =100), 5),
                 y = as.numeric(dnorm(x, mean=means[group]) > 
                       0.4*runif(10)),
                 dummy = 1) # dummy variable trick

, & note that as well as appending dummy, you changed data.frame (from the base package) to data_frame (from the tibble package). This gives an error with the data.table function because group doesn't yet exist when you try use it to use it to create y—but doesn't with the data_table function, which is an alias for tibble. When the editor was fixing your code they presumably changed data.frame to tibble rather than back to the OP's data_frame because data_frame is now deprecated.

So there were good reasons for the edit in this case. But if there were someone going round replacing data.frames with tibbles (or vice versa) just for the sake of it they oughtn't to be.

$\endgroup$
1
  • 5
    $\begingroup$ Ah, I see the group issue now; this (original) answer required a lot of digging around in what mgcv was doing and at the time it also wasn't clear how well mgcv handled tibbles (now we know this doesn't matter). In my script that I wrote to prepare the answer I had defined group externally but that didn't make it into the code in the answer. I'll rectify that. $\endgroup$ Jun 17, 2020 at 15:44
5
$\begingroup$

Should trivial edits be acceptable?

What is trivial? Not all 'trivial' edits are useless (sometimes a comma can change the entire meaning of a sentence). When it comes to review suggested-edits, use your best judgement and stick to the guidance provided in the review dialog.

For example, that edit should have been rejected as it changes the content from OP's post. I would have rejected as:

clearly conflicts with author's intent

This edit deviates from the original intent of the post. Even edits that must make drastic changes should strive to preserve the goals of the post's owner.

Other two reasons from the review dialog would also fit:

  • 'no improvement whatsoever'.
  • 'attempt to reply' (should have been written as a comment).
$\endgroup$
1
  • 3
    $\begingroup$ Thanks; I have clarified what I mean by trivial $\endgroup$ Jun 16, 2020 at 21:11
0
$\begingroup$

When providing an edit, the editor should also include feedback as to what and why the changes are being made. It sometimes happens that what appears to be trivial to the author is not at all trivial. For peer reviewed journals, the default procedure for answering a reviewer is to accept whatever the reviewer says. This site is structured more ineffectually than standard peer review, as points are awarded to those of moderate reputation who do edits, so that there can be more garbage reviews than for a standard peer reviewed journal. (Edit) There is an opportunity to provide a justification for not agreeing with peer review when rolling back an edit back and explain why in the edit summary box on this site, whereas for a typical journal this would be a bit less ambiguously directed communication, by just explaining why not to make a change to the reviewer. Herein,

  • When providing an edit, make it absolutely clear why this is being done, there is only a one time opportunity for doing so.
  • When reversing an edit be absolutely certain that this is correct. There may be no other opportunities to correct an error that may be insidious.
  • In cases of disagreement as to edits, the author should consider providing a comment to the attention of the editor. Although this is not an obvious feature of the software, it is actually present and should be used for dispute reconciliation.
$\endgroup$
18
  • $\begingroup$ @AndreSilva We differ only as to emphasis. If a reviewer for a peer reviewed journal makes a suggestion that is not harmful, I usually default to that opinion. To do otherwise is to argue just for the sake of stylistic issues, and granted some people cannot resist arguing about unimportant points, however, there are more interesting ways of spending time. If you want I can change the style of the above, or, if you wish, you can do it for me, please. $\endgroup$
    – Carl
    Jun 20, 2020 at 3:10
  • $\begingroup$ @AndreSilva I would do more on this site, but have concerns about authorship rights, treatment of users, misuse of downvoting (e.g., for kicks), etc. On the plus side, I have learned how to communicate with statisticians a bit better, which is no mean feat since the concepts are sometimes quite isolated from the rest of scientific language usage. I did the edit you were suggesting, nothing wrong with that suggestion. Thanks for the compliment, you made my day. $\endgroup$
    – Carl
    Jun 20, 2020 at 4:43
  • 1
    $\begingroup$ You've made negative comments and posts about downvoting before. If I wanted to stir things up, I'd say that there is more negativity in those remarks attributing poor motives to downvoters than is really manifest in the downvotes themselves. I don't know what people's motives are unless they explain them! Does anyone else know more? . Anecdotes: (1) I've experienced occasional unexplained downvotes and they are puzzling, but what the heck, there is plenty of evidence that I irritate some people. $\endgroup$
    – Nick Cox
    Jun 20, 2020 at 7:53
  • $\begingroup$ (2) In 7 years here I think I've experienced just one "enemy" who dislikes much of my posts but regard hundreds of people here as virtual friends because I recognise their identifiers and appreciate what they do. (3) Least personally, I suggest it's exceptional here to see any post with more than say 5 downvotes, and in the vast majority of cases it is a bad post and "piling on" is a collective way of flagging spam or rudeness or something outrageously wrong. That's voting working as it should. (In politics too, many of us vote to keep poor candidates out rather than mediocre candidates in.) $\endgroup$
    – Nick Cox
    Jun 20, 2020 at 7:56
  • $\begingroup$ On editing, explaining edits is not worse than not explaining them, but the suggestion here is even the wrong way round. If an edit needs explaining, it's an inappropriate edit. Edits should be of presentation only, so start with spelling, punctuation, grammar, style, Markdown, etc. Is it the case that people with poor English or sloppy presentation want to be told emphatically that is the case? (And, yes, in many cases it's because English or American is not their first language.) Naturally, if people want to see messages like "small edits to your English" as a summary, that's a view. $\endgroup$
    – Nick Cox
    Jun 20, 2020 at 7:59
  • $\begingroup$ @NickCox Geepers, cool it will ya? I react to stimuli different than you do, and have a different approach to problems than you. That does not make me wrong about everything, just different. It is fair to discuss our differences but not if it is causing you pain. How is "If I wanted to stir things up, I'd say that there is more negativity in those remarks attributing poor motives to downvoters than is really manifest in the downvotes themselves." not stirring things up? I do not understand why criticizing downvotes sets users off. $\endgroup$
    – Carl
    Jun 20, 2020 at 8:54
  • $\begingroup$ @NickCox ...But it does. All I said is that it dissuades me from wanting to contribute to this site. Try to understand that. If you want me to explain, please ask. If not then don't ask. $\endgroup$
    – Carl
    Jun 20, 2020 at 8:58
  • $\begingroup$ @Carl Goodwill here, really. You missed the attempt at irony, or tongue in cheek. "I disagree with you" is always pretty close to "I think you are wrong" but having a discussion at all shows respect for the other side, and here too. You have criticised downvoting repeatedly, even when Meta thread after Meta thread shows if anything that you are in a small minority on that issue. I don't expect to change your view but won't let your throw-away comment pass without expressing a different one. $\endgroup$
    – Nick Cox
    Jun 20, 2020 at 9:05
  • 1
    $\begingroup$ When I referee papers for journals I do not expect authors to accept what I say but I do expect them to provide a reasoned explanation of why I am wrong which, being human, I may well be. I think that more nearly parallels what should be happening with edits. $\endgroup$
    – mdewey
    Jun 20, 2020 at 15:44
  • $\begingroup$ @NickCox I was captain of my high school chess team and the most successful player, but didn't like it. Took a long time to figure out why I didn't. It boils down to engaging in spurious competition in a zero sum game whereas cooperation, e.g., in a science project, is open ended positive sum. Sure, there are people who are enamored of competition in zero sum games, may you enjoy it, although if I do not I hope you will at least try to understand. $\endgroup$
    – Carl
    Jun 20, 2020 at 19:21
  • $\begingroup$ @mdewey My last serious research paper had 12 referees refuse to do the review, My response in the last review round by 6 reviewers went to 14 pages. Most of the changes were cosmetic, many of them were correct, some of them asked for tons of unnecessary extra work driven more by curiosity than reason. If I would have argued that the extra work was unnecessary, I would have had to do the work to prove that anyway. Did I disagree with the reviewers? Yes, but only about things that did not matter anyway, and that only for a small minority of things. $\endgroup$
    – Carl
    Jun 20, 2020 at 19:32
  • $\begingroup$ @mdewey If I had addressed the reviewers as if they were competitors rather than what they actually were, sincere readers concerned with creating something good, it would not have gone well. As it was, it took 5 or so review sessions to get the paper accepted. Why so many? Because the theory was new enough and so different, antithetical to existing material, that presenting that new material was very difficult as even the terminology used was at ground zero. In turn when I review something, I expect the authors to pay attention, and often they cannot because of lack of adequate training. $\endgroup$
    – Carl
    Jun 20, 2020 at 19:41
  • $\begingroup$ @mdewey Now a naive approach to writing anything is to view it as a competition between the text and "good enough" to make the cut. When I review my criterion for acceptance is whether the material is constructive and represents something that adds to knowledge or destructive and represents a step backwards. I recommend acceptance of 1 paper in 5, approximately, which is in line with the acceptance rate for the nearly two dozen journals I review for. Am I wrong about my comments during review? Sure it could happen, but nothing comes to mind. For me, it is a minor consideration. $\endgroup$
    – Carl
    Jun 20, 2020 at 19:57
  • $\begingroup$ @Carl Your approach to paper reviewing shows that your standards are high, and long that may continue. Editing and voting (down as well as up) are some of the means we use here to maintain standards, including variously improving what we can and criticising poor questions and answers. Everyone's contributions here are utterly optional -- except for moderators who have obligations while they serve -- and yours like anyone else's are just whatever is congenial to you. $\endgroup$
    – Nick Cox
    Jun 21, 2020 at 9:44
  • $\begingroup$ @NickCox Thank-you. I hear you, downvoting is what is done currently, and my claims are heresy. My downvoting claims go to efficiency. Downvoting tends toward zero sum competitive gamesmanship, whereas editing and commenting (and upvoting) are more cooperative and open ended in nature. I would not design a site to include ineffectual competitiveness. It is possible that the quite negative reaction to my downvoting heresy is more due to confirmation bias than due to consideration of merit. $\endgroup$
    – Carl
    Jun 21, 2020 at 10:24

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .