[EM] danger of coercion (Re: First U.S. Scientific Election Audit...)

[Abd ul-Rahman Lomax]

At 12:00 PM 5/18/2007, Juho wrote:
>On May 18, 2007, at 6:45 , Abd ul-Rahman Lomax wrote:
>>My own suggestion has been merely to require registration. The
>>proposal Juho makes misses the point of write-in candidacies. They
>>are for candidates who were unable to get on the ballot.
>
>Maybe it would be ok to require all the candidates to wake up already
>let's say two (?) months before the election.

Once again, Juho shows that he does not understand the purpose of 
write-in votes, nor, indeed of elections themselves! From my point of 
view, increasing the burden on candidates is exactly the opposite of 
what is needed. I'm only willing to require registration under 
circumstances where there are security concerns beyond the normal. 
Elections are a device where *the public* chooses officers or 
representatives. I actually think that elections are a totally bogus 
way of choosing representatives. You are not represented, except by 
accident, if you cannot freely choose your representative. 
Originally, subunits of government and those in their juridisdictions 
were often appointed by the sovereign. You did not get to choose who 
your representative was. The fact is that we kept this system, 
substituting only an election process, and usually a quite defective 
one, whereby the sovereign (the people) chooses your representatives 
for you. You don't get to choose! You merely get to exert a very 
small influence on the process, within very narrow constraints.

Eliminating your right to name someone at the last minute, with total 
freedom, with at least the possibility of, if enough of you do that, 
the person is elected (and could accept or decline), is eliminating 
one more hint of true democracy, the unconstrained will of the 
people. I'm not at all proposing that the "will of the people" be 
unconstrained, there are necessary constraints, but the ability to 
freely choose representatives is properly not among them.

(The fact is that the people can choose representatives if they want 
to, even if the government does not recognize them. And if enough 
people -- a majority is enough, for sure -- do this, the government 
will eventually follow or it will fall. But generally the people are 
not awake to this power, and they have no idea how it could be 
exercised. It is this that I have taken on as a project, to awaken 
the people, starting with a few, starting, indeed, with a handful, to 
how this could happen and what the benefits could be. I am sure there 
are also hazards, but, frankly, few people are even considering the 
possibility enough to identify real hazards; instead, the few 
objections we see are knee-jerk, ill-considered ones that are 
actually trivial or moot.)

Last minute changes can make a write-in campaign necessary. A 
candidate can die or become incapacitated or there is some huge 
scandal that suddenly takes the candidate off the map of political viability.

So if there *is* going to be some restriction on write-in votes, it 
should be minimal. What I proposed was minimal, the use of a 
registration number, with a booklet available at polling places for 
voters to find the number. It actually solves another problem, which 
is that if the clerk counting votes does not recognize the name, or 
if the name is not sufficiently specific, clerks have a tendency to 
lose the vote, or the votes are never actually amalgamated. That's 
unlikely to happen too much if a write-in is actually winning, but 
..., that is exactly what happened to my wife, we think, though we 
are not sure.

>Another approach would be to use ballots that do not list the names
>of the candidates but just contain space where the number of the
>candidate can be written (that's what I'm used to - adds some risk of
>handwriting recognition when compared e.g. to just ticking boxes, but
>I can use my left (weaker) hand if needed). Very much like what you
>propose below.

You can use checkboxes for the numbers. Remember, this is under 
conditions of a need for security, so handwriting would need to be 
discouraged. However, any manual marks on the ballot could be used to 
make it identifiable.

Now, here is a scheme that is short of my imaging proposal, but still 
preserves almost as much security, and which would eliminate almost 
all of the alleged risk:

A standard program is developed which can use an ordinary scan of a 
ballot to generate vote data. This program runs on commonly-available 
computers. The ballots are, again, serialized as I described before. 
There is a boot CD which is distributed. Registered observers, 
including representatives of political parties and other interest 
groups, including media, may bring and set up a computer and scanner. 
The ballots are fed through the scanners, and the software analyzes 
the images and generates vote data as a set of tables, including the 
serial number, and writes the analyzed data to simple tab-delimited 
files, again on a CD, which is closed so that additional data cannot 
be written to it. The computers may be examined, and may not have any 
apparent nonvolatile information storage devices in them other than 
the normal boot devices. They must be standard motherboards, from an 
(extensive) list. When the data CDs have been written, they are 
identified as to ownership; they may then be examined by any other 
observer to verify that the only files on them are the simple vote 
data files, there are no ballot images. The boot ROMs are reset so 
that all non-volatile information in them, other than the fixed 
information which as in them as manufactured, is erased. (That 
information is written in storage that might as well be hard-wired.)

With these reasonable and fairly easy precautions, the ability of any 
observer to obtain and use the actual images, and thus extraneous 
non=-vote information, would *mostly* be limited to what the 
observers can casually view from the limited ability they would have 
to actually see the ballots. An observer in cahoots with a coercer 
would not be able to guarantee that he had seen all the ballots, or 
even a majority of them, and therefore a lack of seeing confirming 
marks would not be telling.

Now, any seriously determined and technically competent coercer who 
was seriously interested in defeating this system, and making off 
with ballot images, could do it. But it would be expensive and would 
require resources that the bete noir of Juho, the coercive husband, 
would certainly not do it.

Further, possession of the ballot images would, in fact, under 
present law, be illegal. They would represent data theft, data that 
was explicitly protected and which could only be accessed and held by 
circumventing security measures.

What happens is that multiple observers end up with data from scans, 
tagged to the ballot serial numbers. It then becomes possible to 
compare their results. Where there are discrepancies, it's possible 
then, to arrange an examination of the sequestered ballots, if the 
differences are not moot. Systemic manipulation of the counts would 
be almost impossible.

I would prefer that there be multiple analysis programs that are 
tested against an assortment of actual ballots, with various 
difficult markings. The observers could choose which program they 
would use; all would be open-source.

Once again, by making all aspects of this open to public scrutiny, 
the vast resources of the public to detect fraud can be employed. 
Attempts to encode a back-door in the public-source programs would 
probably require infecting the compilers, something very difficult to 
do on a large scale without detection.

But the system does not have to make it *impossible* for a dedicated 
hacker to defeat, it merely needs to make it difficult enough that 
some isolated coercer can harm someone.

But my opinion remains that all this is not necessary. There never 
has been a guarantee that ballots could not be examined by the 
public, because they *are* examined by the public. It is merely very 
difficult for the public to actually do it, and the difficulty is not 
caused by concerns over voter privacy, it is cause by concern for the 
safety of the actual ballots. If you can examine them, you might 
alter them. That is the reason why you have to pay an officer to 
stand there and watch you if you want to examine the ballots. They 
*must* allow you to see them, it's just a matter of the right of the 
public (through the government) not to be put to extraordinary 
expense merely because you want to look at the ballots.

What I've done is to suggest that it be made so easy to examine the 
ballots that everyone can do it. This was not technically feasible 
(it was possible but not feasible) until quite recently. The 
conditions that allow it are, essentially, cheap imaging, with 
widespread access to imaging tools. However, it did become possible 
quite a while ago, the possibility predates computers. Faxes. 
However, until fairly recently fax images were not stored, they were 
printed immediately as transmitted, so what really made this 
inexpensive was widespread storage of fax images, thus eliminating 
the two or three cents per image cost of printing the faxes. Faxes 
*are* quite acceptable as scanners for the purpose given here. The 
basic resolution is sufficient.

So all the concern about privacy of ballots, as distinct from voters, 
is very new. If you were a friend of the clerk (or custodian), and 
you wanted to look at the ballots, you always could have done it, in 
practice, though in some cases it might have been illegal. Someone 
coercing votes is already willing to break the law, so there was no 
protection, really. It's a red herring. There is no right of privacy 
of ballots. Your ballot can be seen by others, now, period. Privacy 
of ballots would be a *new* proposal, not something already there 
which we are proposing to take away.


>>>I propose simple ballots and separate ballots for each race in
>>>addition to what you said.
>>
>>Not a snowball's chance in h e double toothpicks, as my wife's
>>father used to say.
>>
>>The public expense is relatively large. What is the evidence that
>>the cost of not doing this is greater?
>
>Simple ballots would be cheap to print.

Showing, once again, how shallow can be consideration of something 
new. It's not the *printing* that is expensive. It is collating them. 
You have to produce sets of ballots that have no duplications, so 
that every voter gets one and only one copy of each ballot. But it 
could be done, and it is not all *that* difficult. Since  there are 
some good reasons to have separate ballots -- I agree with this, that 
it's a good idea -- my expression about the expense is coming from 
this as a political obstacle. What will be pointed out if I tried to 
actually advocate this is that there is no problem that this solution 
is solving. If it ain't broke, don't fix it!

It is cheaper and simpler to give voters one ballot with many 
questions. So to motivate a change, there must be some problem being 
solved by it. Until there is a perception that there is a problem, 
there is no good reason to change!

Vote coercion is not perceived as a problem here in the U.S. There 
are two perceived problems, and which one is more important depends 
on whether you are a Republican or a Democrat, apparently.

There is voter fraud, generally registration of ineligible people, or 
voting by people who are not legally qualified to vote. The evidence 
is that this is actually, in most environments, a tiny problem. The 
whole current scandal in the Justice Department is apparently over 
serious political pressure from high Republicans to prosecute voter 
fraud, and the alleged failure of certain U.S. attorneys to file 
cases. Even in the presence of very serious pressure, being enforced 
by selective dismisal of attorneys who did not produce results, very 
few cases were actually filed. As one writer pointed out, in response 
to claims that voter fraud is an epidemic, this is one of our less 
serious epidemics, as a major effort produced very little prosecution.

The most common "voter fraud" is where a non-citizen registers and 
votes. In the large majority of cases, the cause is really that the 
voter didn't realize that it was illegal, did not notice the language 
on registrations forms warning about it. With a large population, a 
few people will make just about every mistake you can imagine. It 
does not mean that they are stupid, for everyone makes "stupid" 
mistakes once in a while. You have a  hundred million people do a 
thing, it only takes a small and rare oversight, some error in 
interpreting what is read, or a moment's inattention, to cause every 
conceivable mistake to take place. From the cases that were actually 
filed, it looks like people who made innocent mistakes had the full 
weight of the federal government come down on them....

Sometimes fraudulent registrations may have been part of a some 
campaign to scare up votes by buying them. You go and round up a 
bunch of idle people and pay them to go and register and vote. It's 
illegal, and not very effective, and the recent case that comes to 
mind happened in Florida. It was a Republican candidate who, as far 
as we can tell, was unaware of the activities of the person who 
actually did this. It is highly unlikely that the election turned on 
the actual votes involved. However, because the first election did 
not result in a majority winner, there was a runoff, which was run by 
the Republican, with no allegations of voter fraud. The court, 
however, decided that because it was impossible to tell how many 
votes were involved, they would discard *all* absentee ballots -- all 
the alleged bought ballots were cast absentee --, which was enough to 
give the Democrat the victory in the *first* election, hence the 
overall result switched. To me, this was an example of a court 
ignoring the purpose of elections in favor of a technical 
construction of the process. And I'm a Democrat. The Republican 
involved was not necessarily one of my favorite people, he was 
involved in the Florida 2000 fiasco. But he, and the voters, were 
apparently robbed by clever legal action....

The other kind of fraud is very real and is a serious problem, and it 
is election fraud. There is overt fraud involving deliberate 
manipulation of the counts, clearly illegal if it could be discovered 
and proven. However, the only evidence we have regarding it is 
statistical. Statistically, there *has* been manipulation. But 
statistical evidence does not prove who did it, and statistical 
evidence has, so far, not been admitted as evidence to turn election 
results, even where statistical analysis can produce effective 
certainty that an official result was incorrect. The law requires 
actual vote counts.

The other kind of election fraud is more subtle. If you are a voting 
officer, and somehow you fail to provide adequate equipment to 
districts, and, surely by coincidence, it is Democratic districts 
which don't get proper equipment, it is very difficult to prove 
intention, and even if you can prove intention, it is not necessarily 
a violation of law, or at least not of criminal law. Can a state 
officer selectively perform his or her duties?

Such discrimination is not criminal, it is only grounds for civil 
prosecution, if I'm correct.

Anyway, the *big* problem is not voter fraud, nor is it vote 
coercion, nor is it vote buying, it is fraud in how the election is 
conducted and how the ballots are handled and counted. With the 
suggestion of paper ballots and public imaging, I'm directly 
addressing two aspects of the problem, definitively and simply:

Paper ballots are minimal cost, and, in fact, if you consider 
amortization of equipment, cheaper to count than through using 
complex voting machines. There should be no problems with 
distribution and maintenance of equipment. If the pencils aren't 
there, someone can go down to the local convenience market and buy 
some pencils! The only thing to be distributed is paper ballots. It's 
also possible to design ballots so that anyone could print them, but 
this causes other security problems, so, at the present, I'd prefer 
ballots to be not easily printed at home.

And official counting of the ballots becomes faster and easier if 
images are used rather than the ballots themselves. Observers do not 
need to watch every move, since alteration of the images is moot, it 
*will* be detected. Further, anyone can count the votes, and 
collectively, the public can count the votes with multiple 
independent redundancy. Any problems, such as ballot ambiguities, 
will be easily identified and will be quantifiable. Needless recounts 
will be avoided, reducing expense both for the public and for 
candidates who suspect that something is wrong with a close election. 
And recounts that are truly necessary can focus on the ambiguous 
ballots that are the only reasonable cause, under this system, for a 
recount (besides image alteration which will be easily detectable).


>>Oh, dear. But vote coercion is already a serious crime. Why would
>>such a rule make any difference at all?
>
>Keeping the system healthy is easier and cheaper that healing it
>after it has become seriously ill. Light rules may be enough.

The *present* rules are enough! If you are not ill, you don't start 
taking medicine to stop becoming ill unless there is specific reason 
to think you are exposed to a need for it. If you do, you are quite 
likely to do more harm than good.

Do you want to have to search voters when they enter the polling 
place? In certain places, that's necessary, but not to prevent 
cameras. It is to prevent weapons from coming in, as well as 
explosives. We do not have this problem here, particularly. I don't 
recall going through serious security at any polling place. It is 
more difficult, by far, to take a plane.

>>>Bad design. I have never voted in an election with several races on
>>>one sheet. (Usually there has been only one race per election, but
>>>when there have been more the ballots have been separate.)
>>
>>Really? Where do you live? What elections have you participated in?
>
>Finland. Mostly local, national and EU level PR elections and
>national single winner elections.

Single ballot per race makes certain aspects of counting easier. As I 
mentioned above, the printing problem can be solved. You print a 
booklet, essentially, then there is a paper cutter at the polling 
place. You pull out one booklet and cut the binding, which causes the 
booklet to become a pile of sheets with one ballot per race. The 
binding is only a staple or two in the folded edge, standard 
side-staple binding, cheap and fast, the cheapest form of binding, 
actually. If there are two pages that accidentally got folded 
together, the whole booklet will be *very* different to handle, and 
it can be detected and destroyed by the election workers, or before 
(it would stand out like a sore thumb in a stack of ballots viewed 
from a certain direction, it's twice as thick as a proper set.)

>>>Sounds a bit dangerous from privacy point of view. The next step
>>>could be to include the voter's profession, age etc. Better be
>>>careful with these.
>>
>>Juho, nothing was suggested about collecting private information.
>
>Ok, my words exaggerated a bit, but the point is that risk of losing
>privacy grows when larger amounts of person related data is linked
>together.

But votes are not personally related data. They are *generated* by a 
person, but in a very narrow context, and, normally, there is nothing 
about a vote pattern that would reveal who a person is, unless the 
person has some very rare combination of positions *and you already 
know what they are*.

This is so rare a combination of two circumstances that it is likely 
that it simply does not occur in reality. People do not ordinarily 
look for things that are this rare. Nobody is going to be pouring 
over ballots to see if they can figure out how their neighbor votes. 
What would they be looking for? What they already know? Why would 
they need to look at the ballots if they already know it?

If it is a crime to be a Republican, then you might as well shoot 
him, you don't need the ballots. After all, how would you know that 
it was his ballot?

But, of course, this would be exceedingly stupid, which is why it 
does not happen, even though people do exceedingly stupid things 
quite frequently. It makes no sense to shoot people simply to prevent 
them from voting, because to have any effect on outcomes, you'd have 
to shoot quite a few. And people dislike being shot, and, in fact, 
they dislike their neighbors being shot almost as much. Even if they 
are Republicans.

(There are places, of course, where people can indeed be shot because 
they are suspected of holding unpopular political positions. But it 
does not happen here. What *does* happen here that might be similar 
in some respects has nothing to do with voting. It has to do with 
selective prosecution, harassment, or deportation, an entirely 
different matter, and it is not common.)