[E-voting] 'Integrated end-to-end testing has been carried out and papers are available'

Colm MacCarthaigh colm at stdlib.net
Mon Apr 12 12:08:42 IST 2004

On Mon, Apr 12, 2004 at 10:58:21AM +0100, Shane Hogan wrote:
> The subject line is a direct quote from Peter Greene (Dept Environment) at
> the Joint Oireachtas Committee on 25th Nov 2003. I was kind-of surprised at
> this comment, as my requests for test results/papers/documentation from
> summer 2003 had not shown up anything that could be called 'end-to-end'
> testing. So I emailed Peter shortly after and requested a copy of the
> 'available' papers. I got repeatedly fobbed off or ignored, so I finally
> submitted an FOI request in early March, and a pile of test papers were
> provided to me last week (on the final day of the 30 day limit for response
> to FOI requests - Nice helpful attitude eh?). Here's my analysis & findings
> - I can provide soft copies of the papers to Margaret for the ICTE site (or
> to others on request).

At some point, we should compile a summary of all our experiences of the
departments "open attitude" to requests for information and file a 
formal complaint with the ombudsman. This is not the first time they've
used the 30 day limit so cynically. 

> In general, I am surprised at the absence of formal test documentation, i.e.
> test objectives, test methodology, bug logging/prioritizing process, roles &
> responsibilities - all the stuff that would be 'best practice' for a formal
> test of any IT system. 

Personally I wouldn't be the least bit surprised at this :( This is the
real problem; the department staff don't know what they're at, they are
not qualfied for all of this.

> In relation to the first set of tests (Buncrana & Athy), I noted the
> following;
> *	For the Athy test, the test result matched for the first 8
> candidates, but for the 9th candidate, a different person got elected on the
> eVoting system! DW explains this as follows; "some minor differences in the
> final results table can be attributed to the sampling of votes and probably
> a couple of errors in inputting the 2448 votes. However, the first 8
> candidates were elected while the 9th was different, given the closeness of
> the votes, and it should also be noted that there were the same number of
> counts in the IES programme as in the paper count." I really need to get a
> copy of the original paper count results to investigate this further - Any
> ideas as to where I might find this (http://election.polarbears.com/ doesn't
> have local elections). 

It will be in Nealon's guide, which I don't have to hand right now
though :/ You could ring the returning office in Donegal though, they
should have them to hand.

> *	The test papers were printed on 2nd Jan 2002 (Buncrana) and 11th Jan
> 02 (Athy) from IES version I-0.05 (so it precedes the ERS tests of IES which
> found many bugs);

... showing the complete irrelevance of these tests, and indeed *any*
tests until the actual software we will use on election day is
delivered, *sigh*.

> *	The 'open poll' statement as printed on the Nedap VM shows the
> polling date of 11/6/04. This indicates that either a) they set the clock
> forward on the machine, or b) the printout shows the polling date, not the
> current date. This may be a subtle point, but we can possibly use this to
> highlight the importance of VVAT, i.e. the machine can print/store anything
> it likes, but unless/until there is a paper audit trail verified by the vote
> which overrides anything stored on the machine, we can have no trust in the
> end result.

When you view the demo machines, they all say 11/06/04 as well, seems to
be that the machines are just constantly in one day.

Colm MacCárthaigh                        Public Key: colm+pgp at stdlib.net

More information about the E-voting mailing list