14

Should Wolfram Alpha Notebook questions be considered on-topic?

Here's an example: Calculating double integral bounded by domain in Wolfram Alpha Notebook

Here are some related meta Q&A:

Wolfram Alpha Notebooks are a new WRI product that hybridizes W|A and Mathematica. I looked only briefly, but it resembles a Mathematica notebook in which the only valid input starts with single-equals (probably without having to type =), though the sample inputs are sometimes interpreted differently in the examples shown than in my Mathematica.

(For those who may not know, it was decided to consider questions about W|A off-topic.)

8
  • 3
    FWIW I downloaded the W|A notebook thing when it came out and it's just a shell on the FE which calls into WolframAlpha as a function. Basically it's just nice formatting on a everything through Ctrl-= type of NB. So as far as FE questions are on topic, I think these are. OTOH since we don't support W|A questions from a syntax perspective, I think those parts are off topic.
    – b3m2a1
    Mar 1, 2021 at 4:54
  • “ the only valid input starts with double-equals (probably without having to type ==) ” Nope, it's singal-equal i.e. the default input of W|A notebook is free-form input.
    – xzczd
    Mar 1, 2021 at 6:48
  • @xzczd I guess that's right. It's not the exactly same as what I get from single-equal input, but it's certainly much closer to single- than double-equals. I didn't look closely enough at first. (Or maybe I did ctrl-=,.)
    – Michael E2
    Mar 1, 2021 at 14:35
  • 2
    I think the distinction @b3m2a1 makes might be hard to make in practice. (1) A question about the FE could be asked about the FE without involving W|A NBs, theoretically, unless the W|A NB FE has special restrictions/capabilities. (2) Users usually come to the site with their attention on solving their problem, not dividing it between FE and W|A issues. The same occurs when someone comes to the site saying "<query>" didn't work on W|A, and some helpful community member changes the question to WolframAlpha["<query>"] didn't work.
    – Michael E2
    Mar 1, 2021 at 14:44
  • 2
    A similar thing happens with questions that are mainly about mathematics or computational mathematics. Problems solely about mathematics etc. are off-topic; how to compute the solution well in Mma are on-topic. In the latter case, the solution to the OP's problem often turns out to be more about mathematics than programming Mathematica. This surely happens in other domains as well (e.g. physics, judging from comments). While W|A has some distinct characteristics from Mma, it seems the distance between them is diminishing, somewhat like the distance between math and Mma.
    – Michael E2
    Mar 1, 2021 at 15:16
  • 3
    I don't think we'll be inundated with them anyhow, so why not.
    – Chris K
    Mar 2, 2021 at 14:21
  • 2
    If we allow questions about W|A notebooks, should we allow questions about W|A (change our policy)? Distinguishing them seems problematic to me. (@ChrisK I expect we would become inundated with them in proportion to the extent W|A notebooks are adopted by schools and colleges.)
    – Michael E2
    Mar 2, 2021 at 15:23
  • @ChrisK, MichaelE2 I mean my sense is that anything related to a W|A notebook query is off topic, which I assume will be 90% of the questions we'll get anyway. If it's clearly distinguishable that there's really an FE question that can be fixed with Mathematica code that's on topic. That said, my inclination would be to say their off-topic, since we simply don't have the expertise on this site to answer W|A questions.
    – b3m2a1
    Mar 3, 2021 at 3:33

3 Answers 3

6

No, Wolfram|Alpha questions should be explicitly off-topic, regardless of whether W|A is being used through the web interface or Wolfram|Alpha Notebook.

Vote here if you agree (and post the opposite answer for voting, with appropriate arguments, if you disagree).

Arguments:

  1. The interface is a minor detail that should not decide whether W|A is off-topic or not. It would be quite ridiculous to tell people: "Your questions about this W|A input is off-topic because you used the website. Had you downloaded the GUI and typed the input there, it would be on-topic."

    As for asking about the GUI vs W|A input: as @MichaelE2 says, "I think the distinction @b3m2a1 makes might be hard to make in practice." I do not think most people who might post questions would be able to understand the distinction, unless they are also very familiar with Mathematica.

  2. The reasons from https://mathematica.meta.stackexchange.com/a/267/12 still apply:

    " I know that W|A runs on mma and it kinda sorta understands mma syntax. However, opening the door to such questions will only lower the bar and result in hit-and-run questions from folks who just want a quick result from W|A."

  3. Having expertise in Mathematica / Wolfram Language (which is what users of this site have in common) does not translate to Wolfram|Alpha. While W|A understands some Mathematica-like input, it often interprets it differently from Mathematica. W|A does not have a documented syntax. Its understanding of natural language is constantly evolving and often unpredictable.

  4. W|A (not W|A Notebook) has a much larger userbase than Mathematica, most of whom have no familiarity with Mathematica whatsoever, and would not fit in the current community. Additionally, there is good reason to suspect that many W|A questions would be of the hit-and-run type, coming from students looking for a quick answer or homework solution.

  5. It is difficult to maintain a community like the one organized around Mathematica.SE. There are even concerns that things are not going as well as they used to. In my opinion, inviting W|A questions (which are orthogonal to Wolfram Language questions) will bring nothing of value to the community. In fact, I worry that it might even be the last straw that breaks the camel's back.

Note that there are already places for W|A questions: WebApps.SE and Wolfram Community. There is no pressing need to create another one.

6
  • Items 2 and 4 above mentioning "hit-and-run" are the most important reasons for not including questions about W/A in any form.
    – JimB
    Mar 24, 2021 at 16:39
  • 1
    Yes, I think it's fair to point questions about WA notebooks directly to Wolfram Community (in fact, perhaps we can add a close reason with that, next to the "needs professional help" reason). I think it's potentially fine to have a slightly blurry line in case there are extremely interesting and worthwhile questions that may have a larger impact, but I would support a rule of thumb pointing those users to the community site instead.
    – Carl Lange
    Mar 24, 2021 at 18:47
  • 1
    (Hot take ahead) I don't personally expect W|A notebooks to be around for all that long anyway and I have a mild suspicion that they will end up on the WolframTones zombie project pile in a year or two. Maybe I'm alone but that prejudice on my part would prevent me spending any time thinking about W|A questions anyway.
    – Carl Lange
    Mar 24, 2021 at 18:51
  • 1
    @CarlLange don't forget CDF, Wolfram data drop, Wolfram functions, the NKS summer school, and so so many more. Oh and of course the Wolfram Data Science Platform ("Coming Soon!" since 2012). OTOH I think W|A is a big product for them in the university setting, so if this convinces schools to not drop Mathematica licenses it might have more staying power than expected.
    – b3m2a1
    Mar 28, 2021 at 17:24
  • 1
    @b3m2a1 Well, I gotta say I do use data drop pretty often, and to be fair one could have said the same about W|A itself, the NN repo, function repo, Cloud, etc, but yes, indeed. I would shy away from explicitly allowing questions on various WRI offshoot projects until they have shown themselves to be quite established.
    – Carl Lange
    Mar 28, 2021 at 17:41
  • I disagree with this answer. The example about the integral in the question is anyway very informative. The question of that example could be answered just using the language of Mathematica (disregarding completely the software used by the person who asked it)... but you voted to close it because the user who asked stated that s/he was using Notebook Edition, which makes no sense at all (it was a question about the Mathematica language). You can find below my answer (together with my answer to the example question using with Notebook Edition the same syntax you use with Mathematica). May 24, 2021 at 9:42
4

Thanks to Penelope Benenati, it came to light that Wolfram|Alpha Notebook Edition does not behave like Wolfram|Alpha. Unlike Wolfram|Alpha (which frequently interprets input differently from Mathematica), it appears to be able to understand Wolfram Language syntax.

Based on her screenshots, it appears to me that W|A Notebook Edition might work as follows (please correct me if I got any of this wrong):

  • It contains a Mathematica kernel which can evaluate Wolfram Language code locally.
  • It does not allow inputting Wolfram Language code directly. Instead, it interprets free-form input using Wolfram Alpha, in the same way as Mathematica does when we start an input using "=". Then it shows its interpretation as the actual Wolfram Language code that will be evaluated.
  • In most cases (in all examples shown so far), if the input happens to be valid Wolfram Language code, it interprets it literally. It is unclear to me if this is always the case of only sometimes.

Here is my updated proposal on what questions should be on- or off-topic:

  • Naturally, any question about the Wolfram Language should be on-topic regardless of software used.

  • However, questions about how to coerce W|A Notebook Edition to interpret free-form input in a certain way should stay off-topic. This is because there is no clear specification of what free-form input is allowed to look like. It's also unclear how frequently its interpretation changes. For the sake of example, if W|A Notebook Edition happens to interpret "Sqrt[x]" not as Sqrt[x] but as Surd[x,2], questions about how to get it to understand it as Sqrt[x] should continue to be off-topic.

  • If it is indeed the case that W|A Notebook Edition understands most Wolfram Language input with the exception of some specific functionality, it should be fine to ask how to solve a problem without using that functionality. For example, if some standard packages such as TriangleLink are not available, it is fine to ask how to solve a problem without TriangleLink. This is contingent on the asker being able to clearly explain what is and isn't available. In my view, this is no different from asking questions about the cloud which also does not support all functionality (but note that it does support most functionality and it is quite clear what isn't available and why).

It is crucial that W|A Notebook Edition explicitly shows the Wolfram Language code that it is evaluating. This ensures that there won't be any debates about differences in how W|A Notebook Edition and Mathematica interprets input. Without this certainly, I would not be in favour of allow W|A Notebook Edition questions. But thanks to this feature, it is easy to separate the interpretation of free-form input (questions about this should be off-topic) from the usage of Wolfram Language (which has always been on-topic, even for subsets of the Wolfram Language).


The question that triggered this debate is the following (now closed):

Plotting a complete graph with a given image as vertices

According to what I wrote above, I still consider this question to be off-topic. Notice that the asker mentions that the following code did not work:

g = CompleteGraph[8, VertexShape-> Import["~/google-chrome-yellow-png-image-69499.png"]]

However, this is perfectly valid Wolfram Language code that works in Mathematica. Neither the asker, nor any of the commenters seem to be certain why W|A Notebook Edition does not accept this input, other than the usual explanation that it accept unspecified free-form input, and it doesn't happen to consider this one valid.

Also notice this screenshot from the question:

Here we see valid Wolfram Language code, given as free-form input. W|A Notebook Edition fails to understand it as Wolfram Language code for unknown reasons. This is precisely the kind of problem which I believe should be off-topic here.

Finally, notice that it is completely unclear in what way the code in the accepted answer differs from the code that OP claims did not work. It probably has to do with the idiosyncrasies of how W|A interprets free-form input.

For these reasons, I did not vote to reopen the question.

2
  • I basically agree with your answer here, including the decision to close my question of yesterday about importing images. However, there is something that is not clear to me. Why do you write "It does not allow inputting Wolfram Language code directly"? I think NE (Notebook Edition) transforms free-form inputs into Wolfram Language, but in most cases when the input is in "pure" Wolfram Language, it just behaves as Mathematica. May 24, 2021 at 11:38
  • 1
    Szabolcs Now I understood that the square root issue was just an example. Anyway, I agree with your answer update. Summarizing, as you wrote, "any question about the Wolfram Language should be on-topic regardless of software used." And I also agree as I said that my question about importing images should have been closed as you did, because it resulted to be specifically about the Notebook Edition syntax instead of being about the pure Wolfram Language (I did not know it before to ask the question). May 24, 2021 at 12:32
2

Yes, most of the time (but not always). It seems to me that many users do not know well Notebook Edition and are not aware that Notebook Edition accepts almost all code syntax of Mathematica.

To answer this question, one should first define "Wolfram Alpha Notebook questions." I use Notebook Edition, and during the last nine months, I verified that for all the problems I had to solve, all the answers to my questions that I received here were beneficial (you can take a look at my questions and discussions here). The only difference I had to implement to use the suggestions I obtained was changing the variable name using only one character. About all the rest, I never found any difference.

As I said, Notebook Edition accepts almost all code syntax of Mathematica.

A question cannot be on-topic or off-topic based only on the software used by the user who asks it. Indeed, in principle, the user could also ask an interesting question about the Mathematica language by using his/her smartphone without any computer or math software (at all).


Hence, the answer is simple: questions about the language of Mathematica are on topic independently of the software used by the user who asks them; questions that are specifically related to the syntax of Notebook Edition (that cannot be used with Mathematica) or that are about the software Notebook Edition itself are off-topic.


Questions are on-topic or off-topic, not the software used by people who ask them. Otherwise, you should close all my questions of the last nine months, which were considered useful by several people here, just because I am "revealing" that I always used Notebook Edition while implementing the suggestions obtained (just using variables of one character solely).

Finally, since you mentioned above Calculating double integral bounded by domain in Wolfram Alpha Notebook as an example to clarify the above question, I am glad to show the screenshot I obtained by using Notebook Edition. The question is now: Do you solve that problem differently with Mathematica, in a way that I cannot implement with Notebook Edition? Thank you. enter image description here



I also add a screenshot of the result I get attempting to use the answer I received here: Plotting a complete graph with a given image as vertices :

enter image description here

7
  • 1
    Indeed it makes sense to accept questions about the Wolfram Language, regardless of interpreter. I was not aware that "W|A Notebook Edition" interprets input differently than W|A. However, personally I maintain that questions should be restricted to the Wolfram Language when it comes to W|A Notebook Edition. If W|A Notebook Edition treats some input differently than Mathematica does, questions concerning that should be off topic.
    – Szabolcs
    May 24, 2021 at 10:02
  • 1
    I do agree with you that "A question cannot be on-topic or off-topic based only on the software used by the user who asks it.", and I think my original answer was in that spirit. What you are probably not aware of is that we had many cases of people insisting that "W|A accept Mathematica input" when in fact it does not, and if sometimes even interprets it differently than Mathematica would. It is good that it came to light that "W|A notebook edition" is something different.
    – Szabolcs
    May 24, 2021 at 10:20
  • 1
    It seems to me based on your screenshots that "W|A notebook edition" contains an (almost?) complete Mathematica kernel, but it does not allow one to evaluate Wolfram Language code directly. Instead, it uses W|A to translate your input to Wolfram Language code, then it evaluates that. Crucially, it shows the actual WL code that it is evaluating. Do you agree with this description?
    – Szabolcs
    May 24, 2021 at 10:23
  • While I agree that what software you use does not matter, looking carefully at your now closed question show that W|A Notebook Edition failed to interpret perfectly valid Wolfram Language code. It seems clear that it still accepts only free-form input. There is no clear specification for what is or isn't valid "free-form input" and what its interpretation should be. It's just not a programming language. You accepted the answer there, but the code in that answer is no different from the code that you showed in your question and said that it did not work.
    – Szabolcs
    May 24, 2021 at 11:00
  • @Szabolcs I am not completely sure I have understood what you meant. I think NE (Notebook Edition) contains a Mathematica kernel, and it accepts free-form input too by translating them into Wolfram Language. However, it also accepts Wolfram Language directly - with few exceptions, e.g., the length of the variable names, which must be one, i.e, one character. I am sure that there are other parts of the Wolfram Language syntax that are not accepted by NE, but I feel very confident that they are very few. Why do you say that NE accepts only free-form input? May 24, 2021 at 11:25
  • @Szabolcs About my question of yesterday, I noticed that although the Input command works like with Mathematica, NE does not accept the variable assignment i=Input... . Hence, my question of yesterday concerns one of the very few parts of the Wolfram Language that are not accepted by NE while being accepted by Mathematica and therefore can be closed. On the other hand, the question example about the integral (here above), was a Wolfram Language question that should have been answered because the author would have been able to use it even in NE, instead of being closed. May 24, 2021 at 11:28
  • @Szabolcs I added in my answer here a screenshot of the implementation of the answer provided in mathematica.stackexchange.com/q/246519/74045 . As you can see, it is easy to bypass to issue that with Notebook edition I cannot assign to a variable the result of the command Import, and verify the provided answer is correct. May 24, 2021 at 14:15

You must log in to answer this question.

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