Possible use of Python for a voting machine demo project -- your feedback requested

P

Paul Rubin

Alan Dechert said:
Our project incorporates a proposal for Remote Attended Internet Voting to
replace the various existing absentee voting methods employed today.

I wouldn't want to use the public internet that way. It sounds like
an invitation to launch DOS attacks against the parts of the network
where one's political opponents live. I don't see the need for any
network connection as long as the election info can be delivered to
all the polling places before the election starts. If every election
can be enrolled in an FEC database a few weeks before election day
(that means the database has all the info that would get printed on a
ballot), then the whole database can get dumped to CD-ROM or DVD-ROM
and shipped to all the polling places in time for the election, no
internet needed.
 
A

Alan Dechert

Paul Rubin said:
I wouldn't want to use the public internet that way. It sounds like
an invitation to launch DOS attacks against the parts of the network
where one's political opponents live. I don't see the need for any
network connection as long as the election info can be delivered to
all the polling places before the election starts. If every election
can be enrolled in an FEC database a few weeks before election day
(that means the database has all the info that would get printed on a
ballot), then the whole database can get dumped to CD-ROM or DVD-ROM
and shipped to all the polling places in time for the election, no
internet needed.
That's a thought. It might be feasible to aggregate all the databases from
all the counties for all the contests (although these databases might be
larger than you think -- especially when you have all the audio files in all
the different languages). However, the main problem I see is the voter
files. You'd also need all the voter files from all the states and some of
these get updated too close to Election Day. In other words, with remote
absentee voting -- with no pre-printed roster like you have at the precinct
polling places -- we need to ID the voter and verify registration (including
the precinct in which registered). HAVA calls for statewide databases and
this should help a lot for cleaning up these files. Remote poll workers
should be able to access the voter file online to verify registration.

Several studies conducted so far regarding Internet voting have turned
thumbs down on unattended voting but have concluded that attended Internet
voting should be feasible. It's already pretty much a given that it will be
available for overseas military (perhaps as soon as 2004).

Alan Dechert
 
A

Andrew Dalke

Alan Dechert:
Point (or click on) again to de-select.

Agreeing with Ian Bicking in his followup, there's no need to grey out the
unselected fields, just emphasize the selected one.
As a matter of fact, we won't let you vote for O'Reilly.

He-he, I was thinking of O'Reilly as the book publisher, forgetting there's
another one more closely involved with politics.

I brought it up because I remember on our old Plato system (last 1980s),
the Plato admin for the department was a "O'Something" and rewrote
some code which didn't allow him to use an apostrophe for his name.
Okay we'll have an apostrophe available. Anything else?

I don't think there's need for accents, umlauts, tildes, and other such
marks,
even if it does mean leaving it out is technically a misspelling.
Mercuri (Mercuri-Neumann, more accurately), suggests the paper ballot be
inaccessible to the voter -- viewable behind glass. This involves some
expensive and proprietary hardware since paper handling must also deal with
rejected printouts.

Huh. Well, like I said, I know just enough to be dangerous. I like your
method instead.
My scheme is cheaper and lower tech. It allows the voter to handle the
ballot. This involves a minor security issue (then again, since when have
we decided we can't trust voters to touch their ballots?).

Agreed. We're trusting people to make a vote, so the little bit of extra
trust needed to handle a ballot seems appropriate.
It is a real life problem. We've given a lot of thought to this issue. The
printout will be designed so that counterfeits can be detected easily.

Cool! I'm feeling all warm and fuzzy about your work now. :)
Isn't that overkill? I seem to recall that already there are provisions
for people with special needs to have someone in the booth to help.
I don't think it's overkill. One of the current [lame] arguments against a
"voter-verified paper trail" is that "Mandating Voter-Verified Paper Trails
Could Deny Voters With Disabilities the Right to Cast a Secret Ballot."

Indeed, and you're right. Objection withdrawn.
Okay, thanks for your input.

BTW, another possibility for a demo is to use Flash. I've never used
it, but I hear it has some sort of authoring environment and it's pretty
popular and documentation about it is widely available. It might be
best to start with this for a demo.

Andrew
(e-mail address removed)
 
P

Paul Rubin

Alan Dechert said:
We'll see. The touch screen we intend to use works well with a stylus. So
if we get too many mistakes using fingers, we may just have people using a
stylus exclusively.

Try to avoid that. Use a real big screen if needed (Wacom makes some
nice ones).
I'll take this as a prediction, not necessarily correct, however. Our team
includes some people with extensive experience with voting machine
evaluation -- they think it will work. But again, we won't know for sure
until we try. But beyond that, most voting machine PCs we are proposing to
use will be mouse driven. So even if it proves to be too dense for a stylus
(very unlikely, imo) it is certainly not too dense for a mouse. Virtually
all of the testers using the web based version will be using a mouse.

I think a production system shouldn't use mice.
On the other hand, a lot of people really really like the touch screens. We
can't make them all mouse driven since a percentage of the voters will have
a big problem with that. But there is no reason to give up on mouse driven
systems just because some people can't use them. Mice are very cheap and
most people are used to them.

What do you mean by "most people are used to them"? Do you really
mean "most computer users are used to them"? I remember hearing that
most (i.e. more than half) of the people in the world have never even
made a phone call, much less used a computer. How that maps onto US
voters, I don't know.

I hope you will do user testing with people from all backgrounds and
social strata. Go to some senior citizen centers; get a few street
people ("will work for food") to try out the system; etc.
I appreciate your taking the time to write.

Do you talk anywhere about auditibility, i.e. how does anyone verify
that your published source code is the same code actually running on
the machines?
 
M

Marc Wilson

In comp.lang.python, "Andrew Dalke" <[email protected]> (Andrew Dalke)
wrote in <[email protected]>::

|> My scheme is cheaper and lower tech. It allows the voter to handle the
|> ballot. This involves a minor security issue (then again, since when have
|> we decided we can't trust voters to touch their ballots?).
|
|Agreed. We're trusting people to make a vote, so the little bit of extra
|trust needed to handle a ballot seems appropriate.

FWIW, in the UK, the ballots are normally done by the lowest-tech method:
you're given a ballot, and a pencil, and you put an "X" in a box next to the
one you like. We don't have problems with chads, etc.

We've also started, in some areas, voting by phone or internet. Very simple
system- you get a polling card with a number and a PIN, you go to a HTTPS
website, log in and vote. Has to be done *before* the "manual" ballot
opens, so the returning officer knows who's already voted.
--
Marc Wilson

Cleopatra Consultants Limited - IT Consultants
2 The Grange, Cricklade Street, Old Town, Swindon SN1 3HG
Tel: (44/0) 70-500-15051 Fax: (44/0) 870 164-0054
Mail: (e-mail address removed) Web: http://www.cleopatra.co.uk
_________________________________________________________________
Try MailTraq at https://my.mailtraq.com/register.asp?code=cleopatra
 
P

Paul Rubin

Marc Wilson said:
FWIW, in the UK, the ballots are normally done by the lowest-tech method:
you're given a ballot, and a pencil, and you put an "X" in a box next to the
one you like. We don't have problems with chads, etc.

In the US you have the inalienable right to vote for anyone you want
for office, including for yourself, your girlfriend, your
mother-in-law, or whatever. So for example, the optical scan ballots
for the presidential election in Volusia County, Florida in 2000
looked something like this:

( ) Bush
( ) Gore
( ) Buchanan
( ) Nader
( ) Other (write in): ___________________________

If you wanted to vote for your mother-in-law for President, you'd
check "Other" and write her name there. If enough other people did
that everywhere in the US, hey, your mother-in-law could become
president! (Don't hold your breath though).

Trouble starts when you ask what happens if more than one box is checked.
If you check both "Bush" and "Nader", there's no way to tell what candidate
you wanted, so your ballot is declared invalid. Fine. If you check
"Bush" and also check "Other" and write your mother-in-law's name in
the blank, same thing. So the company that programmed the scanning
machines thought this issue easy to deal with: if more than one box
is checked, the ballot is labeled as an "overvote" and is invalid.

But what if you check "Bush" and also check "Other", and in the blank
you write "Bush"? Then there's no ambiguity, it's clear who you want,
so it's a legal vote and it must be counted (and Florida law requires
that it be counted, though establishing that takes some careful
reading). But the machines make no attempt to read the write-in
blank. They just don't count those votes despite what the law
requires.

And who would be dumb enough to mark their ballot like that? It turns
out that 488 people in Volusia County alone (some for Bush and some
for Gore), and more in other counties, did so. Whether this happened
because of some confusion in the printed voting instructions, or bad
instructions given verbally to voters, or some other reason, is
unknown. It was not discovered until after the state totals had
already been reported, because the lawyers busy fighting over the
ballot counts in other parts of the state didn't realize that so many
such "overvoted" legal ballots existed and didn't go looking for them.
Had they figured it out and noticed, that could have forced deeper
examination of a number of other results in the state, and a different
guy might have ended up in the White House than the guy who's there now.

There are a lot of not-so-obvious possibilities that any counting
scheme has to take into account. Getting anything wrong can have
potentially far-reaching consequences.
We've also started, in some areas, voting by phone or internet.
Very simple system- you get a polling card with a number and a PIN,
you go to a HTTPS website, log in and vote. Has to be done *before*
the "manual" ballot opens, so the returning officer knows who's
already voted.

That doesn't sound like a great idea. Suppose you cast that vote
a week before the normal election day. Then two days before election
day, it comes out that the leading candidate committed a number of
murders that got covered up til then. You voted with less information
than the other voters had. In fact, that candidate, fearful that
the news might come out at any moment, may have encouraged all his
supporters to vote early. For a fair election, all voting should be
done on the same day. This is another reason mail-in absentee voting
should be curtailed.
 
M

Marc Wilson

In comp.lang.python, Paul Rubin <http://[email protected]> (Paul Rubin
<[email protected]>::

|> FWIW, in the UK, the ballots are normally done by the lowest-tech method:
|> you're given a ballot, and a pencil, and you put an "X" in a box next to the
|> one you like. We don't have problems with chads, etc.
|
|In the US you have the inalienable right to vote for anyone you want
|for office, including for yourself, your girlfriend, your
|mother-in-law, or whatever. So for example, the optical scan ballots
|for the presidential election in Volusia County, Florida in 2000
|looked something like this:
|
| ( ) Bush
| ( ) Gore
| ( ) Buchanan
| ( ) Nader
| ( ) Other (write in): ___________________________
|
|If you wanted to vote for your mother-in-law for President, you'd
|check "Other" and write her name there. If enough other people did
|that everywhere in the US, hey, your mother-in-law could become
|president! (Don't hold your breath though).

Ah. That's a whole load of trouble we don't have. To stand for election,
here, you have to pony up a deposit, which you lose if you don't get a
certain percentage of the vote. It's to discourage "frivolous" campaigns,
supposedly. IIRC, it's around GBP 1000; about USD 1500.
--
Marc Wilson

Cleopatra Consultants Limited - IT Consultants
2 The Grange, Cricklade Street, Old Town, Swindon SN1 3HG
Tel: (44/0) 70-500-15051 Fax: (44/0) 870 164-0054
Mail: (e-mail address removed) Web: http://www.cleopatra.co.uk
_________________________________________________________________
Try MailTraq at https://my.mailtraq.com/register.asp?code=cleopatra
 
A

Alan Dechert

Paul Rubin said:
Here, you have to get a certain number of voter signatures on
petitions, plus pay a bunch of fees, to get your name actually printed
on the ballot.
I'm not sure if this is "uniformly" true. I once ran a petition drive to
get a candidate on the ballot IN LIEU of the filing fee (for 1986 primary,
U.S. House of Representatives, 12 CD in CA). It took 3,000 valid
signatures, IIRC.
But when voting, you can write in the name of anyone
you want.
This too seems not uniformly true. In some states (e.g., Florida) write-ins
are only valid if the candidates are "qualified." Write-ins have to file a
petition or something like that. Otherwise, the write-ins don't count. So,
if a contest has no qualified write-in candidates, the line for write-in
does not appear on the ballot. Doug Jones has some on his web site. Here's
one for Clay county in FL. Note that some contests have the write-in line
but some don't.

http://www.cs.uiowa.edu/~jones/voting/intent/samples/clay.pdf

The way that write-ins are handled would probably change [for the better] if
our uniform PC-based-open-source-with-a-printer voting system gets
implemented. Write-ins pose significant overhead in election
administration. In some cases, the actual name written in only gets read if
there are enough write-in votes to impact the outcome. In other cases,
write-ins have to be tallied by name regardless. With our system, write-ins
would be taken care of (almost) automatically. So, a lot of the rules that
are designed to cut down on the manual labor involved in tallying write-ins
would no longer be needed. There would still be some issues but, for the
most part, these will be easy to deal with. Spelling variations cause some
challenge. For example,

B WRIGHT
BILL WRIGHT
WILLIAM WRIGHT
WILL WRIGHT
WILIAM WRIGHT
WILLIAM WRITE

might all refer to the same peron. In the extremely rare instance where
sorting this out could impact the outcome, it will be much easier to deal
with on our system than on any system where votes are written in by hand.
Existing DREs on the market also have this advantage over other systems but
they don't have the penetration we hope to achieve.

Alan Dechert
 
M

Marc Wilson

In comp.lang.python, Paul Rubin <http://[email protected]> (Paul Rubin
<[email protected]>::

|> Ah. That's a whole load of trouble we don't have. To stand for election,
|> here, you have to pony up a deposit, which you lose if you don't get a
|> certain percentage of the vote. It's to discourage "frivolous" campaigns,
|> supposedly. IIRC, it's around GBP 1000; about USD 1500.
|
|Here, you have to get a certain number of voter signatures on
|petitions, plus pay a bunch of fees, to get your name actually printed
|on the ballot. But when voting, you can write in the name of anyone
|you want.
|
|Senator Strom Thurmond, who died recently at age 100, was originally
|elected to the US Senate through a write-in campaign in 1954. He went
|on to serve in the Senate for longer than anyone in history. Today,
|of course, someone winning a major office like a US Senate seat by
|write-in would be practically unthinkable. But occasionally a
|write-in candidate wins a local office, or affects the outcome of a
|local election. Tom Ammiano ran for mayor of San Francisco as a
|write-in candidate a couple of years ago and got 25 percent of the
|vote (apparently getting over 70% in his core neighborhoods), coming
|in second in a race with three non-write-in candidates listed on the
|ballot. That was enough to get Ammiano into a run-off election (which
|he lost with about 40%, but he was considered to have serious chances
|of winning). Ammiano spent about $25K on his write-in campaign while
|the guy who came in first spent over $3 million.
|
|So write-in campaigns are not necessarily "frivolous".

I mean "frivolous" in the sense of someone standing for election "for a
laugh", as used to happen when the deposit was (IIRC) GBP 25. There are
plenty of "fringe" parties and demented individuals about, and it wasn't
unknown to have 20-odd[1] names on a ballot.

[1] Some *very* odd.
--
Marc Wilson

Cleopatra Consultants Limited - IT Consultants
2 The Grange, Cricklade Street, Old Town, Swindon SN1 3HG
Tel: (44/0) 70-500-15051 Fax: (44/0) 870 164-0054
Mail: (e-mail address removed) Web: http://www.cleopatra.co.uk
_________________________________________________________________
Try MailTraq at https://my.mailtraq.com/register.asp?code=cleopatra
 
J

JanC

Alan Dechert said:
This ballot has 45 candidates in 10 contests

We often have more than 150 candidates for 1 contest here...
I guess that's why we only have 1-3 contests/ballots on the same day. ;-)
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Members online

Forum statistics

Threads
474,079
Messages
2,570,574
Members
47,207
Latest member
HelenaCani

Latest Threads

Top