[E-voting] Current state of evoting in Europe

David Glaude (PourEVA) dglaude at poureva.be
Fri Dec 16 23:27:26 GMT 2016


It is indeed difficult...

Here is for a quick status update about Belgium.

A lot of stuff can be found in French on http://www.poureva.be/ but even
there, we have no summary or recap.
I will try to translate in English presentation I did recently:
https://docs.google.com/presentation/d/1OaV7TY_VcSBTsUysp05LTDBAk3QSi6_arj4nA3EA12w/edit?usp=sharing

Belgium is a Federal State made of 3 Regions (geographical region) and 3
Communities (language). Yes, this is in the very begining of our
constitution.

Local elections are organised by the Region (with the exception of local
election in the tiny german speaking community that organise it's own local
election despite beeing fully included in Walonia, one of the 3 regions).

Other elections (Federal, Regional and European) are organised by the
federal authority.

The reason it is important is that potentially (and in practice) the rules
or equipement used might be different depending on wich Region and/or
Communities you are in, and what level of election (and hopefully Local
election = every 6 years are almost never simultaneous as
Federal+Regional+European = every 5 years).

Last election were Federal+Regional+European in 2014
Next election are local in 2018.
Next Next election are Federal+Regional+European in 2019 (but if the
government fail, then early Federal election could take place anytime
between now and 2019)

So how did we vote in 2014:

1) In Wallonia (mostly french speaking region on the south) all voted on
PAPER except for less that 40 localities that used very old electronic
without paper trail and that store the vote on a magnetic card. It is to be
noted that 100% of the localities in of the German community voted using
that system.
In some place (mosly Luxembourg province) while the paper voting was not
modify (not even the law), computer assisted counting (e-counting = DEPASS
= Dépouillement Assisté par Ordinateur) has been used for the counting. In
short, two team are "encoding" each and every vote in the same order. If
the second team encoding does not match the first team encoding, then the
computer ask to resolve the inconsistency.

2) In Flanders (mostly dutch/flemish speaking region on the south) 50% of
the localities were voting on paper where 50% were voting on new voting
machine from Smartmatic adapted to more or less follow the recommendation
of the BeVoting report made by 7 (= all) Belgian university. This is a VVAT
enabled voting system with an hybride printed parper voting ballot.
To sumarize, the voter receive a smart card to enable him to vote, then he
goes into the voting booth and make the vote selection on a touch screen
voting machine, the machine print a paper ballot that contain two major
part, (1) the human readable version of the vote (2) a 2D barre code
(likely QR code) supposed to contain the vote in computer readable format.
The voter verify the ticket, hide it's content, goes to the voting box,
scan his QR code, then put the ticket/paper ballot in the ballot box.
No one else but the voter does seen that paper ballot, there is not risk
limiting audit, no provision to do a human recount, ... at most, a judge
could order a recount if some loosing candidate can convince him of a fraud
that would require such a recount... however the law seems to say that the
QR code is the valid recording of the voter intent.

3) In Brussels Region (central bi-lingual region) 2 localities were using
the new Smartmatic/BeVoting system and 17 where using the old magnetic card
system.

The election was organised by the federal state. The same localities were
voting with the same hardware as in 2012 for the election (local election
organised by the 3 separates region). However, the source code was
modified. On the Smartmatic system, it was to correct glitch that occured
in 2012 and for JITES the magnetic card system, it was adapted to take into
account multiple simultaneous election and other difference in the user
interface. Apparently, the source code of the election was a bit of a
mix/merge between previous version (2012) and previous simultaneous
election organised by the federal authority.

All software are audited by a company (PwC branch) to check it if works OK
and respect the law. The audit company is paid by the voting system vendor
(the one writing the software). Based on that report, the minister of
interior validate the use of that software for the election. The source
code is made available to a team of official expert (that have no time to
analyse it) and to one representative of each party that request it. After
the election, that same source code is made public on the official web
site. And this is true for both JITES (magnetic card) and Smartmatic (QR
code).

Something went wrong during the election day... and that was detected
during the election night. On the JITES system (the one with magnetic card
only recording). Some vote (more than 2000) were oddly encoded on the
magnetic card (the content of the magnetic card is recorded and encrypted
on a floppy disk) and the counting software (that aggregate the floppy disk
result from each voting area) detected a "problem". Apparently those vote
were having a party number selected, but no candidate selected, not even
the list check box. The ministry of interior first blocked the publication
of result from those voting place, then assumed that they were just missing
the preferential vote, but the party vote was OK, so they started to
release result based on that. After further analysis (fully described list
of event exist in the official expert report) the party vote was also
wrong!!! This was not the validated voter intent, but a previous choice
(visit) to a party list that the voter did not confirm and where the voter
changed his mind and selected another list... there are two ways to do that
in the user interface, one of those was buggy and was not zeroing the
previous selection. Depending if the final choice of the voter was a party
with a lower/greater list number, the bug would be triggered.

So for 2000+ votes, the real voter intent was lost (not recorded) with no
way to know what he wanted. It took a few days to undestand the bug and
recompute result of the election by not counting (disfrenchise) those 2000+
votes. In fact there were bugs in trying to correct the bug and re-decrypt
the floppy disk content and the final result would only be known weeks
later, the day/night before election validation.

The blame was given to the old hardware that still use floppy (unreliable)
and magnetic card. So the political message was that we should stop using
those old machine and start using the new machine... in reality, the
software did exactly what it was program to do. There was no "hardware" bug
or issue to read the floppy (as originally stated) but just a 2 lines bugs,
not detected by the vendor test, not detected by the auditor nor anybody
else that had access to the code/hardware before the election.

This story made the news for a week at least. Judge (or president of voting
place) were asked to validate result that were wrong without much
explanation. The ministry of interior or the IT specialist declared what
was supposed to be the result (with errors).

What went even more wrong is that those 2000+ votes cancelled could have
change witch candidate were to be elected (because of low marging between
candidate in Brussels election) and in German Speaking Community, one party
lost a seat by 17 votes... where 21 votes were cancelled and 17 of those
could have been for that party. Of course the result of the election was
contested by that party (and other looser) but since the elected candidate
are also the one validating their own election... it was judged unlikely
that the result would have change.

So how are we going to vote in 2018 (to the best of my knowledge based on
trends and political positioning).

1) In Wallonia, the decision was taken to get rid of JITES (the magnetic
card system) and to vote on PAPER only.
However the German Community asked to be able to continue electronic voting
with the new Smartmatic/BeVoting system. So Wallonia decided to give them
the right/power to organise local election. If the German Community find
the money and get a acquisition path to have the few Smartmatic machine it
require for the election, they will do it. They are currently
writing/voting their legislation to organise local election.
Finally, it is very likely that once again electronic counting will be use
in some places, without any legislation to give a few minimum guarantee on
the proper functionning of that system. We are lobbing the Wallonia
government to make sure they either BAN that eCounting system or the put a
legislation in place.

2) Flanders will likely continue to vote as they did in 2012 and 2014 with
Smartmatic/Bevoting. If they have money, they might tright to increase the
number of localities that do not vote on paper. It will likely be a budget
issue, but if they had the money, everybody would vote on computer.

3) Brussels Region made debate on electronic voting and decided to go for
Full Smartmatic/BeVoting system (so 19 localities out of 19 using that
system). However they changed a bit the law for local election to add some
safeguard such as mandatory counting of minimum one voting place per
locality and bigger printing of the party number (the only part they plan
to count). Because Brussels Region does not have the money to acquire all
the required voting machine, they plan to rent them (leasing plan). The
detail are still unclear, but that is officially their plan.

It is likely that in 2019, the vote will be organised by the federal using
the same machine/system as in 2018.

One big uncertainty is if we have early election (when the goverment fail).
In that case, it is unlikely that Brussels Region can be ready with
Smartmatic system, and the old system as been send to the recycling
already... so maybe it would be paper.

If you follow my Google Slide presentation, you will see maps of what is
used where (or will be used where).

We at PourEVA are unhappy with the waist of money in acquiring (or renting)
Smartmatic/BeVoting.
We are also unhappy about the lack of citizen audit (aka Risk Limiting
Audit) into that system.
We believe that eCounting could be acceptable if proper law is put in place
to have a mandatory counting to validate the absence of bug or fraud.

I believe this sumarize the recent past and announce the near future.

I am available to answer questions if needed.

Regards.

David Glaude



















2016-12-16 18:48 GMT+01:00 Margaret McGaley <mmcgaley at gmail.com>:

> Hi all,
>
> I'm writing a thing, and I'm finding it hard to get up-to-date
> information on e-voting in Europe. I'm particularly interested in the
> current status of:
>
> Belgium
> Germany
> The Netherlands
> Estonia
> Switzerland
>
> Does anyone still subscribe to this list? And if so, does anyone know
> the latest on these (or any other) countries?
>
> Thanks,
> Margaret
>
> _______________________________________________
> E-voting mailing list
> E-voting at lists.stdlib.net
> http://lists.stdlib.net/mailman/listinfo/e-voting
> http://evoting.cs.may.ie/
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.stdlib.net/pipermail/e-voting/attachments/20161217/e44be3cf/attachment-0001.htm 


More information about the E-voting mailing list