Musics I done

Sunday, January 23, 2011

StatChem

I was Playing Space chem, as you can see from this picture.

what's extra-great about spacechem is how it rates your performance against everyone else's, as you can see from the three frequency graphs on the left.

look at that top one. a perfect positive skew* - but then, a massive modal spike just underneath1000 cycles.

the spike clearly doesn't fit into the population, totally bucks the trend - you can see the shape of the rest of the data is from one population, the spike is an interloper. a separate group hiding in the population, revealed like an x-ray by looking at the data.

So, why the spike? This level has an unlockable achievementy thing - complete it in under 1000 cycles for bonus something. Well, that explains the position of the spike.

But why no smooth curve to blend it in with the rest of the data? Why the anomalous rise and fall? As I said, it's a separate population. people are either just trying to complete the level, or going for the 'just under 1k' mark - there's no points for getting nearly to 1k, it's all or nothing - digital.

I think this graph says something very profound about the effects of achievements on game play, and i'm not exactly sure what. My role as a statistician ends with me pointing out the anomaly. why it's there, and whether it's right or not, is up to the philosophers.

over to you, dan.








*I can never remember if that's a positive or negative skew - neither seems to make any sense.

Edit:
What's also a little bit annoying that StarChem leaves out is the correlation between one form of game play and another - i want to know if the people who didn't use many reactors did it with a payoff of a longer run time, or did they manage to fix both of them? on one level, i was well under for reactors, but massively high for run time - is this usual? is it possible to score well in all areas?

Edit 2:
I just completed this achievement. it's funny how completely optimised you think something can be, but then to be given the incentive to reduce it further, you find something else - one little pathway to shorten. eventually i shaved it down to 1014 cycles, and couldn't find any other way to improve my machines - so i just made the pipes longer to serve as an ad-hoc overflow container and hey presto - while it felt like a cheat, prolonging overflow rather than actually improving the machines, it just about worked. and then i said 'right, got it under 1000. now i can move on.' and that's why achievements are bad.

No comments: