Q: Chris, in a nutshell, why digital paper for voting? 

A: I think if you have been following voting systems, especially since the passage of HAVA (and frankly that’s where most people entered this discussion) you can see the tremendous pressure exerted against DREs.  Some of the criticism has been politically motivated, but no matter what you think it’s very hard to find someone that is completely satisfied with pure DREs (no paper audit trails, etc.).  Not many people are willing to get up and profess their complete confidence in them.  The law requiring the VVPAT is now in place in something like 27 states and the Holt bill is closing in fast.

Plus, look at Florida; they have 15 counties with DREs (no VVPATs) that now have to get new equipment.  They can’t be very happy but this is the way the industry is moving.  What are their options?  Well, they can get optical scan from ES &S or Diebold, or perhaps Sequoia.  Not much of a choice really.  Keep in mind that some people have been using optical scan (mark sense) equipment for voting for nearly 20 years.  There are even some electronic voting counties that go back about 15 years.  So the move to optical scan seems like a backward move to me. 

When you investigate what’s out there on the market, there isn’t much development.  We hear that Diebold may want out of the business and is quietly shopping their election division.  We know that Sequoia is for sale.  Most of the firms just made their money with the HAVA sales, and now times are tight.  So don’t expect much in the way of development from the big 4.

Many of the smaller firms have DRE devices of their own. Sure they may have a paper trail component, but they are all post-HAVA DRE machines in essence; and that includes firms like TruVote, Avante and so forth.  Most of these firms either have had no sales or very little sales in 5 or more years.  So it would be hard to imagine them starting this all over again with a new product.

I just think the timing is right to start to think about new ways of voting, and with that respecting the general movement away from fancy electronic voting and the return to the primacy of paper.

 Q: Can you tell us more about digital pen voting?  How does it actually work in practice?

 A: This is described pretty well on my web site but I will cover it here again. But before we do that, let’s look at tabulation systems as a whole.  I break it up into three parts: pre-voting, voting process, and post-voting tabulation processes.  Most people that enter the business fail miserably because they focus just on the actual voting process and they ignore the pre and post voting processes.  Those are very complex and simply have to be done right.  In the pre-voting situation you have to build an election management system.  You either have to link to a vendor’s voter registration system, or you have to build it yourself.  But here you have to do all the things necessary to develop ballots.  That means you must understand what it takes to generate ballot STYLES.  To do that you have to know the precincts, precinct splits, district/precinct assignments, candidates, issues, races, etc.  All of that has to be hashed together in an algorithm to make ballot styles.

In addition you have to have ballot design software.  For example, if you are making DRE ballots for Sequoia you use Visio.  Having worked with those I can tell you it isn’t a simple process (it should be but isn’t with their stuff).  It’s extremely time consuming and a bit frustrating to create ballots with their product.  But you have to have that component as well.  And those ballot designs have to conform to state law in multiple states.

Then we have the actual voting process. This is the simple part.  The voter arrives at the polling location, goes in and votes and leaves.  All of this probably takes about 5 minutes.  It’s fairly well explained with DREs and with optical scan.  The only real change with optical scan is that the ballots are now scanned at the precinct before the voter leaves.  This allows the voter to correct any overvotes and potentially undervotes (if they are set to be flagged).  We’ll have to talk a bit more about HAVA to see the need for this.  In the past, the voter could deposit the OpScan ballot in the ballot box and the ballot box was returned to the election office and they were scanned in on election night with a high speed scanner (such as a Sequoia 400c unit).

With Digital Pen Voting (DPV) the voter receives a paper ballot and takes it to a booth.  There is nothing special about the booth.  It could even be the old punch card booths that we all know and love (if we still have them).  You take your paper and put it on the booth table—you can place the paper any way you like.  You grab a digital pen that writes with real ink.  It is a bit bigger than a normal pen but it should fit comfortably in your hand.  The pen is “chained” to the voting booth so voters won’t walk away with it---just like pens at the bank.  You then mark you ballot placing and X in the required boxes, or filling in the circles.  You can even select write in votes if they are permitted.  When you have looked at your ballot and are convinced you are finished, you return the pen to the little pen holder (a cradle with a hole in it).  You wait a few seconds and then a small screen alerts you to any overvotes or undervotes.  You can then request a new ballot in the case of overvotes, or you have the opportunity to take care of undervotes.  In any case, after you are finished you press a button on the screen and your vote is confirmed. You then take your ballot and place it in the ballot box.  That is the process in a nutshell.

When the pen is placed back in the holder the votes for that person are added to a laptop or other device that stores those votes.  Yes, it is possible to even have them sent wirelessly via 802.11 or Bluetooth devices.  However, that might not be legal in many jurisdictions.  Redundant storage can be assured as well.  At the end of the evening the data (could be a thumb drive, or other removable media) are securely stored and returned to the election office—just as it is done with both optical scan and DREs.

When that data arrives back at the office they are gathered together and the total tabulation for the election takes places.  Then you use your standard reports or web exports to display the total reports.

However, unlike OpScan ballots, these ballots can be also displayed on screen.  Therefore it is possible to view the actual ballots with the pen marks from each voter.  This is very useful in cases where you have to make a call about voter intent.  It is possible as well to make very large “blow ups” of the ink marks and see what the voter’s pen was actually doing.

Q: Why haven’t we heard about this technology for voting?

A: The paper itself is a development of the Swedish firm Anoto.  It’s actually be around for several years.  But the most natural markets to pursue would be healthcare.  They use a TON of forms and are really backwards.  I used to think that real estate people were the most technologically backwards until I realized how bad healthcare is. It’s almost a national disgrace. Elections traditionally haven’t been cutting edge, but, in fact, I think our track record is better than many other industries.

There is a test case of this technology now underway for the City of Hamburg Germany (for 2008) and that will be closely watched.  I have a document on my site about their use of this technology.  However, in the US there are so many barriers to entry that many people just haven’t really taken it seriously.  There are a FEW vendors out there that are digital paper experts and are “claiming” they have developed a solution for voting, but I’d be very skeptical about such claims.  One, if they have no special expertise in voting I’d be deeply worried.  Those same firms also tout that they are serving healthcare, homeland security, education, transportation, banking, insurance, and hog farming.  You can’t serve all of those industries if you only have three people in your firm.  So I’d be very skeptical about any firm claiming to have a system of voting on Anoto paper if they aren’t entirely devoted to the election industry.  If you look at the firms that serve the election industry, virtually all of them have one line of business and it is voting.  Diebold, with their ATM business is probably the exception and even they know now that voting is probably less than 10% of their revenue and 99.9% of their negative publicity.  Do the math.

By the way, you can buy this stuff for your own use.  Logitech sells their IO2 pen system (with paper) for less than $150.00.  With that you can take paper notes and when docked with the cradle will convert your handwriting and send it to an MS Word document, or Outlook, etc.  It will also capture your drawings or doodles as a graphics file.  So if you just want to learn more about it, you can get involved fairly inexpensively.

Q: How did you get involved with this technology?

A: Well, I’m always on the lookout for new developments in elections and I have several other projects in the works.  I hope that you’ll hear more about those as they get more fully developed.  Some of them are kind of hush-hush right now.  A good friend of mine, Steve Hilsman, has told this story on his Blog site (see, http://sbclist.blogspot.com/2007/06/pen-and-paper-new-thin-client.html ).  Steve is an expert on thin-client computing and server based computing and has his own uses for this technology.  Since we collaborate on various projects, I thought I’d see what he thought of the paper/pen idea.

It happened, as do many interesting things, as a bit of blind luck.  I bought the paper at a closeout store because I was out of legal pads.  I just thought it was some kind of gimmicky paper, but it worked for taking notes.  Then as I read the cover to it (where you can order more of it) I felt that I was missing something.  This paper was somewhat special.  It was then that I discovered what the paper actually was and that to use it effectively you’d need a digital pen.  Well, I didn’t really need a digital pen, but the use of this kept rattling in my mind until one day I started to wonder if it could be used in voting.

That led then to a series of calls and contacts to digital paper companies.  I found they were quite excited about the possibility of using it for voting, but they lacked the expertise to tackle the market.  Since then we have been discussing ways to move forward with a test case, or a “Proof of Concept” for digital paper voting.  That’s where it stands now.

Q:  How do you envision an end-to-end solution with this technology?

A: First of all, I doubt that anyone can tackle this all without creating some partnerships.  Voting is an extremely complex vertical market with a long history of littered bodies of those that have failed in it.  So no digital paper vendor in their right mind would try to take it on and develop all those pre and post tabulation processes we discussed earlier.  You would want to partner with an established tabulation firm.  Now, as we discussed, many of those firms may not want to get involved with new ventures, but since the digital paper only addresses the actual voting process, the hope is that you could simply drop that in the middle and have a complete A to Z solution. 

Q: How does this product compete with OpScan?

A: It competes very well with optical scan.  To understand the issues surrounding optical scan systems, I would recommend reading all the technical documents from Professor Douglas Jones (Counting Mark Sense Ballots).  This is a great primer to the issues and problems with Mark Sense ballots.    As you will see there are various types of optical scanners each with their own issues and concerns.  However, one of the biggest problems with OpScan is that you have to have an outside firm print your ballots.  We all know the story there.  You send the samples, and they return a proof.  Yet, we constantly hear about printing mistakes.  They are printed on expensive card stock paper and when a mistake is made it can be costly—either to the election office or to the printing firm.  All of that is factored into the cost of OpScan ballots, but they aren’t cheap.

Keep in mind that with optical scan all the ballots either go to one or two scanning stations.  That creates a natural bottleneck.  It doesn’t take long to scan a ballot, but still you may have several booths where people are filling out ballots all coming over to one central scanner.  With digital pen voting, the “casting” of the vote takes place when the pen is returned to its cradle. So if you have 5 booths, for example, then votes are being cast at 5 locations—much as it is done with DREs.

Optical scan readers are prone to LED reader failures.  If you lose one of the LEDs that read the ballots, then (potentially) you have to take that unit out of service.  If that is the only reader in a polling location, then things will really slow down, and a replacement unit has to be put in place or delivered and set up.  With digital pen voting, if the pen dies or malfunctions, a new inexpensive pen can be placed in that voting booth.