6
$\begingroup$

My favorite thread on this site so far is this one: Is there a free package with robust CASSCF functionality?, because six different people chipped in, and somehow many of us decided to follow the same format, even though there was no written rule. I could easily have listed several programs in my answer, but I decided to focus on one: This meant more effort into quality rather than quantity, but also gave others better opportunity to participate, which I think is good "sportspersonship" (and incidentally raised our average number of answers per question 😊).

When similar questions came up, I encouraged people to follow this type of format, but did not say anything further when they decided to reject the suggestion.

However, after seeing this:

I think we should decide how our community wants to treat this type of situation. Does the 4 upvotes for MRCC and 10 upvotes for Psi4+PySCF mean that the community is less in favor of MRCC than Psi4 or PySCF? Or could it mean that 4 of us like the suggestion of MRCC, 9 people like the suggestion of PySCF and 1 person likes the suggestion of Psi4?

Tyberius wrote in one of these cases: "I would say a question like this will work best if people stick to one framework per answer, so just doing Monte Carlo for now is perfect." which I 100% agree with.

  • If we do choose to stick to this format for certain questions like these, do we want to encourage each person to provide just one answer? All these questions have something in common, which is that a large number of people could easily list a large number of items, but so far most of us have chosen not to "hog" the whole question and instead to focus on quality and in-depth explanation rather than quickly listing as many examples as we know. For some questions like this one of mine the user can specify that they would be happy to see multiple topics per answer or one topic per answer (I did this because I knew it was a difficult question that likely only someone specializing in basis-sets would be able to answer, so limiting them to just listing one basis-set family would not make sense). However in the above examples that followed the proposed format, most people that answered could easily have given several other suggestions too, but refrained from doing so to give others a chance.
| |
$\endgroup$
6
$\begingroup$

This is a good point, I think we should try to encourage more focused "single-answers" than longer "multiple-answers".


I agree with everything you've said and will only point out a few more reasons answering with one answer is better:

  1. The opportunity to vote for specific answers is a central feature of stack exchange. Voting for the answers you like raises them to the top. Therefore, the more popular answers will rise to the top. While more popular does not always mean better, there is usually some correlation between the two! In questions like What are the milestone achievements in density functional theory in recent years? and Is there a free package with robust CASSCF functionality? I imagine that the votes will indicate the community's preference, which is valuable to know.

  2. It looks better. The extra whitespace, nametag, and voting bar help to break up the answer into easier visual pieces, which is not possible even with a well formatted multi-answer.

  3. It's easier to write a focused comment or edit. In the examples you gave it was interesting to see the focused discussion in the comments which would be less obvious if the question listed several answers.

  4. It's easier to answer. This may be specific to me but I'm lazy. I'd prefer to write a short succinct answer to the DFT question listed above than the entire history of DFT!

Finally, with regards to sportspersonship. I think people can answer multiple times if they use discretion. I like User Uhoh's answer which basically said it should only be done rarely. The reason is people can somewhat try to game the system that way.

| |
$\endgroup$
  • $\begingroup$ Great answer. You raise some especially good new points, such as the fact that answers become easier now that you can just focus on one topic rather than feeling the burden that you have to give a comprehensive overview of all possible answer topics. Also, and especially good point that you made, which I hadn't thought of when I wrote my question, was that comments can be more focused. Imagine commenting on each milestone for an answer that lists 10 different DFT milestones!! $\endgroup$ – Nike Dattani Jun 4 at 1:11
  • $\begingroup$ Good points. I have a small concern though. If we stick to single answers, isn't there a potential lose of great answer. What if a person A had some really good points and he restrict himself answer only one and no one else posts the other points as answer? $\endgroup$ – Thomas Jun 4 at 12:14
  • 1
    $\begingroup$ @Thomas maybe I overstated Uhoh's answer a little. He said it's fine and has done it a few times. I think discretion is key. Maybe if you have more than 2 answers for a single question you can write one or two, and then comment on the main Q that there are others like X,Y, Z but someone else can discuss those in more detail. $\endgroup$ – Cody Aldaz Jun 4 at 16:38
  • $\begingroup$ @CodyAldaz That's a good suggestion. May be we can follow this as a norm. Two answers maximum if deemed necessary and can leave comments to encourage others to participate $\endgroup$ – Thomas Jun 4 at 16:43
  • $\begingroup$ @Thomas I'm only suggesting this for cases where there's several easy answers that several people could easily give. In my question I referred to one example of mine, where the question was hard, and if someone has more than one example, I encourage them to give multiple answers because limiting them to one answer would be too restrictive. However in questions like "what software can do CASSCF?" I could easily have given 6 answers: OpenMOLCAS, PySCF, COLUMBUS, BAGEL, etc. and left nothing available for others. Usually people that do that give low-quality answers rather than in-depth answers. $\endgroup$ – Nike Dattani Jun 4 at 19:11

You must log in to answer this question.

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