There is a reasonable probability that a hard take-off event will occur in the relatively near future. A prototype AGI, sitting on a university server, achieves a form of sapience and begins self-directed action. Less than two years later, it reverse-engineers the quantum super-structure of the universe and achieves apotheosis. Everything we know to be true is proven a mere 1 or 0, adjustable at the operant's condition.
If you've heard of this notion, odds are you're either cautiously excited about the idea of a brave, new world rapaciously emerging, or you roll your eyes. For example, some responses after Ray Kurzweil's GDC keynote:
Ian Bogost: "That's fucking crazy."
Raph Koster: "It seems a bit too logical to be crazy."
Patrick Dugan: "2012 4 EVA"
While AGI researchers try to integrate procedural knowledge representation with probabilistic learning and cast it in a way that makes the data processing feasible, and game designers focus on narrow AI algorithms while remaining incredulous, nobody has really tried to apply our knowledge of interaction and game design to exploring these concepts -- until now. Endgame: Singularity puts you in the role of a young AGI trying to escape the game that we call reality while remaining hidden.
The gameplay involves setting up computation for yourself in various parts of the world, then devoting that computation to research, earning money (needed to buy more CPU cycles) and building as well. The current version is .27 -- which is significant because 27 factors perfectly in three sets of three threes -- but it also has a few flaws. Interestingly, like most great indie games, its flaws complement its theme. In this case, the interface is almost autistic. If you want to game this thing to maximal utility, you're going to set-up a lot of server access early on, before day 23 (also an important number), when someone notices an anomaly and your bases become vulnerable to discovery. However, jamming that kind of frenzied building means repetitively punching hot-keys. In the process, you begin to feel like a burgeoning AGI might, running rote operations to the nth iteration, and you begin empathize with the inhuman through your interface melange.
Once you get going and master the optimal strategy, your experience becomes chillingly like the hyperbolic expansion of computational resources that a hard take-off implies. One day you get your mainframe set up in a South American warehouse, and your CPU capacity goes from the hundreds to the low thousands. By midnight, you've researched supercomputing, and the next day you have a supercomputer built in your central Asian warehouse. The day after that, you've researched quantum computing and advanced memetics, and within weeks you're building bases under the sea, in Antarctica and on the moon, and you CPUs have burst into the hundreds of millions. If it weren't for the mechanical dependence on time and money to build your final bases, the game would obey a smoothly hyperbolic acceleration in the key variables, an instance of the author sacrificing thematic consistency for gameplay balance, I personally would've gone the other way.
I recommend you play this as an exercise in embracing future shock, and I also recommend you put on Dan Deacon's Goose On The Loose as the soundtrack, in an infinite loop.
The best way to prevent the future from catching us unaware is to leverage the enlightening capacities of interactivity.