I spend a lot of time with my wife playing Pandemic. We love the fact that it is hard to win, but a cooperative rather than competitive play.

I'm in the same boat. I have two old Dell Inspirons and the battery has failed in both. I just run them off of power, and hibernate/suspend-to-disk when done working (which I have configured to happen automatically when I shut the lid). It's not as nice as having a working battery, but it's useable. I could buy new batteries, but last I checked, the official Dell ones are several hundred dollars, and third-party knockoff ones are almost a hundred. That's not worth it for my 10 year old laptops that I got for free.

<troll>Ah, Windows... the gift that keeps on giving.</troll>

Seriously, though... this is pretty ugly. It checks back every five minutes for each machine. You would think that Sony IT would notice that network traffic (or, say, the fact that all of their Windows desktops started listening on port 443). The moral of this story is run an IDS, scan your network, and pay attention to it all! :(

A few notes:

Python's newer abstract base classes allow you to make types that specify the presence of abstract methods and properties and you can use isinstance(thing, base_class) to achieve something similar (and thereby making handling types more familiar to foreigners)

With respect to Java and call stacks; Java has no easy way to dispatch to a function by name. You either need to make a class hierarchy so that you can use virtual call dispatch; or you need an if-tree (which is ugly but underappreciated). I've seen a few Java brains melt when I give them something like:

        cast_spell[spell_type](spell_data, casting_context)

Of course, I could just make a base class for spells and grow a giant, sparse API; but the benefit of playing the default-implementation / function overriding game is suboptimal. At the end of the day, the class hierarchy is just a data structure to determine how different "types" of functions get dispatched. As it happens, it spreads things out in a way that makes them hard to visualize. It often forces me to shadow unrelated sections of other APIs that grow on the same base-class.

In short, it's a crappy data structure for the purpose and it makes a pain out of gathering the knowledge to know how that dispatch happens. Sure, the code is "type-correct", but that doesn't say very much about being logically correct. As it happens, type errors are easy errors--but figuring out how the class hierarchy isn't serving your logical needs is *not* an easy error.

Don't get me wrong, I'm not fundamentally opposed to a good type system. First-class functions in Haskell or the like could handle this with type-safety. However, our industry just isn't ready for all of the pipes, beards, and sweaters. So, given the choice between Python and Java, I pick the one that give me methods for dynamic dispatch that don't abuse an already incredibly limited inheritance mechanism. I'll take the language that allows extensible, generic function application instead of getting it as a side-effect of their implementation of an already stunted type-system.

Uh...Philae was a European probe, not NASA. Your argument is invalid.

Gregory Benford had a great column about this, all the way back in 2000. It also involved a nuclear powered satellite.

It's human nature to react more extremely to new things, especially if they seem "unnatural." This might have been a survival instinct in bygone days, when the hominid who noticed that bush was out of place could take another path and avoid getting eaten by the sabertooth tiger behind it. But like so many such instincts, it translates poorly into the technological era.

Google has published a similar study: http://research.google.com/pub... a few years ago. I wonder if technology changes have had much of an impact.