Hunting The Muse

Precinct Reporting Is Weird

Here were the precinct results in WA with 16 precincts reporting:

       John Kerry      Dem     593     35.0%     0
       Howard Dean     Dem     486     28.7%     0
       Dennis Kucinich Dem     432     25.5%     0
       Wesley Clark    Dem     66       3.9%     0
       John Edwards    Dem     58       3.4%     0

(update: This probably was supposed to be 16% precincts reporting.)

And here they were with 140 precincts reporting:

       John Kerry      Dem    1321     49.3%     0
       Dennis Kucinich Dem     499     18.6%     0
       Howard Dean     Dem     487     18.2%     0
       Wesley Clark    Dem     151      5.6%     0
       John Edwards    Dem     138      5.2%     0

(update: I corrected a typo in the above table. Plus, 140 precincts has to be wrong, because there are an average of four delegates per precinct in WA. This had to have been a typo on the AP's part. It's more like 1300-1350 precincts.)

That strikes me as odd. What's with Kerry practically sweeping 124 4% of the precincts, except for Kucinich having significant support and Dean coming in last with only one additional delegate?

Here's a history of the WA precinct reporting.

Update: I also wrote about this over at daily kos where it is starting to raise some concern. I'm relieved because I thought it would just look like I was trying to make an issue out of nothing. I just want to nail down the loose ends. Others are sharing stories about how when a precinct came up with votes, the computer operator input a different number of votes. Please read on or participate if you have ideas for next steps, if any.

Update: There is a follow-up post at dailykos detailing one voter's experience of witnessing delegates being unfairly taken away from Howard Dean.

Update: There is another another follow-up over at dailykos that shows status from the above WA voter in trying to verify precinct data.

Update: Make sure to read my main weblog to see more status reports.

Read on for more details...

This is the screenshot of when 140 ~20% of the precincts were reporting. Dean gains one delegate out of 124 4% of the precincts, while Kucinich gains 67.

Then for a short while, CNN had the following numbers up - with 21% (around 1376) precincts reporting. Kerry's numbers are slightly less, and Kucinich's are signifcantly less, enough for there to appear to be a much bigger lead for Kerry. The Kucinich situation is strange because he had a strong total for a more than one phase of reporting, and here his numbers go down significantly. If this was a simple error, what was it? These numbers, used by CNN and Fox, were not AP numbers, and Fox was overheard saying they got their numbers from the Democratic National Headquarters.

Almost simultaneously, the AP updated their numbers with 1397 precincts reporting. Notice that here, Kucinich's numbers are still high, and Kerry is below 45%. This set of numbers stayed up on the AP wires for several minutes.

The previous CNN batch of numbers was only up for a couple of minutes, before they were replaced by this next set of numbers. Notice that these numbers still have little relation to the AP numbers, and Kucinich's numbers are still low. This set of numbers with Kerry at 50% is the one that was reported in all the articles and repeated ad infinitum by Fox News. Note that they say 32% precincts are reporting.

Here is where it gets stranger - soon at 2:31 PST, the AP released new numbers, with 2248 precincts reporting, later revised to be 2125 precincts reporting. This has more precincts reporting than the above CNN/Fox numbers, but Kerry has less delegates, and his delegates are more in line with all the other AP numbers. Also note that Kucinich is still up in the 500 range - also consistent with the AP numbers, but in conflict with Fox and CNN.

This makes me curious about what the heck was up with that set of numbers that Fox/CNN was reporting, of Kerry at 50% with 1924 delegates - which was the set of numbers reflected in the wire articles. They don't even appear to be based in reality, even when these were the numbers reported all over the news.

The revision from 2248 precincts reporting to 2125 precincts reporting:

At 2:49, CNN finally revised their numbers. The Kerry@50% numbers were at 32% reporting, while these are at 34% reporting. Note that Kerry's numbers are revised downward, significantly. Why were Kerry's numbers inflated for the set of numbers that all the news articles talked about?

Note that as of 2:54, Fox was still referring to the other set of 34%-reporting numbers that had Kerry at 50%.

Soon, at 2:58, CNN updated their numbers again. While everyone else's numbers grow significantly, Kucinich's numbers are again revised downward, which again, is weird.

Things start progressing more regularly from this point. At 3:22, 55% (3588) are reporting.

CNN, WaPo, and AP project Kerry at 3:43.

At 4:20, all three sources release numbers from 71% precincts (4624) reporting:

Additionally, the AP has a first draft of delegates apportioned - this is not the complete slate of delegates, but perhaps only from certain congressional districts:

From this point on, they grew basically proportionally until the latest results, 97% reporting:

---

So here are my questions. For someone like me who doesn't work in the Secretary Of State's office, I would expect that once a number is received from the precinct, it is vetted before it is released to the AP. And when 500 precincts are reported, I'd expect that all candidates' delegates should grow or at the very worst, remain the same, when 1000 precincts are reported. Instead, we're seeing a lot of examples of delegates going down. So most of my questions concern where the errors are happening. Do we know that certain precincts vastly misreported their delegate count? Who is it that has the power to revise the numbers? How could Kucinich's numbers have sunk by over three hundred delegates when each precinct didn't have more than twenty or thirty delegates to award? Could that many precincts have been wrong all at the same time?

But my bigger question is, where's the opportunity for monkey business? How do we know there isn't any cheating going on here? I'm a programmer that has to deal with security issues on a daily basis, and I know that security isn't just about monitoring the entry and exit points - you have to be vigilant at every step along the way. When a precinct captain calls in their numbers, how do we know that the other end is recording those numbers? When the precinct captain goes home, they can't trace it back in reverse. Where's the list that shows exactly how many delegates each precinct got, and how that adds up to the total, so every precinct captain can then refer to the same document and verify that that was their total? I heard that in other states, this information is not published because the parties want to keep that demographic information private from other political parties. If that is so, it is definitely opportunity for fraud through obfuscation.

I am not alleging fraud, or even opportunity for fraud, because I honestly don't know enough about how it works to point out the opportunity. But it did strike me as weird that there was so much downward revision of delegate numbers after they had already been reported.

Update:

This is with 99% precincts reporting. Note that there are still over 4,000 delegates left to be awarded out of 27,000. There are 55 precincts left to report. Remember that there is an average of four delegates per precinct. What is it that is up with the remaining precincts? What are the chances that only 55 remaining precincts - with 4,000 delegates between them - would mirror the percentage split of candidates that has already been reported?

Posted by Curt at February 7, 2004 03:07 PM