Elections software and Shibboleth

[Peter Schober]

* Russell J Yount <rjy at cmu.edu> [2011-10-06 14:03]:
> Does anyone have experience with software that could be used to
> provide Student or Staff Council election services using Shibboleth?
> The software would preferably be able to use attributes such as
> affiliation to determine voter eligibility and be able to handle
> some type of nomination process for candidates.
> The software could be on open source, proprietary, or software as a service.

I wrote a short (unpublished, german language only) "white paper" on
this a few years back when our national government wanted to
test-drive an electronic voting with the student union elections.
The devised (and used) system was based on the national citizen card,
a government-issued, sector-specific (targeted, though not pairwise)
smart card with full PKIX, qualified sigantures, etc. Not surprisingly
to some on this list, it has also seen virtually not use.

In short, I think that -- ignoring the general problems of electronic
voting or trustworthiness of voting mechanisms in general -- Shib/SAML
used with only affiliations (if required by the use case) and
persistentIds (a.k.a. eduPersonTargetedId) would allow for an
extremely simple machinery at the SP end (i.e., the voting system),
potentially even allowing storage of the persistentId together with
the vote -- if you're confident that the SP will not be able to
"reverse" some of these identifiers with help of the issuer/IdP
(otherwise storing them seperately in a database transaction would do).

The identity assurance could also be increased by issuing OTPs of some
kind to the voting population (e.g. via wire transfer of 0.01$ or text
messaging or postal services or whatever) beforehand, and then
transfering the data (encrypted attributes or via attribute queries;
certainly not via the browser) to the SP so the SP could ask the
principal for the secret token before accepting the vote (and grabbing
the answer from the assertion/environment), etc.

So in this naive view a Shib SP with a reliable storage service (that
supports transactions, in case you'd wanted to store persistentIds and
votes seperately) and very few lines of trivial code would do here.

The general problems of electronic voting remain, of course, e.g. how
to be sure that the code written is also the code executed during the
vote, that the storage is not manipulated during or after the vote,
etc. I have nothing to offer here and would expect that this is where
the black science of companies selling voting machines would come it.
-peter