Slashdot Mirror
Voters News Service: What Went Wrong
ddtstudio writes "Baseline Magazine has a pretty good recounting of how even the national TV networks can have a computer network go wrong -- in this case the night of the last U.S. election. From the article: "VNS had been trying to rewrite and retool the system for years. This was just the most recent attempt and it failed miserably."
Oracle, IBM, BEA Systems -- all crashed."
The systems in question were mainframe computers running IBM's Operating System 390.
Not that i'm a linux fanatic, just wanted you to get your facts straight.
... and one page version of article.
Well, no where in the article is it mentioned that there was a problem with IBM and Oracle. It just says that there were delays in transferring data. So I don't know why Oracle and IBM were named in the original post.
What's under yellowstone?
From article:
Now, it does not mention what OS they were running WebLogic on in the article, but it was definately not OS/390.
Forget the whales - save the babies.
From the article:
Back up to Election Day, Nov. 5. The balance of power in Congress was up for grabs. Yet by 10 a.m., the TV networks confirmed what they had feared for months: They couldn't derive any meaningful exit-polling data from a system they had just spent between $10 million and $15 million to overhaul.
That's 2002.
OK Folks,
I'm one of the lead programmers of one of the members
We KNEW this was going to happen a LONG time before November. At the end of the article, they talk about "set a deadline 2 months ahead of the real deadline"
Guess what? They did!!! They were supposed to be ready for the NJ Primary, which was before the summer - they missed it, BIG TIME - That's when the alarm bells went into overdrive for me
I understand (This up in the levels above me) that the steering committee didn't realize that the technical committee was saying "We've got a BIG problem"
Another warning sign was when their test data generator that they sent us in the spring generated XML that didn't match their own XSDs - and that was with all the fields declared as cdata - the field names didn't match
The first test, which was supposed to be months ahead, came weeks ahead, and even the most basic message (just a heartbeat) didn't work. That's they day I knew it was doomed for sure. Our prime efforts switched to our backup data source at that time. THAT worked fine. I had a boring election night, watching VNS crash, and laughing
One stipulation: That the new system use more flexible and current programming languages--Java and the Extensible Markup Language-- rather than OS 390 to gather, compute and deliver data to the media outlets.
That sounds great. People who have no idea how to accomplish the goal telling the people tasked with doing it, how it should be done. I can't believe it failed. They should have laid out what they wanted to acheive and left the rest up to the designers on how to meet those goals...
Also, some interesting older information on the VNS can be found a the Votescam website. Although they seem to have a few extreme views, along with some wild conspiracy theories..
-- -- Warning. Do not stare directly at the sun.
In New Zealand we have pretty much a total political news blackout on election day until the polls close. They can talk about voter turnout estimates etc but nothing much else. Exit polling is illegal. All billboards must be done by midnight before election day. Party volunteers giving rides to elderly people etc to polling places can have coloured ribbons on their cars but no party or candidate names.
IMHO, these rules all make good sense. For one thing, we don't have billboard trash lying around the roadsides for weeks after an election.
I remember a radio interview with someone in South Korea the day before their recent election. The interviewer asked 'how was ???? doing in the polls this week?', the answer 'we don't allow polls in the week leading up to an election'.
First, the "cast a vote free from the influence of network prognostication" issue is a red herring in the case of the 2002 election, as that only really affects Presidential elections. US networks generally voluntarily refrain from projecting the results from any given state until all or most polls in that state have closed, so particular Congressional races would probably not be affected by early projections.
.sig: be the majority of voters.
But generally, it is important to note that America essentially places no special restrictions on election reporting. With regard to Presidential elections, part of the issue is that the contiguous USA spans four time zones (UTC-5 thru UTC-9), with Alaska and Hawaii extending at least another couple of zones west. Each state (and in many cases lower levels of government) sets its own polling hours for the convenience of its citizens (usually closing c. 8 or 9 Pm local time), so the poll closings roll East to West. Since the Eastern part of the country holds enough electoral votes to elect a President, it is possible for the networks to project a winner before the most populous state (California) finishes voting. Bills to introduce a uniform poll closing time for Presidential elections have been introduced in Congress (notably following the 1980 election which was called for Reagan before 9 Pm EST (UTC -5)), but they have never passed. Florida was hit in the 2000 race by the fact that the western panhandle extends into the Central time zone, and the state had polls close an hour later in that (demographically very different from the southern part of the state) area... so the networks (which at that time were following a "most of the polls have closed" criterion) called the state while people were still voting in Pensacola.
-- Remainder of my
Save Maine's economy: write stuff down. All comments are exclusively my own, not my employer.