Why polls are stupid

I'm really surprised what a big deal the media makes of polls, because they're very nearly irrelevant. The percentage of people who would vote for a specific party doesn't directly affect the number of seats that party gets. What's important is the distribution of those people among different ridings. Because the person with the most votes in each riding gets that seat, and the party with the most seats gets first dibs at forming a government.

Example:

There are 308 ridings in Canada. For mathematical simplicity, let's say there are 100,000 voters in each riding, and 100% voter turnout.

The Purple Party gets 51% of the votes in 155 (or just over half) of the ridings.
51,000*155=7,905,000
So the Purple Party received 7,905,000, and holds 155 of the 308 seats in Parliament, thus forming a majority government.

Meanwhile, the Yellow Party receives 100% of the votes in 100 (or just under one third) of ridings. 100*100,000=10,000,000
So the Yellow Party received 10,000,000 votes, but holds only 100 of the 308 seats in Parliament. The Purple Party still gets to form the government.

However, polls would have shown the Purple Party polling at about 25%, and the Yellow Party polling at about 32%.

I chose these numbers for mathematical simplicity, but there's room for a lot of variations in between. For example, because we have at least four parties running in each riding (and often more), the Purple Party could easily have won with something like 35% of the vote in 155 ridings, which would only show up as about 18% in a poll. There are entirely too many variables.

To be effective, polls would have to poll each riding. Which would be hella useful for people considering a strategic vote! But as it stands, polls of the country as a whole are uninformative. Please stop reporting on them as though they were useful!