Re: Proposal for an Emacs User Survey

[Ihor Radchenko]

> Let us say you wish to ask if users are using MELPA or what else?
> There is nothing else to compare.

At least we can get ratio between people using MELPA vs. not using
MELPA. As I stated, it might be useful later, when we have nongnu ELPA.
Or it might, for example, reveal that most of Emacs users are actually
not using MELPA (though I don't believe that it is the case).

Note that I am not arguing that we have to include this question in this
specific poll (there will be many other valuable inputs regardless of
this question). Just wanted to point out that MELPA statistics will not
answer the question what fraction of users are _not_ using MELPA. If
that fraction is significant (unlikely), taking only MELPA statistics
might be not representative.

Best,
Ihor


Jean Louis <bugs@gnu.support> writes:

> * Ihor Radchenko <yantar92@gmail.com> [2020-10-12 10:56]:
>> > Reason for that is that it is obvious that people do, people ARE using
>> > MELPA and Marmalade software repositories; AND more important reason
>> > not to ask is that information about usage of those repositories,
>> > likes, number of contributors, it is already available on the
>> > Microsoft Github "Insights" link on the MELPA page. There is no point
>> > in asking users what is already obvious. There are stars or likes on
>> > Github.
>> 
>> I disagree that there is no point asking. Assuming that we are
>> interested to know about MELPA/Marmalade usage, taking information about
>> usage/stars/contributors/etc from third-party sources will represent
>> different subset of Emacs users - it cannot be compared with other
>> results of the presently discussed poll.
>> 
>> All the MELPA/marmalade statistics is inherently biased. It only
>> represents Emacs users using those repositories. On the other hand,
>> asking about package repositories in this poll will provide us with a
>> concrete estimate how popular are MELPA and Mermalade. In future, when
>> nongnu ELPA is going to be up and running for a while, it may also be
>> interesting to see how the popularity changes.
>
> When doing an opinion poll, it involves investment of money, time,
> efforts, people, so the answers have to be evaluated with purpose to
> do something.
>
> Marmalade is anyway not working any more.
>
> Let us say you wish to ask if users are using MELPA or what else?
> There is nothing else to compare.
>
> ELPA is already built-in, there is no other package repository. Now
> you wish to find statistical information how many users use MELPA, but
> why not simply ask MELPA developers to provide such statistic? They
> probably have statistics of downloads or accesses to the
> website. There is no point to asking around the corner, when one can
> ask directly.
>
> Let us say one wish to evaluate which packages are more popular, you
> can take a list of package from MELPA, then run Emacs Lisp through the
> list, and then probably use Github API or web scraper, you can find
> which package has more stars/likes or number of forks. Similar results
> you can get from Github search. None of results, neither survey
> results nor Github API/search results are correct results, so it would
> be expensive making survey in that way.
>
> So, for number of MELPA downloads, anybody wishing to find out, one
> can ask there. Another question how they are evaluating the website
> statistics. But for which practical use?!
>
> For popularity of various packages, it is possible to use API (I just
> guess so) or web scraper, and obtain statistical information. But for
> which practical use?!
>
> Most practical is when user reports a bug, then it can be solved, or
> user reports enhancement to software, it could be implemented.
>
> This mailing list is a small survey itself, help-gnu-emacs mailing
> list as well, various answers on non-GNU websites also represent
> valuable data for improvements, at this moment there is so much data
> that demands improvements.
>
> Jean