It's got to end some time, Kevin

I got an email from Dad basically saying, jeez, I hope you didn't upset your flatmate when you said you wanted him out by Christmas. It's funny that both my parents seem more concerned with his well-being than mine. I took him on for the extra money, not out of the goodness of my heart, although I think I have been pretty good to him - I've given him a lot of leeway (probably too much, partly because his past made me a bit wary). He has it pretty good here, with all those cooked meals (OK I'm not that good a cook) and all that housework that magically gets done, often when he's still in bed. He has the whole place to himself when I'm out, and has almost free rein on the TV and whatever the hell else he likes. His classes are a ten-minute walk from here. It's all very convenient for him and I can perfectly understand him wanting to stay. But...

The World Cup continues to excite and surprise, not that I can watch much of it. I did catch some of this morning's game between Brazil and Mexico; the Mexican keeper was in inspired form. I still have my booklet from the 2010 tournament, with all the results filled in. The group games from four years ago look a bit dull now; some teams (um, England?) operated entirely in binary, with every game finishing 0-0, 1-0 or 1-1. I'd written in some notes from my hotel in Bali - "Channel 13 (reception awful) or 8 (not much better)" and names of countries in Indonesian that I'd picked up. Ivory Coast was "Pantai Gading", Greece was "Yunani", New Zealand was "Selandia Baru". Watching the action in a foreign country, and not having complications like work or flatmates to contend with, made the whole thing more interesting.

Warning: if you don't like maths or logic or tennis, or probably all three, please ignore the next paragraph.
Last year I promised to do a hypothetical calculation based on the Wimbledon qualifiers, and post the result here, so twelve months later... Qualifying for the men's singles at Wimbledon (which is going on now) involves winning three matches: two best-of-three-setters and a best-of-five. There's no tie-break in the deciding set of any match. The question was, how dominant do the servers have to be to make a best-of-three, on average, last longer than a best-of-five? The reason why that is even possible, in any situation, is because you're more likely to reach a long third set than a long fifth set. If you make that final set long enough (and you do that by making both players demons on serve), the greater likelihood of reaching the final set in a best-of-three starts to outweigh the fact that you've played two fewer sets to get there. If you assume that both players are equally dominant on serve, and their dominance is constant throughout the match, it turns out that we need the final set to last, on average, 14 times longer than each of the normal (tie-break) sets. That might sound ridiculous, but to achieve this the server "only" needs to win 86% (or more) of points. That's a very high percentage obviously, but you might expect it to be higher. The average length of a no-tie-break set rises steeply once you get above 80% of points won on serve. By the way, if both players are winning less than 14% of service points (and I've certainly played matches where it's felt that way!) that would do the trick too.

A guy at work is due to become a father next month. So at work today they had a baby shower (if that's what they call it when it's the father). I ended up spending over thirty bucks on this event, which is quite a lot when you don't really know the bloke. And it meant you had to talk and mingle and all that tricky stuff. And I realised that there's at least an 86% chance that I'll never have kids.