Looking much better.
At the moment, I don't quite see how the replayability will be extended beyond the number of edicts I choose to take and the units I can set loose. It reads to me that the creator controls almost every variable in the eponymous creation of the island. In this regard, the blurb does not the reveal potential sources of randomisation:
In terms of controlling every variable: no, you don't. Most of the land tiles that pop up are not by your choosing, and the bandits popping in are also not by your choosing. We also have some other stuff that we probably won't introduce before profiles reach a few levels in (to give players a bit of breathing room at the start). Josh and I have talked about a "suggestions" mechanic from The Master, but lately I've been thinking a "propositions" (not in that sense) mechanic from units themselves might be more interesting.
There's also randomization in a very butterfly-effect sort of fashion. In other words, just having a few tiles different, or a guy making a random roll slightly differently, means that the outcomes are different. For instance, I had a scenario that I was testing just last night to make sure something worked: Adamantine, a mythological token. It gives the one dude who picks it up 100x his normal health and attack -- holy heck, right!? But it also spawns 20 bandits at the end of that turn. In one outing of this, he killed all the bandits within a few turns and had 65% of his health remaining. In another 14/20 of the bandits were remaining after he died. The difference there was both in which bandits appeared, and where.
Anyhow, there are already a triumvirate of goals in any game as it stands:
1. Make it to the end without failing your edicts or having genocide.
2. Make the highest score possible... because, come on, it shows you're awesome.
3. Work on the 100 challenges, which unlock new stuff, and which are not something you'll blow through in a couple of hours.
In other words, for $5 the replay value is completely off the hook. I wouldn't say that it has AI War levels of replayability by any stretch, but neither did AI War when AI War first came out. If Skyward takes off, I hope to do with this what we've done with AI War, in terms of the combination of free DLC and paid DLC to keep it growing for a long time. Of course, I've said that with every game since AI War and that's never happened -- but we've never broken even, let alone made a profit, on any game but AI War either.
Anyhow, I get what you're getting at above, but I'm not yet sure how to really address that beyond what is already stated. It already talks about the challenges in the bullet points, it makes a big deal about both bandits and free will, and it implies heavily that figuring out an optimal strategy is super hard. All those things, when I read them, say "replayability." I don't mean that in a snarky way: different things pique different people, so I'm just genuinely not sure what is the missing element for you and am trying to find out.
- When I create the island, do I play a game of Carcassone with all the tiles in my hand?
It's funny you mention the randomization of what you can place in Carcassone-like fashion. That's exactly how this game started and was conceived. And oh MAN was it not fun.
Or do I play whatever is available to me at the time so I can't easily do things like place village on hill -> make killzone with marsh -> stock archers -> village become invincible to melee units.
Bear in mind that everything costs resources, and you are pursuing multiple objectives at once. If you had no secondary objectives, then sure you could just set up a stalemate in various ways and everyone would be safe and happy. However, if you don't create military units then your cities crumble into crime. And your military units won't stay still if they have access to enemy towns or enemies in general: they will run off and attack. So that archery stronghold you mentioned would instead be a breeding ground for archers running around the map, not staying where you wanted them to. If those archers prove TOO effective, you're going to be struggling against yourself on the other side to fix what you just wrought.
On the other hand, if you block off your archers so that they can't reach the enemies directly but can just shoot at them, that actually would work... for a little while, until you die.
See, the military units won't actually move unless they have an enemy in their sight range or an enemy town center that they can path to. So if you make the enemy fortifications perfect, you'll get a backup blockage of guys in your "perfectly safe" town. That sounds fine, until you learn that more than one unit can't stand on a tile. And that military production facilities can't produce units while someone is standing on them. And then you remember the crime factor, and in a dozen or so turns that perfectly safe town belongs to the bandits from forces within.
The whole "do I do whatever I want" sort of argument is kind of like saying the same thing in any any strategy game. And I know the next argument in that: "but you're playing against a (human or AI) opponent there, rather than playing both sides." Which is true, but here you are playing against an equally challenging... let's call it "environmental situation." If you just doodle around, the game kills you.
- I can bring into being any god in every game?
Yes, but for purposes of challenges and otherwise you're encouraged to choose different ones at different times. Also, depending on the map or on other circumstances (ie what else you are trying to accomplish in a specific game), you'll find that some gods are way better suited for some things than others. Lots of bandits around? Yeah, you're going to pick Ares most likely. Working on lots of trade? Foolish not to pick Pan.
But the thing is, you're rarely doing just ONE thing at a time if you're playing at an advanced level... so the choices aren't so obvious. Or even if the choice of a god is obvious, choosing when and how to use his/her powers certainly is not.
- Does there exist drand() in the creatures' decision algorithms? (Is it is just irand() or no rand() at all?)
I'm not familiar with drand vs irand, so I'm not sure what you mean. But there's a fair bit of randomization, yes. The units can be somewhat predicted if you made a spreadsheet and looked at the exact board state at the moment. Or just have an intuitive sense for the game from long playtime. But the thing is, the circumstances are so ever-changing and the board state grows and changes after every turn in nontrivial ways, that there is a lot of "oh man now that just happened" stuff thrown your way, heh.
The blurb seems to me as a more of a bloodied-zen exercise in discovering best moves. After I find satisfaction, I will stop playing (that's not actually a bad thing). Essentially, my primary fear is 100% player-driven optimisation. While it's true that the game is highly complex, it will not take long before a small set of seemingly-optimal strategies are found - such as chess. The key decision for freshness is when to step off the grid, there is no pRNG to push you. Whereas in AIWar much of the fun was in the pRNG - nothing like discovering teleport raider+battlestations (although not really viable for high-difficulty), or Zombard+TDL, to decidedly change your options.
Well, the text specifically mentions that there are no optimal strategies. The edicts themselves also have WAY more influence here than the AI types do in AI War, I should add. And the challenges really are going to throw monkey wrenches into your usual strategies for a good couple of dozen hours or so. Beyond a couple of dozen hours, Josh and I are aware that players could fall into something of a rut (once all challenges are complete, basically). That's where our thinking on either suggestions or propositions come in, as those will add basically a fourth layer of randomization to this (bandits, map growth, and unit AI being the first three).
The reason that the situation in AI War is so interesting isn't because of some singularly brilliant AI algorithm -- what you would need for grandmaster-level Chess. Rather, it's because of all the layers on layers on layers of AI and randomization and so forth. Skyward is turn-based and so some different things apply, but rest assured we are thinking about ways (prior to 1.0) to address the "I've played this to 'completion' in terms of challenges and am now done with it" effect. The effect, which I might add, is what eventually has driven me away from every RTS or TBS I played, even the ones I loved. Usually took between 0.5 and 2 years for that to sink in, but it always did.
Another key thing with AI War has been that the game has actually been constantly changing over 4 years. All those expansions and free DLC... well, I don't think people would still be playing the game without those. Presuming enough interest in Skyward, that's again what I want to do here.
I can see where this may end up going - a strategically masochistic 4X-god game. One can only hope the tokens implemented will play off each other and produce desired outcomes. I foresee complaints along the lines of "I buffed the Greek archers with invincibility, but they didn't go on the attack!".
Free-will is a fickle thing.
Your archers will never just sit around if they have any route to enemies. If they are sitting around, it's your fault. The complaints would come in the form of "argh, you made small decision X instead of Y, and now my larger schemes need some adjusting." That's part of what Josh and I both were adamant the game needed:
somewhat predictable AI in the units. If you have archers, and they have somewhere to go, you can be 100% sure they will start heading out. Which exact place they go, or who they meet and how they fight along the way... that's a different matter. But since this plays out over turns, you can kind of see how things are developing and airdrop minotaurs or whatever where needed.
The free will here isn't terribly fickle: it's where the rand() you're looking for comes from.
