<p>Wow. And we don’t even have hanging chads. :eek:</p>
<p>[Kucinich</a> Granted New Hampshire Recount - Politics News Story - WMUR Manchester](<a href=“http://www.wmur.com/politics/15028468/detail.html]Kucinich”>Manchester, New Hampshire News and Weather - WMUR Channel 9)</p>
<p>I’m kind of glad, I really want to see if Diebold fears are legit or just hype.</p>
<p>Guess we are lucky we even have paper at all, some didn’t even want that, sad to say</p>
<p>How are they going to recount the machine votes?</p>
<p>I always wondered what happened to the ballots after you feed them to the machine that looks like a shredder. How would you recount the machine votes? 
</p>
<p>Good question.</p>
<p>let the CC community get creative!</p>
<p>a local radio show host was talking about this yesterday. He starts off many of his suggestions with (in a self-mocking way), “When I’m king, I will do _____ to fix this problem.” His suggestion for accurate voting involves marbles. When you sign in at a voting precinct to vote, you are given a certain number of marbles, reflecting how many positions/issues you have to vote for. You take your marble and dump it in the bucket with your candidate’s name on it. In the end, you count up the marbles and the one with the most, wins. </p>
<p>So people were calling in, playing devil’s advocate. Some questions were resolved with a slight change in logistics, such as having a tube that leads to a locked container (this would also prevent someone from grabbing a handfull of marbles out of one bucket and putting them in their candidate’s bucket). When you insert your marble, it feeds down the tube to the locked container. </p>
<p>Other glitches in the system… how do you keep someone from bringing in extra marbles and ‘stuffing the box’? Uniquely colored marbles might help. If one is found in a box that is not the proper color for that race, then it gets disqualified. Could you put some sort of sensors in the marbles in order to detect illegal marbles? </p>
<p>You could even call results quicker because maybe you can weigh the containers and the heaviest one wins, as opposed to counting them all (assuming all marbles weigh the same). </p>
<p>So, I’m sure there are some very bright people here who could come up with possible flaws, and then solutions to those possible flaws. Any takers?</p>
<p>Not just the shredder - the computer! There are documented cases of votes changing on the computer right in front of the voter’s eyes.</p>
<p>
</p>
<p>All voters in New Hampshire use paper ballots (with SAT like bubbles). Some districts hand count while many districts feed them through optical scanners, which is where they are saying the problem occurred. The percentages of the hand scored districts (Obama ahead) don’t match up with percentages in the scanned districts (Clinton ahead).</p>
<p>What the article you quoted doesn’t say is that if he wants the recount, he has to pay for it.That’s the NH rule. As of this morning, I haven’t heard whether or not his campaign is going to cover the costs.</p>
<p>When I voted, the attendant standing at the optical scanner (that looked like a shredder) said to put the ballot face up. When my husband voted, he put his ballot face down. Does the scanner read both ways??</p>
<p>Maybe if you looked suspicious enough they told you to feed the ballot in upside down so the vote would be not counted… ah ha… That must be it. It was a left wing conspiracy. :D</p>
<p>It looks like the recount process has started. The $2000 initiation fee has been paid.
[Kucinich</a> claim spurs N.H. recount - BostonHerald.com](<a href=“http://news.bostonherald.com/news/2008/view.bg?articleid=1066072]Kucinich”>http://news.bostonherald.com/news/2008/view.bg?articleid=1066072)</p>
<p>Hand scoring was likely done in the smaller districts which, if you look at a NH electoral map, are in the northern part of the state and in an arc along the western border. The areas where Obama ran well are almost exactly the same areas where Dean ran well. </p>
<p>Clinton’s strength was in the larger cities of Manchester and Nashua…which almost certainly used machine reading of the ballots.</p>
<p>And the news in NH just won’t go away for the Clinton campaign.:o</p>
<p>[CNN</a> Political Ticker: All politics, all the time - Blogs from CNN.com](<a href=“http://politicalticker.blogs.cnn.com/]CNN”>http://politicalticker.blogs.cnn.com/)</p>
<p>
</p>
<p>I don’t think that’s the optical scanner. I think that’s the “lockbox” that stores the ballots until they are taken to be scanned. It just keeps a running tally of how many people have voted.</p>
<p>This silly recount is just part of the continuing effort to undermine the legitimacy of the winning candidate…much like the Obama campaign’s ongoing effort to blame the result on racist voters in New Hampshire.</p>
<p>If it is the same contraption we use here, the “shredder” is actually a scanner, and the paper ballot is retained within. This is acutally an electronic system with a full paper back up.</p>
<p>The NH scanners do read both ways, face up or down, and sound exactly as mafool stated.</p>
<p>I checked. It is the scanner. It is as mafool described. I have no idea what a lockbox is. Maybe that retains the ballots after they are scanned.</p>
<p>the black box reads the ballots both ways and at the end of night sends the final count to the election headquarters. At the same time a ticker tape tally of all the precincts & ballots cast is printed out and publicly posted.
Some ballots are not able to be read, those ballots go in the side and are hand scored at headqtrs in a process overseen by reps from both parties.( same with the mail in ballots- mail in ballots are confirmed with signatures)</p>
<p>yes, sokkermom, I believe that is correct.</p>
<p>A simple system, with instant electronic recording and a complete paper trail.</p>
<p>I keep wondering why other locales go/went for something more complicated and error-prone?</p>
<p>I worked in the scanning and data extraction business for about ten years. This is something I actually understand!</p>
<p>The technology used in those “scanners” is a combination of several components. First, there is a paper scanner that transports the paper past an imaging system that makes an electronic image of the document. Any imperfections in the mechanical system can produce an imperfect image. Any of you have ever had a weirdly stretched Xerox copy can understand this – it’s the same problem. Scanned images are imperfect. The scanned image will be a single-bit (black and white) image, probably scanned at 200 dots per inch, although it might be scanned at 300 dpi. Rarely is a mark-sense application scanned at the sort of resolutions we’ve come to expect in photo scanners, and rarely would it be done in grayscale or color. The image sizes get very large, very quickly when done at higher resolutions or in multi-bit color, and the results can actually be worse than in single-bit mode at lower resolution. Sometimes the simple answer is best.</p>
<p>Once the image is created, the software takes over. There are marks on the page that the software can use to standardize the image, and to detect whether it was scanned upside-down, or even bottom-end first. The standardization marks allow the software to rescale the image to a standard form if the paper transport was not running at exactly the right speed, or if it skewed during transport. The standardization routines can fix may errors, but they can’t do magic. It is possible to generate a bad image. By the way, the paper transport can work perfectly on one page, mess up the next one, and then work perfectly on the third one. Something as simple as a voter who ate a glazed donut just before handling their ballot can cause problems in the transport. The software that detects orientation and page order is pretty robust stuff, and works well. It’s not likely to introduce errors of its own, but it may miss problems with the source image.</p>
<p>Once the image is oriented and standardized, the optical mark recognition (OMR) software will attempt to tell whether a certain box is filled in or not. It does this my comparing the number of dark pixels in a certain area of the image to the number of dark pixels in other areas. In other words, the person who set the system up will program the software to look over 200 pixels from the edge of the image (or actually from the standardized registration marks on the document – but lets keep it simple), and 300 pixels from the top edge of the doc, and then count the dark pixels in the box from 200-250 pixels on the x-axis, and from 300-350 pixels on the y-axis. The proportion of dark to light pixels is analyzed to determine “is this box marked.” It then will conduct the same analysis for every check box on the form. I’m not sure how the ballot scanners are programmed (I worked on the core software, but we had thousands of customers using it to develop their own specific applications, like voting), but most people doing OMR will then program logic like, “Box 1 is 5% dark, box 2 is 90% dark, box 3 is 1% dark, and box 4 is 11% dark, therefore the user meant box 2.” The OMR software just tells the developer how many black dots there are – the developer needs to figure out what that means. They could have a rule that says, “To be a valid selection, only one box can have more than 80% dark. If there are two boxes that are more than 50% dark, I’ll reject this selection. If no box is more than 80% dark, I’ll reject it. If one box is more than 80% dark, no other box can be more than 20% dark. I will only accept an answer if one and only one box is more than 80% dark, and the others are all less than 20% dark.”</p>
<p>I dragged you through that to demonstrate that even ballot scanners are computerized processes, that they could be manipulated by software developers of fairly limited skills, and that even if the paper is sitting in front of a person, they have to develop rules, because people will make stray marks, will get jelly donut on the ballot, will not follow instructions, and generally will act just like they are trying to mess things up on purpose.</p>
<p>As a long-time professional software developer, paper is about the worst choice you can make. Humans suck at repetitive processes, and if there is anything more repetitive than hand-counting ballots I don’t know what it is. Scanning paper generates its own problems. In most applications, this scanning and OMR process is acceptable with an error rate of up to a couple of percent. This would have had a material effect on several national- and state-wide elections in the last ten years. I am NOT a big fan of using paper for voting.</p>
<p>If you are willing to accept that the system was not written by someone with active criminal goals, a pure software system with a computer terminal of some kind is by far the most likely to reflect the wishes of the data entry operator (the voter). </p>
<p>I’ll leave you with this, why don’t we fill out little slips of paper and then insert them into our ATMs for our personal financial transactions? It’s because it is time-consuming and error-prone compared to letting users simply use a keypad and screen to enter information. Why do we trust ATMs but don’t think the same people can build a trustworthy voting system?</p>
<p>Washdad:</p>
<p>To answer your final question: I get a receipt from the ATM at the time I make the transaction, and then I receive a statement from the bank that reflects the same transaction. I can verify the proper entry and computation at both the “voting” end and the “counting” end. If any bank in my town were to routinely miscount the votes, er, transactions, I promise that there would be he** to pay.</p>