57 responses were received so far. I should note in advance that I do not believe the sample (i.e. visitors to this site) to be representative of all users. Keep this in mind, and draw your own conclusions.
Mathematica versions
84% are on the latest version, 7% on 9 or earlier, 3.5% on 10.x and 5% on 11.x (excluding the latest).
Operating systems
For the following question, keep in mind that multiple answers were possible. Everyone who ticked the cloud or Raspberry Pi also ticked a desktop platform.
Using packages
Only about a third of the responders use packages ...
... even though most of them have tried packages at some point.
For how long have you been using Mathematica?
Most responders are long-time users. This, to me, seems to indicate that the sample may be heavily biased. Especially with the recent popularity of high-level languages (what physics student learns C or Fortran as first language anymore?), I would expect most users to have started with Mathematica recently. However, new users may not be frequenting this forum. A healthy and growing system should have many "fresh" users, so hopefully this result is not representative of the broader Mathematica userbase.
What about any correlation between regular package use and the for how long a responder has been using Mathematica? Here's a breakdown for the answer to "Do you regularly use third-party packages?"
- used M for 10 or more years:
<|"Yes" -> 11, "No" -> 19|>
- For 6-9 years:
<|"Yes" -> 2, "No" -> 8|>
- For 5 or fewer years:
<|"Yes" -> 8, "No" -> 9|>
It would appear that people who are newer to Mathematica are more likely to use packages, but the dataset is too small to draw reliable conclusions.
I am not going to publish the full dataset to protect the responders' privacy, but if you have any questions / requests, I will try to satisfy them.
I have also written to Wolfram Support to ask if WRI has any statistics on the current share of various versions. I have not yet received a response, and frankly I will be surprised if my request is not refused. While I am sure that WRI collects such data (remember that Mathematica regularly contacts Wolfram servers to check for paclet updates, so it is technically possible), I can imagine numerous reasons why they would not want to share such statistics publicly. However, if they do respond, and if they do agree to sharing this information, I will post it here.
I will keep the survey running for a little longer and will link to it from other websites too. The date of each response is recorded, so it will be possible to differentiate between the old and new responses.