From: Alan Dechert <adechert_at_earthlink_dot_net>
Date: Sun Aug 24 2003 - 14:31:16 CDT

Thanks, Matt, for your comments. Your participation in this project is
greatly appreciated.

> There are only a few things that I want to address:
> (1) Will we actually tally the votes in the end? ...
The end of what?

I see a couple of scenes for the demo. One will be a press conference.
We'll have the PCs set up in a room with some tables and chairs (think
handouts sitting on tables). We'll discuss what we did and then let people
(including reporters) try it out and ask questions. They will ask a lot of
questions (like programmatic ballot construction) to which we will answer,
"that is something we thought a lot about but was beyond the scope of work
for the demo. We have applied for an NSF grant to work on full development
of this voting system, which will be conducted at some of the top
universities in the USA. We are looking for other sources of funding., yada
yada." There will be a lot of questions where we make very good points
about why our system is great.

It makes no sense to try to schedule an "end" to the press conference where
we will tally the votes for everyone. Reporters will stick around long
enough to get whatever information they think they need. In some cases,
they'll just grab the handouts, take a look around and split.

Another scenario for the demo. Say we set up the PCs (one with touch
screen, one with mouse, a laptop w/headphones for the visually impaired, a
laptop with the scanner, a third laptop for the tabulation demo) where
passersby can try it out. I will try to set this up in the Capital Building
here in Sacramento CA. Someone (maybe more than one someone) will attend
the set up to assist users and answer questions. In this case, again, there
is no clear "end" to the demo -- it goes on for as many hours, days, weeks,
etc we can manage.

Likewise with the Internet version -- no end of the voting; just people
trying it out.

Also keep in mind we are demo-ing voting at a precinct. Tabulation is
important at the precinct level but it's a small task compared to
aggregating votes through out the county where there may be, for example in
Sacramento County) 250 ballot styles (different races different districts)
at 900 polling places. We can only demo this type of vote aggregation with
dummy data. Furthermore, the ranked preference tally would be meaningless
at the precinct level. We might produce a chart that says how many first
place, second place, etc votes each candidate received from this precinct --
but it would not be meaningful to run ranked preference vote scoring
routines on precinct level data.

When people ask about tabulation, we will explain how the votes will be
written to CD when the polls close. We should have a disk with dummy data
on it that we can pop into a laptop and show them what it looks like.
Additionally, we will have dummy data for demonstrating various aspects of
vote aggregation and tabulation.

>Has anyone thought about
> how to do this? This may be fairly trivial, but it may be the most
> important part!

This part of the demo (item 20) was recommended by Arnie Urken. Professor
Urken will likely have more to say about this very soon. Arnie has all
sorts of tabulation routines written in C and I hope that he will take over
this part of the demo. We need to give him a file format to work with, then
he can build the tabulation demo around that.

You can expect to hear more about this from Professor Urken soon.

> (2) From below:
> Selecting "Write-in" will bring up an on-screen typewriter. A QWERTY
> will appear on the top half of the screen (three rows plus space bar
> Under the space bar, we'll have two (or three rows) of letters arranged
> alphabetically.
> Why will there be both a QWERTY layout and an alphabetical layout?
This was debated some time ago between several of us. Some suggested giving
a choice. I strongly oppose giving a choice for two reasons:

1) It's involves an unnecessary step for the voter (FEC guidlines say
minimize steps!)
2) Some voters will not know what to do. They will have no idea what you
are talking about if you ask them, "Do you prefer a QWERTY layout or
alphabetical layout."

We should give them both because a lot of people don't know how to type and
will have trouble with QWERTY.

Give them both with instructions to pick the letters from either keyboard.
We are providing large monitors in the demo so there should be no space

> (3) Using commodity printers: I have nightmares about printer jams,
> etc. I don't see any way around it, but it's just something to be aware
I have a lot of experience with printers too. I'm not too concerned about
this. In the early 90s, I worked as PC support at a major corporation (for 3
years). Most of the support calls began, "I can't print."

This is not really a very demanding application for a printer -- should be
no problem although we want to make sure we have some reliable printers for
the demo.

David suggested blowing some extra $ to make sure we get reliable printing.
I think we should spend as little as we need while testing the printers to
make sure they are reliable. I don't see printer age as critical. I have
an HP LaserJet 4 sitting right here that I bought brand new 10 years ago.
It works perfectly. All I have ever done in the way of maintenance is
change toner. This printer has never had a single paper jam with normal
printing in 10 years! The only times I ever had some problem had to do with
unusual paper stock or with duplex printing.

I get a little bit of paper curl with the LJ 4 so re-inserting the paper for
duplex printing is sometimes a problem. But, printing the ballot should be
very straight forward.

I also mentioned maybe using an HP 5L for the demo. David said he had some
problem with his 6L, but I bought a 5L on eBay for $75 dollars that I needed
for a project several years ago. It performed flawlessly for over almost 2
years at which time I re-sold it on eBay for $50. The 5L is very compact
and light. I would lend my LJ 4 for the demo but I think it's a little too
heavy and bulky (actually, it might work well for the demo since it delivers
the printout horizontal and face-down. I also have an HP DeskJet 820 CXI
(color inkjet) that I bought 6 years ago. It also works perfectly -- only
maintenance = changing cartridges (wife and kids use this printer a lot).

As you can see, I like HP printers and I'll probably try to use some with
the demo. However, I'm open to other possibilities. Some of this depends
on space considerations and how the printout comes out of the printer (face
down is preferable for obvious reasons).

> (4) Security. I'm no security expert, but I think we need to assume that
> the system is hackable. Maybe it's beyond the scope of this project, but
> perhaps we should consider defining how to detect hacks.
Yes. Some of it is certainly beyond the scope of the demo, but we will need
answers. People will ask questions and we need to have some good answers.
The Dev Lead for the production system will likely be a world-class expert
on security issues. I had Bennet Yee of UCSD lined up for this but he
recently took an industrial postion and will be leaving UCSD "soon."

So, I guess we'll have to get someone else. BTW, if Bennet was picked up by
MSFT I'm gonna be pissed. I suspect MSFT but Bennet wouldn't say. I guess
we'll know soon enough.

> In addition, Alan has stated that there should be clear boundaries between
> QA and Dev. If it's acceptable to the group, I'll withdraw from any
> development to work exclusively on QA. Does anyone have any issue with
I don't know that you should work exclusively on QA. I see no problem
contributing to development whatever relevant ideas you have. Right now we
are in transition with no Dev Lead so I encourage you to work with David and
other developers as much as possible right now. When we get a new Dev Lead
then we might want to re-think our roles a bit.

Alan Dechert

= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
Received on Sun Aug 31 23:17:15 2003

This archive was generated by hypermail 2.1.8 : Sun Aug 31 2003 - 23:17:18 CDT