That's a total of 2,514 votes out of 46,000,000 people - 0.005%, to shift 11 seats between victory for one group and victory for another.
While any voting system has points where a shift of a few votes will tip things over the edge, FPTP seems particularly prone to this. By breaking things into 650 individual elections, you are going to get many more where things are incredibly close, and 3 votes means everything.
AMS/MMP (as used in Scotland for MSPs) uses much larger groups for the additional proportional representatives, so there's much less chance of an individual area being that close to a tipping point. And STV (also used in Scotland, for councils) also clusters representatives together, for a similar effect.
It just seems ridiculous that targetting such a tiny number of votes can trigger such a massive difference in result.
Original post on Dreamwidth - there are comments there.