I decided to do an informal survery of SPF record publishing in .edu domains,
specifically US colleges and universities. I chose this group partly out of
personal curiosity and partly because it is easy to find a fairly complete
list. (Yahoo is pretty reliable on colleges and universities.)
This was not a carefully designed or carefully recorded survey, but I came away
with some impressions I think are accurate.
1. SPF record publishing is not concentrated by region, state, administrative
unit or religious affiliation. Non publishing, although harder to judge, also
appears unaffected by the same parameters.
2. Total SPF record publishing so far is a small fraction of all the colleges
and universities.
3. Most SPF records are published by small to medium sized institutions. The
big exceptions are UC Berkeley and Columbia
4. Private, church affiliated and public institutions are all publishing. I
would say that the private and church affiliated schools are somewhat ahead, in
percentage, of the public institutions. This may be secondary to size.
5. Schools well known for computer science departments are no more likely,
perhaps somewhat less likely, than others to have published. This may also be
secondary to size.
6. "Technical" schools are no more likely to have published than general or
liberal arts curriculum schools.
7. Probably 2/3 of institutions with spf records have registered them at
http://www.spftools.net/register.php before I got to them. (I registered the
ones I found.)
8. The error and warning rate appears higher than recorded at the registry.
Most of the errors have been typos or simple omissions. A couple indicated
serious misunderstanding of the syntax.
Conclusions:
1. SPF record publishing is probably being decided on a one by one basis, very
likely by very small groups of people, perhaps only one, at each institution.
2. People composing records are not as likely to use the wizards and validators
as one would like. Essentially all of the errors would have been caught by the
online tools. None of the records appears to have come from the faulty
Microsoft wizard.