SteamOS is Linux based but I doubt it's going to be anything even Stallman would call GNU/Linux.
Everything I've seen makes this sound like it's more aimed at being a 'Console which runs PC games' than a normal computer. I'd expect it to load into a 'Big Picture' mode Steam client, and allow the user to launch their games and specially-modified applications from that which could well run as overlays like the existing Steam browser. Whether this machine even needs a command line is debatable, it shouldn't need GCC (I'd expect a fully binary-based OS) or a full-featured window and compositing desktop like Gnome.
If there's no climate change though then there's no extra damages to pay. If they believe firmly in climate change they weight it heavily in their predictive risk models, if they're certain that climate change is incorrect then they can ignore it in their predictions.
Either way it's going to hurt if you're wrong should other insurers have different predictions. If you think climate change will occur and it doesn't then those who predicted correctly will will been able to sell cheaper than you and you'll have needlessly lost market share, if you predict that it won't and it does then you've underestimated risk and will be paying out a lot more damages than you expected to when setting the price.
Blizzard's main priority with World of Warcraft is getting people to keep paying their subs, and to do this they make the game as engaging as possible. This goes against that by both managing to destroy the sense of immersion by dragging gamers out of their game world, and also by forming a link in the player's mind between Warcraft and real-world scenes of suffering. Not a connection that most players will want in their recreation time.
Where things may work better is where it's possible to both turn the work itself into a game, and also to wrap it in an appealing layer to stop it having too strong a connection in the player's mind with the reality behind it. An example of this would be the recent Facebook game developed to help identify some genetic factors in Ash tree dieback, as detailed in this BBC News story. Here the presentation is cute, and the focus is on making it a game. The only problem I could see here is that I can't see how it's cheaper/more efficient to develop and serve the entire content for even a simple game compared to just doing the pattern matching in a more traditional manner, but for other tasks I could see it working.
The basic idea is there though, make the work part of the game rather than making it a task which detracts from the game. Something which this story doesn't seem to recognise.