German Elections Go Open Source

Get Behind the Mule writes "The Heise news ticker is reporting that the software used by the German government to handle the results of the Bundestag election (that's the national parliament) on September 22nd will be based on open source platforms. The system will be written in Java and deploy Tomcat, JBoss and MySQL, and is being developed by the Berlin software firm IVU (here's their press release), working with the Statisches Bundesamt (the federal statistics office). It's not clear from the announcements whether the source code of the application itself, and not just the servers it runs on, will be publically available. Nevertheless, one is reminded of the argument of Peruvian congressman Dr. Edgar David Villanueva Nuñez (seen recently in Slashdot) that open source software enables citizens of a democracy to see for themselves whether the work of government, such as elections, is conducted as it should be. All of the announcements are in German, so go fish. The software, as described in the announcements, will compute preliminary results (which are announced as soon as possible after the polls close), run plausibility checks, and determine the Bundestag membership as well as distribution of seats to the political parties. It will use web clients for entry of voting data, data import, presentation of results, and preparation of printed results. It will be based on a three-level architecture (apparently standard J2EE) and deploy Enterprise Java Beans."

Comments

Better than Babel?

[clickety6]

Try the following translator instead of Babelfish.
I think it gives a more readable result, especially as it keeps the paragraph formatting.

http://translator.abacho.de/translate.phtml

MySQL?

[Anonymous Coward]

Why MySQL? Postgresql would be more reliable surely?

Judging from the amount of posts that Slashdot drops anyway. You need to accurately record every vote, you can't drop 1 in 100,000 even.

Maybe if they are using MySQL 4 with transactions and all the other stuff, then fine. But really, Postgresql is a better match. And preferable is a solution where you can sue someone if it all goes wrong...

Re:MySQL?

[Tet]

MySQL is fairly reliable; the process on our mail server has been up for hundreds of days

This is exactly the sort of anecdotal evidence that open source advocates need to avoid. The fact that it's reliable to that extent is completely irrelevant. Businesses (and in this case, governments) don't care if it goes down occasionally. Sure, they'd rather it didn't. But what they do care about is that if it does go down, they don't lose data integrity. It's far less costly to have 2 hours downtime than it is to have garbage data in your database (potentially without you knowing about it). MySQL doesn't have the ACID properties, that provide this level of assurance, and until it does, it won't really be suitable for this sort of use.

Imagine a concept like this being used in the US

[dmouritsendk]

The election results would be prosponed forever, because every single candidate that lost. Have hired a horde of computer scientists to find possible problems with the software being used, that could have effected the results ;)

Re:Whereas...

[Tom]

Actually, mostly Tony Blair himself. Ever noticed how much the guy loves to be in the company of Bill Gates?
In contrast, the german government had a left-swing in the last general elections, and with the leftist green party came a bunch of people into the parliament that had actually heard of or even - gasp - used Linux. Microsoft only realized when the parliament was publicly discussing using Linux for all its computers, and retaliated with massive lobbying, winning at least a compromise.

So this is only the latest event in a number of battles for the european governments.

May not release source

[CDWert]

They may NOT release the source, and they dont need to.

If they are using OpenSource components, such as the server enviroment application servers etc.
They dont need release the source. People ranting that they MUST release the source, etiher are lost in a fantasy.

If I write an application for counting chicken egg hatching probablity and It runs under Tomcat, JBoss and MySQL I needent release a single thing as long as I dont use any GPL suff in the code I am handling myself.

That said it would be nice if they do, who cares if they dont. A private software company is doing the development, they may or may not have some kind of agreement or future plans for the software being written.

I am getting pretty sick of all the OpenSource neophiles barking they must release the code blah blah blah. I think you are probably a large part of the reason MS calls the GPL viral, and people actually belive them. It isn friggin poison fruit. The other reason MS calls the GPL viral of course is projects like this get sold on building upon OpenSource applications, taking gold from a dragon has a tendency to piss it off a wee bit.

Hmmm

[NiftyNews]

"The software will...run plausibility checks"

Hopefully they mean on the votes. If you ran it on the candidate promises you'd have a 95% failure rate!

And in other news...

[Ethelred Unraed]

In a strange result of the September 2002 general election in Germany caused by an unknown quirk in the software, Linus Torvalds was elected Chancellor, with Richard M. Stallman foreign minister. Thanks to Stallman's diplomatic skills, 104 countries declared war on the recently-renamed GNU/Germany. Film at 11.

Before you start thinking the US should try this..

[Asprin]

There it a push in the US to standardize the election process to try to head off the kind of thing that happened in Florida in the US Presidential election of 2000. I don't know how long Germany has used a standard voting system or whether they've done it - to this point - US style, but I anticipate a lot of 1337 s1ah5d07 p0s73X0ring along the lines of 'we should do that too, 'cause Florida 2000 sucked!' I want to head off as much of that as possible.

The US Forefathers were smart - they intentionally left the specific details of how to collect the vote and tally the results to the states, and ultimately, the local county districts. They weren't concerned as much about regional cultural and financial differences as much as they were concerned about the integrity of the election process.

If I wanted to rig an election in the US, I would have to rig it ONE COUNTY AT A TIME, because each election office makes their own choice how to operate on the voting day in question.

With a centralized, standard voting system like Germany's open source plan, I would just have to know how to rig one system.
The Florida election worked exactly as it should have - the election was just really close. It sucks that we couldn't just call the election at 10PM and go to bed, but you know what? Your vote *does* matter.

GNU Free voting project

[bckspc]

If you're interested in this sort of thing, check out the GNU Free voting project at http://www.free-project.org/. From the site:

We are a free software project creating Java electronic voting software released under the General Public License (GPL). With this software we aim to:-

Provide a secure and private system

Create scalable and reliable software

Offer a non-commercial, non-partisan voting alternative

Use the GPL to create an open system that Internet users will trust

Release a system that can be used to support the growth of effective democracy anywhere in the world Additionally, in support of our wider development community, the project aims to:-

• Advocate the free software paradigm
• Evangelise the use of technology to strengthen democracy within a holistic understanding of the current malaise i.e. Internet voting alone isn't going to solve turnout problems

As an official GNU package and one of two electronic voting projects of FreeDevelopers.net our Free Software evangalism aims our particularly important.

Ummm...

[ttfkam]

PostgreSQL is also easy to install, setup, and use (admittedly without replication -- that's true). It is just installed with RPM and "createdb". No biggie.

It's not ACID, but very reliable? That's a bit of an anachronism. ACID isn't a library or a protocol to which you must be compatible -- it's a minimum guideline for reliability. ACID is a contract that says that if you put it in, you will be able to get it out again, unharmed, unchanged, and every time.

Atomicity: In a transaction involving two or more discrete pieces of information, either all of the pieces are committed or none are.

Consistency: A transaction either creates a new and valid state of data, or, if any failure occurs, returns all data to its state before the transaction was started.

Isolation: A transaction in process and not yet committed must remain isolated from any other transaction.

Durability: Committed data is saved by the system such that, even in the event of a failure and system restart, the data is available in its correct state.

----------

In short, by saying that "it's not ACID, yet it's VERY reliable" in reality means "it's not 100% percent reliable, but it's at least 85%."

If I delete a record that is a foreign key reference for other records, will MySQL guarantee that the record cannot be deleted or that every record that points to it is also deleted dependant upon admin preferences? PostgreSQL does.

MySQL has many more utilities to repair and maintain integrity of its databases than PostgreSQL. It's true. But then again, the authors of PostgreSQL have designed and engineered the datastore so that catastrophic data integrity failure cannot occur in the first place. Hardware failures (memory loses bits or hard drive fails to write correctly) can cause it, but even pulling the plug, while putting the most recent changes in jeopardy, will not destroy your committed data.

And before anyone comes forward with tales of PostgreSQL data integrity issues, please check that you are talking about v7.1 or later (available for more than a year).

Now then, on to issues of vacuum. I admit this can be a time sink. As you run updates and inserts on a PostgreSQL database, the indexes are not quite as efficient as they started (hence the need for vacuum) but it's not all bad as that. Most people would only have to run vacuum every night at 3am (or something like that) from a cron job. In short, PostgreSQL requires a minimum level of maintenance for optimum perfomance. Note that I did not say that it needs maintenance to keep your data safe -- only to maintain optimum performance. And nothing replaces good backups for any database.

And before you say, "But PostgreSQL's vacuum locks the table so you can't get work done," please note that 7.2 removes this constraint.

Will someone honestly try to say with a straight face that you would trust a MySQL database with gobs of data for an extended time without performing at least minor maintenance? If someone does, please tell me that they aren't running the German election database...