Wednesday, December 08, 2004

Clint Curtis on Air America...and other tidbits.

Forgot to mention, Curtis interviewed yesterday with Air America Radio though I don't yet know the air date for it. They had originally hoped to schedule him for Marty Kaplan's show which airs on Saturdays, but as things are moving quickly on this story, they may be airing it sooner. Will update as I learn more.

The BRAD BLOG can also confirm that very high level folks have been informed about the Curtis issue (and have, in fact, visited this site).

He's also (as mentioned previously, but you may not have caught it here) been interviewed by several mainstream media folks. With more scheduled.

END OF DAY 2 on this story, and still no major holes in his story. (Or even minor ones yet that I know of!


Anonymous Anonymous said...

Nice job Brad. I thought people might be interested in the following: Personal funds Feeney spent on his campaign: $0

Total Yang contributions: $5250
Not exactly breaking the bank for a tech company with NASA and FDOT contracts...not to mention a "20 year friendship" with Feeney.

12/08/2004 2:11 AM  
Anonymous Anonymous said...

Curtis may testify at Conyers hearing…today…WED 10 AM ET, CSPAN and Pacifica Radio

12/08/2004 3:25 AM  
Blogger Noah Diamond said...

Curtis is on "Unfiltered," on Air America, right now. Outstanding work, Brad...

12/08/2004 6:42 AM  
Blogger Noah Diamond said...

...and apparently the complete, 16-minute Curtis interview will be posted sometime today either on or

12/08/2004 6:47 AM  
Anonymous Anonymous said...

Hi! Yesteday, on a board I frequent, I posted a link to the article on that subject which you had mentioned:

along with a good part of that article. A highly knowledgeable and intelligent poster had this reply:

len harrison
This part of the article is very odd
Tue Dec 7, 2004 5:11pm

"According to Curtis, Feeney and other top brass at Yang Enterprises, a company located in a three-story building in Oviedo, Florida, wanted the prototype written in Visual Basic 5 (VB.5) in Microsoft Windows and the end-product designed to be portable across different Unix-based vote tabulation systems and to be "undetectable" to voters and election supervisors."

There are several anomalies in this single paragraph.

1) VB 5 was not the current version of VB in 2000. The runtime for VB 6 was put on Microsoft's site October 14, 1999.

2) VB is a Windows specific language. Unless you are planning on your finished product being Windows only, it's not a good choice for a prototype, especially since VB programming beyond the trivial tends to rely heavily on Windows API calls none of which are portable outside the Microsoft OS.

3) To the best of my knowledge, central tabulators are more commonly Windows than unix-based, certainly Diebold's GEM system is.

4) Although the paragraph quoted talks of tabulators, a later paragraph mentions flipping votes on touch-screen voting machines. The all electronic voting machine market is essentially owned by three companies: Diebold, Sequoia, and ES&S, which enjoys a familial relationship with Diebold (literally, not corporately). Diebold is the biggest. Sequoia, I believe, the oldest. Both Diebold and Sequoia are Windows based systems. I do not know about the ES&S iVotronic and a visit to their Web site didn't clarify. The question of unix portable code would, at most, seem to apply to ES&S systems at the most, and only possibly.

5) As implied in 4), the article and allegations seem not to distinguish between tabulators which collect the data from individual machines and precincts, and the voting machines themselves.


Now, on the other side of the puzzle there are several possibilities.

1) The unix question. Broward County, Florida, which uses ES&S machines (or more likely their precinct tabulators) had a problem with their count in this years election. The machines began counting backwards. The story, from the Palm Beach Post, is no longer online, but the Google cache of it can be
found here.

This seemingly bizarre phenomenon occured when a candidate received more than about 32,000 votes. From a CS standpoint, it's not that bizarre. The data type chosen for the total was simply not wide enough and was defined as an integer rather than a positive integer. At the machine level, negative numbers are conventionally expressed by setting the high bit which is also set when the data width is exceeded. Presumably the printf() function did not allow for a negative result while the data itself did. This kind of serious bug is indicative of poor project management, inept coding, and incompetent QA, especially since the code exists for the purpose of producing this single number.

It is also the kind of thing that the MS automated test routines catch easily and the kind of thing that becomes less likely when using the Microsoft Foundation Class within Visual Studio (a separate product from Visual Basic), the baseline environment for enterprise level development for Windows. Microsoft made some terrible architectural decisions early on, but there are some things they do halfway right. One can imagine this happening with less sophisticated and integrated tools such as might be found in a shop running linux and picking up non-integrated tools willy-nilly. So perhaps ES&S, aside from producing fatally flawed code for counting election results, is working in Linux or some other unix environment.

2) Sequoia machines produced voting flips in 2000 that were detectable by the voter as did Diebold's machines in the Georgia primary. Both of these are documented in Votergate. It is thus possible that the people in question were concerned about making this happen without detection, though how they expected to load such a program into individual machines is difficult to grasp. With Diebold's President on record as promising to deliver Ohio to Bush and his brother running ES&S, I'd think they would be the folks to go to, not some outside contractor.

3) Specification of VB Quite frequently executives don't have much of a technical clue and good programmers, like filmakers, actors, musicians, and other artists tend to lean left, which is perhaps why the code for these machines is said to be so terrible by those who have seen it. It is possible that it was just ignorance that was involved in speccing an obsolete version of a Windows-only programming language for a the prototype of an app intended to be ported to unix.


Nevertheless, IMO, this story is likely disinformation. It is blatant, has additional problems beyond those in this paragraph, and sensational enough to become the focus, displaying all the other issues related to voter fraud in 2004, many of which include obvious criminal behavior such as harassment and intimidation, and equally problematic issues like major overvotes within counties, statistical anomalies that aggregate astronomically, and changes in counts like the one mentioned in this piece. This story, because of its "smoking gun" nature is much sexier than these others. As disinformation, it would be intended to draw all the attention to itself whereupon it is blown out of the water; and, however illogically, then becomes ground for the dismissal of the real issues.

* * *

I hope this comment is of use to you.

12/08/2004 6:49 AM  
Anonymous Anonymous said...

WARNING: Clint's "affidavit" looks very suspicious. I think it is some kind of trick. The last page uses a different font and has a completely different format. It looks like he took the last page of some document, typed up the first three pages, and then put them all together. Have a look and you will see how obvious it is.

12/08/2004 10:13 AM  
Blogger Noah Diamond said...

Yeah -- but I don't think that's suspicious. I think the last page of the Curtis affadavit is a scan of the physical document, because that's the page with the notary seal, and the signatures.

12/08/2004 10:53 AM  
Blogger BradF said...

To whoever posted the link and answers to the Madsen story. Please be careful to distinguish between what Curtis actually says in his affidavit, and various other elements that Madsen reported in addition to the verfiable Curtis stuff.

For more on that, please see this response to it that I posted here yesterday.

12/08/2004 6:21 PM  

Post a Comment

<< Home