15
$\begingroup$

Something I've seen a lot of in response to questions (mostly low-quality posts from new users) which significantly lack enough detail to understand the issue, is some users here immediately jumping in to ask the OP for their blend file so we can take a look.

While sometimes this may be a quick-ish way to sove OP's particular problem, if someone here wants to take the time to open the file and play around, I would argue this method should be avoided where possible as it seems to me as if it doesn't fit the SE philosophy for a couple reasons.

  1. It's cumbersome for future reference. One of the primary purposes of SE (one may even argue the primary purpose) is not just to answer your particular question, but do it in such a way as to provide a helpful reference for future people with the same issue, to create an encyclopedia of sorts. An encyclopedia which requires downloading files to even understand what a post is about is not a convenient encyclopedia to use.
  2. It might encourage lazy asking practices. As Ray Mairlot alluded to here, being told right off the bat that they can just upload their blend file, users may get the idea that "here's my .blend, plz fix" is an acceptable way of asking questions here. This doesn't seem to fit the SE requirements of minimum acceptable effort and desire for high-quality content.
  3. It puts extra burden on us, the community. Blend files may be great for troubleshooting an issue, but something they can't do at all is explain what the issue is to us. This overlaps a bit with the previous point, but we really don't want people getting the idea that they can get out of explaining their problem by posting a blend file.
  4. It makes questions inaccessible for some people to help. This point may be too specific to me personally, but I suspect I am not the only one in similar circumstances. Most of my time spent on Blender.SE these days is while I'm at work, either over lunch or just taking a peek now and then to relax my brain while something I'm waiting on runs. As a consequence, if a question consists of only a blend file with a few sketchy comments about an issue I am unlikely to be able to do anything. If a post contains enough information to understand the question in text and images however, I may be able to at least offer some helpful comments and at bare minimum be able to handle the post correctly if it shows up in a review queue.

I absolutely don't want to give the impression that I am against posting blend files in questions at all, but I do think we should discuss how quickly we jump to the "can you please upload your blend file?" comments on low-quality posts.

Can we maybe attempt to craft some metric or litmus test by which we determine a post contains enough self-contained information that providing a blend file will only improve its quality? (Similar to how we judge link-only answers vs. answers which contain helpful links.)

$\endgroup$
10
  • 6
    $\begingroup$ Fully agree. sharing a file should be a last resort, and even then most of the time it is a strong sign that the question is too localized and not a good fit for this site. Similarly, I feel posting files in answers is detrimental to the learning process. By providing a ready made solution one doesn't encourage recreating steps nor learning by taking the time to understand the process. $\endgroup$ Mar 16 at 23:44
  • 2
    $\begingroup$ Another side effect is the rubbish tip buildup blender.meta.stackexchange.com/a/2767/15543 $\endgroup$
    – batFINGER
    Mar 17 at 8:05
  • 2
    $\begingroup$ Often just describing the problem in detail will help someone to find the problem themselves anyway, as well as allowing someone else to replicate the same situation to be able to offer a solution or even just to learn about that perticular area, even without an answer. To me BSE is a place for learning so just posting a file and expecting someone to fix it is not appropriate. I kind of agree with @Duarte about files in answers, but solutions can sometimes be extremely complicated and a working solution can be useful to allow someone to unpick exactly how it's put together. $\endgroup$ Mar 17 at 14:33
  • 2
    $\begingroup$ @DuarteFarrajotaRamos I definitely think there are situations where including a file is only beneficial. There are scenarios where OP describes the problem well but the included blend file allows answerers to show the solution directly on OP's file (this not only is nice for OP but also adds clarity for all readers by having direct continuity between question and answer). I've also had some answers where I give a node setup and explanation but include the file so they don't have to copy a 20+ node network from a screenshot. Do you have a suggestion of how we should judge when it is acceptable? $\endgroup$
    – PGmath
    Mar 17 at 15:21
  • 1
    $\begingroup$ @PGmath Those seem like a fair use cases, there are obviously exceptions where a file is indeed useful. I just think mostly new users in the "novelty phase" trying to help every low quality post jump in too soon asking for a file before even trying to understand the issue. For everything else I suppose using your common sense suffices. Question/answer continuity, and situations where recreating the exact OP setup would just be too time consuming to properly answer, are good candidates for asking for a file. $\endgroup$ Mar 17 at 15:34
  • $\begingroup$ For answers I have a stricter view, if it does require such a hard to recreate setup then the question was probably too broad to begin with, but it is certainly not something I'd care to start enforcing. If you feel like providing a file for sake of simplicity, then by all means do, I'm sure future users will appreciate the convenience too. $\endgroup$ Mar 17 at 15:39
  • 1
    $\begingroup$ Certainly for answers the answer should be stand-alone without the file - so it’s only an inconvenience if the file wasn’t there. Same goes for questions - the question should stand as it is without needing the file to understand the issue. I agree with @PGmath that it’s definitely helpful to be able to use the OP’s file for continuity with the answer in some cases. $\endgroup$ Mar 17 at 16:37
  • 3
    $\begingroup$ I feel like the most generally applicable answers are composed of primitives, and as such should be easily illustrated with screenshots. Same could go for questions but the onus is on the asker $\endgroup$ Mar 17 at 16:55
  • 1
    $\begingroup$ In general, I think that while having a file is nice, it's not always necessary. I think it tends be more useful to have images and gifs than files unless the Q happens to be really interesting and just needs a file to completely understand.. But if the question is some variant of "my object doesn't look right" and the answer is "recalculate normals" then it's a waste of everyone's time, OP included. Regarding answers, there are some around that would benefit greatly from having a file, to make learning simpler to some who would benefit years after, but these are mainly shader related. $\endgroup$
    – Timaroberts Mod
    Mar 17 at 22:30
  • 1
    $\begingroup$ @PGmath I understand what you mean, especially points 1., 2. and 4.... there are also sometimes questions where the problem is described well, but (a few) users immediately comment "Please provide your blend file." But in most cases it's not necessary. When suggesting to the commenter no file is needed to find an answer, I'm told that the problem might be identified easy for experienced users like me, but he is still learning and could need a file to understand. That's the moment where I keep calm but think, "well, then let the experienced users find the solution and learn from the answer." $\endgroup$ May 31 at 10:05
6
$\begingroup$

When should we be asking question askers for blend files?
Sparingly, only when necessary.

I very much agree with the question, and have had and stated this position for years.

To quote from this answer:

There should not be such an emphasize on "the blend." Granted, blends are immensely useful, and help get questions answered quickly. Yet there is a balance, not everything needs a blend. Questions should use text as the number one tool through which information is conveyed.

The "fix if for me" type questions have gotten much more numerous in the six years since I wrote that. If we had treated the blend as a "last resort" instead of the first tool we reach for when solving questions I expect the quality of questions would of turned out differently.

I can think of situations where blends really can solve the issue. Cases such as rigging, material settings, or drivers. Its so easy to mess something up, and have no clue where something went wrong. Or even something simple like modeling (I know the bisect setting in the mirror modifier will get many), the OP could make a simple test case, do what they think is the same thing and have it work yet in the original file it does not.

There always will be cases where blends are immensely useful, and so far I have just been talking about the question.

I think answers should always be free to add a blend if the person answering feels it would help.
For example in this answer I laid everything out in text and the pictures, but still felt it would be a benefit to have a premade example to compare the 4 different methods (and apparently it has been a help currently there are 321 downloads on that file).

The trouble comes in defining what is "necessary" to have a blend. I agree we should come up with a standard for measuring and distinguishing LQQWB (low quality questions with blends) vs SQTNB (short questions that need blends).

As an overly broad baseline I'd start by testing to see if the question is still discernible without the blend. (I happen to have an example here) If it is, then it might be a good question simply helped by a blend. On the other hand, if the question loses all meaning without the blend then I'd say that is a low quality question and deal with it accordingly. Likely Closed as "Needs details or clarity" or "Too localized" (doe not really matter which) with a comment somehow or another explaining how blends can not be the biases of a question and questions need details in text.

Here is an exercise for the reader, taken straight out of the current HNQ list: Eevee render output is black What would you do with that question? Apparently that type of question is well received now (its on the HNQ!?!)
Is the blend supplemental, or would you call that a CDLOQotHNQL (cleverly disguised link only question on the hot network question list)? What action should be taken on it and questions like it?

OK so I have a thing for (dumb) acronyms.

$\endgroup$
1
  • $\begingroup$ What (pleasant) a surprise to hear from you! Your 4th/3rd to last paragraphs are pretty much spot on as far as what I was thinking: blends are fine in some (maybe even many) cases but the question (or answer) must be able to be understood without the file. It seems to me as if that's a good starting point, I think it fits with an intersection of the commenters' opinions and we can of course in the future refine the criterion or tighten it as necessary. $\endgroup$
    – PGmath
    Mar 23 at 16:56

You must log in to answer this question.

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