I finally had the breakthrough needed to actually make some progress on Save the Singularity. The current game looks quite a bit different than I described in my last post on the subject, and it’s almost reached the point where it can be sent out for playtesting.
It’s still a 1-many player cooperative push your luck dice game. The Singularity, the moment when a smarter than human AI starts reprograming itself to become even smarter, has an intelligence explosion, and takes over the world over night, is coming. No one can stop it. What you and your allies, the other players, are trying to do, is ensure that when it comes, it does so to the benefit of humanity. This extremely complex goal is known as Friendliness, and you need to finish your research before the Singularity or else everyone is doomed.
You do this by trying to get Friendliness Research to a certain level before General AI Research gets to a certain, different level. You do this by Researching various areas of AI, trying to get as far as you can, making as many breakthroughs and new contributions as possible without overreaching your grasp and accidentally spoiling your code by making so many mistakes that you can’t catch them in time to avoid rendering the code useless (or dangerous!). Then, when you stop either to watch your code compile or to weep at your foolhardiness, progress inexorably marches on towards the Singularity-for sure it continues, but how quickly is impossible to tell in advance.
Each turn is divided into two parts: your research, and the general march of progress. Your research consists of rolling an ever-increasing number of d6 Research Dice, until you either stop on purpose or are forced to, followed by rolling the single d12 Progress Die, until you are allowed to stop.
You start rolling 3 Research Dice. The Research Dice are all identical and have three results on them, in different proportions: Success-gain 1 Friendliness Research point, Promising Lead-roll an extra die if you continue, and Mistake-if you roll two or more of these in the same throw, the research part of your turn ends and you score no points.
So if you rolled S, PL, M on your first roll, you would have one point sitting there hoping to be scored, and you can either press on or stop and score that point. If you press on, you roll a total of 4 dice because of the PL. Say you do and you roll S, S, PL, M. You now have 3 points available to score if you choose to stop here, and if you press on you will roll 5 dice next. Your turn doesn’t end from you rolling a second M because you have to roll both of them in the same throw-for instance, if your third roll is S, PL, PL, M, M, then the Research phase of your turn ends and you score no points. Whereas if your third roll is S, S, S, PL, PL, perhaps you decide to stop while you still can and score your 6 points.
Then the rest of the AI community continues its research. This is represented by rolling the single, 12 sided Progress die. There are three symbols on this, which show up on the various faces in various combinations: Bad, Extra Die, and Stop. Yeah, they need better names. Each time Bad comes up, a point is scored in the General AI Research category-remember, you’re trying to get to a certain number in Friendliness before AGI reaches its goal, so that’s why this is Bad. Extra Die represents research being done that is somewhat useful to you, but also potentially dangerous, and makes it so that on the next player’s turn, they start with an extra Research Die on top of the 3 they normally begin with. Stop means you are allowed to stop rolling the Progress Die.
And that is the current state of the game! The next steps are to figure out good values for the Friendliness goal, the AI Research goal, and the distribution of symbols on the Progress Die. Once a little bit more work has been done on that, it’ll be ready to be sent out for playtesting, to no doubt discover many terrible flaws I didn’t see when thinking it through.