German Court Bans E-Voting As Currently Employed 82
Kleiba writes "The highest German Court (Bundesverfassungsgericht, Federal Constitutional Court) ruled that electronic voting machines like Nedap ESD1 and ESD2 are not permissible in Germany. Der Spiegel, a well-known German newspaper, is featuring article on today's decision (in German; Babelfish translation here) which was the result of a lawsuit by physicist Ulrich Wiesner and his father Joachim Wiesner, a professor emeritus of political science. The main argument against the voting machines in the eyes of the Court is that they conflict with the principle of transparency. 2009 is a major election year for Germany, with parliamentary elections in the fall." Reader Dr. Hok writes "Voting machines are not illegal per se, but with these machines it wasn't possible to verify the results after the votes were cast. The verification procedure by the German authorities was flawed, too: only specimens were tested, not the machines actually used in the elections, and the detailed results (including the source code) were not made public. The results of the election remain legally valid, though."
Re:what's the difference? (Score:3, Interesting)
To be honest though, I'd normally trust the results of a computer collecting tallies over people counting them by hand.
Which is more trustworthy is an interesting question.
On the one hand, computers are more systematic. They don't make random errors and they don't have political biases, per se. However, programming errors or deliberately-introduced bias mean that their speed allows them to miscount lots of votes really fast.
Humans make more actual mistakes, and may be prone to shading the results, but their counting is done in the open and at a pace where other humans with different biases can double-check their results. In fact, recounts are typically done by groups of people, with multiple individuals from competing parties examining each ballot, announcing their results audibly and similar checks on the tallies.
Given that sort of a counting structure, I think there's no question but that humans are maximally accurate. No machine will match them. It's also very expensive and inefficient.
Now I shall enlighten you all by explaining THE SOLUTION:
Voting should be done on computerized machines that print human and machine-readable ballots. Counting should be done by:
All ballots are counted by method (1), every time. This is the nominal result of the election.
A statistically-valid sample of voting machines should have their ballots pulled and method (2) applied.
A statistically-valid sample of the ballots are pulled and counted with methods (2) and (3).
Any discrepancies among the three counts should motivate expansion of the sampling, ultimately to first a full machine recount of the ballots and if necessary a full hand recount.
The definition of "statistically-valid sample" must take into account the margin of win of the closest race. Enough recounting must be done to assure that the election tallies have a margin of error that is small enough to assure sufficient confidence that the closest race is decided correctly. Since the recounting is easy, I'd probably ask for a 99% confidence interval.
Also, after the required sample size is determined, the random selection must be implemented properly, using transparent random number sources such as dice or lottery ball machines. Representatives of the parties should oversee the sampling and recounting, and the parties should agree on precise sampling techniques before the election.
There you have it. The Answer.
Next I shall tackle World Hunger by ensuring a statistically-sufficient average distribution of food.