[E-voting] JCELG meeting of 2003-12-18: an observer's notes

Adrian Colley aecolley at spamcop.net
Sun Dec 21 15:34:06 GMT 2003

On Thu, Dec 18, 2003 at 06:02:09PM +0000, Margaret McGaley wrote:
> There were several representatives from the government's side, including:
> Henk Steentjes of Nedap,
> Jan Groenendaal of Groenendaal BV (the developers of the count software),
> John Pugh of Nathean (formerly of PMI),
> Niall Callan, secretary general of the DoEHLG, and
> Peter Green, who is in charge of the introduction of evoting within the DoEHLG

There was also a Tom Corcoran of DoEHLG sitting to Mr Callan's
right, but I'm not sure what his role is because he stayed as quiet
as Mr Greene.

> Unfortunately, the debate was a little stilted because of the way oireachtas
> committee meetings work. The members had to ask questions, which we answered.
> Everyone felt restricted by the format, I think.

Everyone except the Chairman and the other FF members.  The four
opposition committee members took time to object to it (Mr Cuffe,
for instance, called it Kafkaesque).  Some good exchanges happened
when people started answering points directly instead of waiting
to be called on by the Chairman.

> Any other thoughts from people in the observers' seats?

A few positive things emerged in the meeting, which may be useful in
media interviews as rhetorical points:-

1. Late delivery of development snapshot
    It was admitted (by Mr Groenendaal) that the plan is to deliver
  the IES software as late as possible before the election, and that
  the delivered software would in each case be the latest development
  snapshot.  No reason was offered for this, except for Mr Callan's
  comments about "continuous improvement" and "adding to margins of
  safety".  It should be obvious that having a regular software
  update schedule is a convenient cover for interfering with the
  software on the count PCs, or for introducing Trojan horses.

2. Possible infiltration of supplier
    The only really good pinning-down question was asked by Mr Gilmore,
  who extracted an admission from Mr Groenendaal that it was in fact
  possible to infiltrate his company and interfere with the counting
  software.  Mr Groenendaal of course dismissed this, saying "I'm
  careful about whom I employ" but Deputy Morgan punctured the
  credibility of that comment by remarking "So is Buckingham Palace".

  (Other observers in the room may have noticed that shortly before
  Mr Gilmore asked this question, I passed him a short list of
  suggested questions: I'd become frustrated that no really penetrating
  questions had been asked of the Government-procured experts up
  to then.  He didn't use any of my suggestions, but I didn't mind
  after he'd asked this question.)

3. Sketchy defences to VVAT
    The voter-verified audit trail was of course the dead elephant on
  the table that the Government experts pretended not to smell.  It's
  worth noting that they didn't have a coherent answer to the need for
  it.  Mr Callan asserted that VVAT is not needed because the system
  is "fully tested", and Mr Steentjes agreed, likening it to a calculator
  as something which can be fairly evaluated by black-box testing.
  (No one mentioned the Excel flight simulator.)  Mr Callan also pointed
  to a study which claimed that paper ballots have a 1.1% spoilage rate,
  and tried to imply that VVAT would have the same problem.  When it
  was pointed out (by Mr McCormack) that a VVAT would increase public
  confidence in the system, Mr Callan's only answer was that public
  confidence is the concern of the Minister.  Mr Callan also disputed
  that VVAT is a growing trend in other states considering evoting.
  Mr Groenendaal tried to claim that the ballot modules are actual,
  if electronic, ballot boxes, and that the records stored in them are
  actual, if electronic, ballots, which plainly dodged the issue.  He
  also pointed out that availability would be hurt by paper jams,
  inadvertently running into the Department's standard `availability'
  defence, which is that "there will be a substantial number of spare
  machines available in each district".  When pressed, Mr Callan even
  reverted to a dismissal of vote receipts, and appeared confused
  when Mr Gilmore pointed out that no one was asking for them.

4. Jesuitical "confusion" defence to VVAT/electronic disparity

    A common answer given by Mr Callan and Mr Steentjes was that a second
  record of the ballot would lead to "confusion" over which one was primary.
  It's understandable that a vendor would want to suppress any second
  record which might expose a flaw in its own equipment, but Mr Callan's
  endorsement of this point of view is inexplicable.

5. Aeroplane analogy made

    It was suggested that the blind trust asked of voters was no more than
  the blind trust asked of airline passengers.  Of course, the aviation
  industry has learned (the hard way) to treat its operations as a large,
  complex, critical system, with detailed backups and exhaustive training
  for all involved personnel.  Even the modern fly-by-wire Airbus planes
  have a plan for total avionics failure: the magnetic compass, rudder,
  and engine throttles are all mechanical, so the pilots can continue to
  fly safely (but not necessarily efficiently) even if everything shuts
  down.  The evoting industry hasn't reached that level of maturity yet,
  and it shows.

6. Letter of intent considered binding -- conditions of contract may be an out.

    Mr Callan made it clear that the letter of intent to buy 7000
  machines is effectively binding, but that the "main contract"
  hadn't yet been signed.  In Bobby Molloy's words, Nedap/Powervote
  was chosen "subject to successful testing and satisfactory
  contractual arrangements" (Dáil Éireann, 2001-06-28).  IANAL, but
  I believe that this means the details in the contract are still
  up for negotiation and aren't yet binding.  A requirement of VVAT
  is therefore still a possibility.

7. DBT/Choicepoint buck-passing: Groenendaal onto Dept, Dept onto suppliers

    When questioned about the just-in-time delivery of count software
  to the Department, Mr Groenendaal said that it was the Department's
  responsibility to test the software for acceptability.  When Mr
  Callan was questioned about the trustworthiness of any part of
  the system, he claimed the system was already "proven and robust"
  due to its use in the Netherlands and Germany, and that it has
  been tested by the six independent consultancies.  It's difficult
  to see how these positions can be reconciled.  It reminds me of
  the deal between Florida's elections supervisor and DBT/ChoicePoint,
  whereby DBT/ChoicePoint supplied a list of alleged felons to be
  deleted from the voting register, and claimed that the state was
  responsible for fixing errors, and the state claimed that the
  list was already corrected by the suppliers.

8. Multi-ballot procedure untested

    One of Mr McCarthy's major complaints was that the multi-ballot
  facility of the machines (that is, the ability to conduct two or
  more simultaneous PR-STV elections) has not been tested.  Incredibly,
  Mr Callan's response was that it doesn't need to be, as the
  European Parliament elections are done under the same rules as
  the Dáil elections.  This response is irrelevant to the consideration
  of whether the software will function correctly with multiple
  simultaneous elections.

9. Party-political voting exposes lack of consensus

    In the early legislative record of the Electoral (Amendment)
  Bill, 2000 (which enabled the introduction of the proposed evoting
  system), Government spokesmen placed a lot of emphasis on the
  cautious and consensus-building approach that the Government would
  take with any evoting proposal.  For example, Bobby Molloy promised
  that "this Government will not proceed without unanimity and
  general agreement among the Members here".  A proposed amendment
  which would have required specific approval for any particular
  system by the Houses of the Oireachtas was rejected by Mr Molloy,

>     At this stage, we are simply putting enabling legislation in
>     place.  With regard to the hardware, all that is proposed is
>     the purchase of six machines, on which the testing will be
>     carried out.  Everybody will have to be fully satisfied that
>     the machines and the whole system are operating properly when
>     they have been well and truly tested.  The entire procedure
>     will be open and seen to be open.  I suggest that the Senator
>     consider withdrawing the amendment.  We have a long way to go
>     before any decisions are made about putting the system in
>     place or actually using it.

    It's clear from Thursday's party-line "support the Government
  plan" vote that no pretense of consensus or openness remains.


GPG 0x43D3AD19 17D2 CA6E A18E 1177 A361  C14C 29DB BA4B 43D3 AD19
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.stdlib.net/pipermail/e-voting/attachments/20031221/d074c387/attachment.bin

More information about the E-voting mailing list