> I think the problem with this laudable objective is that there is no way to
> take a scientific sample of (La)TeX users. Only a microscopic fraction of
> them belongs to any one well defined group --- like people reading
> comp.text.tex --- or TUG, or DANTE or GUTenberg or NTG or CyrTUG ---
> or reading Notices of the AMS or SIAM say.
That is certainly the conclusion that we came to several times previously.
If anyone has any good ideas about a sampling frame for such an
activity, for TeX and/or LaTeX, please let us know.
> And if you perform the survey in comp.text.tex you will get an extremely
> biased view of the world!
And how, almost as bad as LaTeX-L:-).
> One outrageous idea would be an access count on packages on CTAN.
> Although this would not take into account redistribution at local sites and
> would be swamped probably by `mirroring' access and `collectors/pack rats'
> (who must simply have everything of any kind whether they need or not)
> rather than `normal' end users.
But it might be interesting data.