Wednesday, May 27, 2009

niXtuff Makeover

Over the next short while (before the expected late July election) we will be giving niXtuff a makeover, and will be making over our projection matrix.

I will share with you a story of how we came to be.

I have always had an interest in politics. In the 2004 election, I remember trying to research local ridings to figure out what would happen. At the time I lived on the east coast and was only 19 years old. The 32 local ridings were enough to fill my plate, but I also took an interest in ridings in Saskatchewan and Manitoba, which seemed easy enough to remember. I was able to make some projections, but many were far off the mark. It was after this time that I first stumbled across the University of British Columbia's, Election Forecaster, found here: http://esm.ubc.ca/forecast.php I've been in limited contact with the Forecaster's creator, Prof. Werner Antweiler (he was the one who sent me the base data that allowed me to 'win' the BC election projection race) I loved the idea of the forecaster and soon became married to it.

Using the forecaster, in 2006, I made a shocking projection that no one in their right mind would believe. I said the Tories would win 10 seats in Quebec. Some of the biggest projectors of the time were laughed at for saying they'd win as many as 6. I was clearly the radical. I was spot on. The problem was I had not really told anyone who cared. Much like catching the soap at the last minute in the shower, this was a personal victory I had not really shared with anyone. It was after this I had the first inklings to create this blog.

During this whole period, I participated in two Political Simulations. Politics UK http://politicsuk.net/poluk/ and Politics Canada http://z3.invisionfree.com/Politics_Canada/index.php? (which is undergoing a restart, if you want to join, now is the time!) In Politics UK, they used the program UK Elect http://www.ukelect.co.uk/ for elections. I wanted to use it for Canadian elections, and in fact, over at Politics Canada, we had in the past. There was only one participant, however, who truly understood how UK Elect worked, and he flatly refused to teach me what he knew! Fear I'd replace him I suppose. I was left in an impossible dilemma, I cannot use the program I need. What to do.

I figured if I cannot use UK Elect, I may as well just build a new one myself. And thus was born my ElectoMatic. The ElectoMatic is superior to UK Elect in some ways in that it is built to simulate Canadian elections, not british ones, though both can be modified to do the other. The ElectoMatic is also cheaper (free, though I keep a copyright on it) and far easier to use (type in your poll numbers and the rest is done for you) though it does lack certain things.

One thing the ElectoMatic currently lacks is how to shift projections in a wide manner. I will explain how the ElectoMatic currently works. Lets take a party and call it Party A. Party A took 40% of the vote last time. This time, they are polling at 44%. 44% is 10% higher than 40%. 10% is therefore the modifier we use. If they took 40,000 votes in Riding A, we assume this time, they will take 44,000. And if they took 50,000 last time, we assume they will get 55,000 this time. That's all fine and dandy, and it works very well for small shifts of a few points, but when you get into huge shifts, and swings, it is just not enough.

Hencefore, I will be doing a makeover of the ElectoMatic to allow for swings, and to properly simulate growth from small parties. Lets assume the Greens somehow manage 60% in the coming election. That's a 1000% increase. Does that mean Riding A, where they got 4% last time, will see them get 40% this time, while riding B where they got 8% last time, will see them get 80% this time? No. The realistic results would be closer to 64% and 68%, though not exactly of course. Hence I will allow positive numbers to increase a result, while not allowing negative numbers to decrease it (a fault of UK Elect, which often elects parties with negative popular vote numbers) I feel these changes are needed to make the program work as best it can.

Lastly, the Blog itself. I have been somewhat lax on updating links that are no longer applicable. This has not been a public transit blog since 2007, and yet I have not removed that data. I will be doing so over the next short while. I will also be revamping our forum where I can more easily post data without cluttering up the Blog itself.

In short, good things are happening, and we fully expect to be ready and raring to go by the time the next election rolls around. Make sure to tell your friends:)

No comments: