That NASA paraded some idiot to push some algorithim or computer generated leaders made us the laughing stock of the world.
The basis of scientific research is that you don't need to study the entire population to come to conclusion about them.
You study a sample of it. How do you get perfect representation - by sampling. Mathematical formulas have proven for example that you need max of 2000 people to represent millions...even if you increase that sample size to 1 million..it will of very little precision utility.
You're better off trying to improve your randomizing function. Random numbers and functions are key mathematical construct. True random numbers are impossible.
Now 50,000 polling stations, with 500-700 votes each, randomly submitting the results (not holding them like in rigging scenario) as soon as they are finished from all over the country, will mean you can call the election in few minutes, with some margin of error. If it's close - like say +-3 - then it will be a long night.
Nobody complained in NO of 2010 - and the numbers did exactly that - they quickly settled on 31% versus 69% - and remained that way.