Luminare - a game of inductive logic
Install on: Android or iOS.
Luminare is an inductive logic game about trying to figure out a hidden rule. Rules can start simple with "White groups contain a red symbol," and progress to "White groups must contain a red triangle or a circle but not both," or "Black groups have a green triangle in the last position."
If that's not clear, don't worry, it has a built-in tutorial that should explain things.
I started working on this prototype quite a long time ago and I've continued working on it every now and again in my free time. It's quite interesting to look back and see how different things are. Some of you may have played builds I sometimes bring to meetups, but this is a new and exciting build :)
If you have an Android device, there is an APK you can download (less than 5 MB), which I recommend. Otherwise there is a flash version which is pretty much the same (performance is slightly worse cause there's no multi-threading in flash).
So if you like puzzle games I'd be really keen to see if you like this, and if you don't I'd still really appreciate it if you had a look and let me know what you think :)
Feedback Questions:
1. Did you understand how to play easily enough?
2. Did you realise the levels continue to increase in complexity, as you go upwards, for quite a long time?
3. Do you typically like mental-heavy puzzle games like this?
4. Do you want to keep playing? Why?
5. What part do you dislike the most?
6. Any other feedback?
Install on: Android or iOS.
Comments
Feedback questions (1 and 2 I was spoiled for):
3. I do play plenty of puzzle games, but I wouldn't typically play something so taxing on the same brain function. This game is mentally tiring for me! Some people really like that though, it's probably a good thing (but also a niche game).
4. I found it very hard to leave a level alone. Especially after I'd failed a lot. It feels like you've got more than enough information, but you're just not seeing it.
As for the meta-game, the exploring aspect is interesting, but I would definitely feel like playing more if there was some kind of goal to work towards. At the moment it feels like the challenge is endless, and therefore you're not making a dent and can't really accomplishing anything. Most of the time you're picking a level at random, which means it can't be an interesting decision.
One way that might solve this (and still keep the exploration mechanic) is to tie "progress" directly to achievements; i.e. an achievement system that makes it feel like completing achievements is what you're "supposed" to be doing. It would probably be as simple as putting an achievement counter on the map screen. Just seeing a number that says 4/30 gives you a reason to keep playing (beyond pure challenge) and a reason to choose certain levels over others. I think you could come up with a bunch of achievement ideas that enhance the exploration mechanic too.
For example:
- Find 10 special nodes!
- Complete a 4x4 square!
- Reach letter Z!
- Complete 20 odd numbered nodes!
- Complete 10 nodes in a single row!
- Complete 10 nodes without a single incorrect guess!
- Make 1000 correct guesses!
- Revisit and complete 5 nodes that you previously retreated from!
Another potential way for the meta-game to go might be to make the exploration space less nebulous. Maybe you're exploring a network of rooms and passages? You could try and create that "Arg! I really need to clear this tile because it's the only one that leads to the next room" moment.
5. This is a tricky one. I guess I found the difficulty frustrating when I started progressing quite far? It's hard to know what the scope of the rules are, you don't really know quite what you should be looking for. I didn't use the hints, they probably would have helped, but also made the game less rewarding.
I do also think that anything that can be part of the rule space should come up in one of those failure hints at some stage. For example, if a rule like "White always has 3 symbols" exists, then a hint of "Is something special about one group's length?" should be possible. Just to make sure the player knows what types of things to look for. Every different rule variation should have a hint that lets you know it's possible. You can even tailor them when the player is doing badly (this might be happening already?).
6. Yes!
I'd really like it if the tiles that you'd failed on had their own colour (or unique appearance), so that I can remember to go back and retry them.
I also think you could rename the level exit button. "Save and quit" made me assume the whole game was going to close. "Come back later" or something similar might explain it better and make the game less frustrating by reinforcing that it's okay to stop trying a level.
1. Yes (but from before). One thing that I did find slightly confusing (and forgot from last time) was the black and white thing. The tutorial says "This group is white" "This group is black", and I waited to be told what they mean. You may want to be a bit more explicit: "Some groups are white" "Some groups are black" "You have to guess which s which". Also, just for the first few times, I looked at the group above the message box. (This is because that group just happened to be white against a few black ones or vice versa). You may need to mark the next group a bit more clearly so that it looks more important than the other groups.
2. I kind of assumed so, but one of the earlier levels was easier than one or two below it... so it did make me question it. The progression is also quite slow, so it is not so 100% clear that t is getting more complex (but I don't think you necessarily should change anything).
3. Hmmm... mental heavy. I would say this is the wrong type of classification in terms of these games. I play a lot of puzzle games; my favorite ones have ways to exploit their core mechanics in interesting and unexpected ways, many of which are hard. Pure logic games (Soduko) rarely fascinate me for long, unless there are some interesting twists (often provided by a human - Battleships is an example). And this game is not hard in the brain-damage way Braid or Portal (at times) are hard; it does not require new insights and wrangling them together. It is similar in spirit to 20 questions.
4. For the moment, yes. I do love the new 2D level selection; it's a bit of a meta mechanic. I was severely disappointed when I discovered the reward was "under construction". I would probably finish the game, if there are interesting things to discover for the amount of levels you have. (Discovery / Exploration is two of the tings I personally like most in games.) One thing that will definitely keep me playing longer is if you group rules spatially (not just complexity along one axis), so that getting to a new region gives you new types of rules (For example, off to one side may be more "color" rules, of to another more "shape" rules, another more "sequence" rules, another more "relation" rules, and so on. You could give these names, "Chromos"). I know this happens implicitly but making it linked will give it more meaning. Your reward scheme may also become a important driver. (Look how well the bears work in Alphabear. They give a lot more depth to a fairly trivial mechanic, and when you think about it, they are just basically a way for giving players a choice for which incentives they like).
5. There is not particularly anything I disliked. As I said, the game may only captivate me for so long. (One thing that did distract me a little was the tick marks in the message box... they don't align with anything and it drives me a little crazy :P)
6. I think the in-game hints are really cool (look at the black groups, is any color special), but I think they should be more relevant to the actual rule in play (they seem to be generic / random). They could be broken down versions of the explcit hints (which you can then remove from the menu).
You know, I was just thinking about Alphabear again... I am quite impressed with how they took a very simple fairly abstract mechanics, added some cute bears as rewards and gave them meaning in the game, to make a game that has quite more depth to it. For instance, imagine your tokens are food types, and you are guessing the diet of a monster or alien or dragon... so opening new levels are not new "rules", but new monsters / aliens / dragons.... and playing the landscape is really a way to discover these creatures. Guessing wrong makes the dragon ill... and you have to feed it to satisfaction to complete the level. There are of course many other things you can do too. Just some thoughts.
Something kinda important I forgot to mention is my overarching goal with this game. What I've experienced personally when playing this, and what I really want to exemplify, is that "Aha!" moment when you figure out a really crafty puzzle that's stumped you. That feels like a really rewarding feeling for me and I want to enhance that as much as possible. It is probably naive of me, but I think that showing people how to experience the reward of solving difficult logic puzzles is super interesting and could lead to them being better at other tasks outside the game.
A fundamental thing about Luminare is that the level outcome is binary - either you beat a level or you don't. Many puzzle games have some sort of scoring mechanism that allows players to get some reward based on their performance. Some examples of this are the star ratings in Angry Birds or the score in Alphabear. I've considered trying to tie performance to the number of guesses or moves it takes you to solve levels, but I feel like this is at odds with my overarching design goal and just doesn't really incentivize players. I don't think this is necessarily a huge problem I need to try fix, but it is worth noting.
So at the moment, by main goals are: continuing to focus on making the core gameplay as interesting as possible, and trying to add a good meta-game that incentivizes players to get into the game. I'm super keen to move away from the abstract look and adding a theme, a la Alphabear, but I feel like that probably requires some nice art I can't really generate myself, and I'm a little cautious of committing to a theme before all the mechanics are nailed down. I have quite a few thoughts on what they did with Alphabear and I think some were at the expense of the gameplay (even when they didn't really need to be), but Spry Fox are super good as "dressing up" relatively simple mechanics with a bucket-load of flair. Another recent game I think does this well is Divide by Sheep. At it's core it basically a game about moving around numbers, but it's dressed up so well it makes me super jealous :)
Good observation, you're completely right - which level to pick is not really an interesting decision. There is some small interest added by showing the question marks when you're close to a special node, but it's really easy to be in a position where you don't see any, and then you don't really have much of a meta goal. I feel like this is probably the main issue with my meta game at the moment. Thanks for helping quantify it :)
Cool idea, and I like your suggestions for achievements! What I'm hoping is that I can improve the meta game to the point where people get sucked in and then keep coming back to the infinite play. I guess this could be pretty similar to DD's "Advisor Tasks" in the way they can keep generating. Funky Smugglers also had a similar mechanic I found interesting. Having a limited number of achievements has a built-in incentive to complete everything, so I might struggle to incentivize completion of infinitely generating achievements. Gonna have to think about this a bit more.
Yeah totally. This kinda ties into my comment about theme, but there are some relatively easy ways I could add some structure to the world, like some well-placed rocks or such.
I totally agree that adding more hints and making them much more tailored to the actual level could help a lot, this is already happening to a much lesser degree. Something you might notice is that the groups that generate are themselves not all that random. If you're doing badly, the game is much more likely to give you more helpful groups. I love working on little opaque systems like this that improve the game without anyone noticing :)
Nice and easy-to-implement suggestions, thanks! :D
I've found this same issue come up quite a bit in playtesting. I feel like it's a combination of bad UI design and a lack of helpful tweening. After a bit of play, people seem to get over the issue, but I need to work on make it a non-issue.
If I'm thinking about the same things, those tick marks line up with the intervals the progress bar can stop at. Was that not so obvious? I thought it might help players get a better idea of the mechanic and their progress, but perhaps it's superfluous.
Hehe, this is the best complaint I have received by far! I was totally expecting no one to get that far, so it warms my cockles to read this. I will endeavour to supply you with a reward post-hate :D
EDIT: While I am thrilled by them, these walls of text might seem super daunting. If you had a look and are worried about this, I do accept shorter comments of the form: "Your game sucks, it needs more cute bears and witticisms!" ;)
"Look at these groups again, come on, it's obvious, pay attention!"
Some nuance could definitely be added in there. I still have a few more ideas I want to try incorporate into the group generation algorithm. At the moment, if the rule is "there must be an odd number of reds," it's super easy to confuse that with "there must be exactly one red." I want to improve the generation algorithm so it more clearly differentiates similar rules like that.
Sure, that's part of the puzzle; but the possibility space is technically endless. I know there won't be a rule like "Every white group has a yellow preceding a hexagon or an odd number of reds after the second symbol", because that's obviously too complex for anyone to be expected to figure out. But I don't know where to draw the line between a rule that is too complex and a rule that is potentially what I'm missing. I think the player needs some guidance to find where that line is.
The menu hints don't fully solve the problem because there will always be a percentage of players too proud to use them (unless you de-stigmatise them in some way). I think the hints for incorrect guesses are probably the best way to clue the player in to what kind of things they need to consider.
- Minor improvements to the tutorial.
- Indicate levels you're busy with.
- Rename menu options to be clearer.
- Remove explicit hint option from menu and fold it into the status hint system.
Still available at the same URL: http://clockworkacorn.com/games/luminare/----
@hermantulleken: exactly, that's basically what I was referring to in my last post. For example, if the rules being confused are "there must be an odd number of reds," and "there must be exactly one red," then I would make sure that groups with 3 or 5 reds are more likely to spawn. I still need to actually implement it, but that's the plan.
@Squidcor: I agree. To be honest, the explicit menu hints were still a work in progress that I implemented and forgot about. I hadn't yet figured out exactly how they were gonna work. So I've tried incorporating them into the status bar hints, but they only show if you're struggling. I think they're still a bit in-your-face obvious, but I think it works better this way, and now I can iterate on the hint system and improve it.
I've been thinking a lot about how to make level selection an interesting choice. I'm sceptical that throwing random obstacles in will help much. I'm worried about trying to group similar rules together, cause if you know these levels deal only/mostly with a specific type of rule, then it takes away a large part of the inductive/deductive logic, and IMHO some of the really interesting challenge. Grouping them by new "mechanics," like more shapes/colours, an additional attribute like size, or the length of the groups, could work really well, but I'm not sure how to fit that in with an infinitely-expanding grid, which I kinda like. Still wanna think about these, but after some thought, none of them really stand out to me.
So I thought a bit more and I came up with something I might try. Instead of each level being worth 1 point, if I make them worth 1, 2 or 3 points, and I connect this (indirectly) to their relative difficulty, then I give players a potentially interesting choice. Do you do a more difficult level that give more points, or perhaps do an easier level now that gives you access to more points for your next options? I made a small mockup to demonstrate:
The idea would then be to tie these points into an achievement system. On that, I think a limited completion achievement system, versus a perpetually-generating one, might work a bit better. That should give the player a relatively straightforward goal to strive for and hopefully make more of the choices interesting. If these two systems work well together, then I can tweak the time it takes to complete the achievements, and hopefully by the time the player has beaten them they'll either not need the motivation to continue playing, or would have had a nice experience and can move on.
I'm gonna give it some more thought, but my current plan it to implement the following next:
Thanks for all the helpful feedback so far, it's really gotten me thinking :)
That build will be at A MAZE so I thought I'd put it up for people that can't make it. I did a bunch of stuff, including a completely revamped UI, but not much of what I said I was gonna do ¯\_(ツ)_/¯
Some time next week I'll do a proper update where I outline what's changed and update the Flash version and the APK.
The "increased special node vision range" means you can see the special nodes from a lot further away. I think this will help a lot to make node selection more interesting, but only time will tell.
I don't really have anything new to ask in the way of questions, but I did create a small questionnaire I plan to use for some testing. So my only real question for here is: What do you think? What are your general impressions?
Android version: sign-up here: http://bit.ly/1j3CQXB, and install here: http://bit.ly/1KPxubw
Flash version: http://clockworkacorn.com/games/luminare/
I really like it, very induct-y! Especially started to appreciate the game when it got more challenging.
I'd like to see the early stages of the game being accelerated a little. I feel like the rigorous process for a win condition later in the game is justified, but I'd also imagine a smaller combo of five correct guesses would be enough for the simpler rules and just help people get past the easy levels a *little* more quickly if they're confident enough.
I also hope to see soon more feedback for how well I'm doing, would certainly like to be challenged to solve each level in as few steps as possible :)
Do you think some statistics like "avg moves per level" and "avg time per level" would satisfy your desire for feedback? Another idea that @Squidcor suggested was adding achievements. Do you think that might fulfil the same need for you?
Like you suggested, I've had plans to revisit the number of guesses you need to beat a level (I call this "the solve streak"), with a strategy to start a bit lower and increase as you progress. I think the trick will be balancing "once I knew the rule it took too long to beat" vs "I guessed a bit and got lucky." I can probably do some math to work out what the chance of getting lucky is, but the math is a little bit on the difficult side considering some of the opaque things going on behind the scenes... maybe I'll just try some numbers and wing it.
Something I'm wondering about is: how should I deal with save game retention from version to version? The levels are procedurally laid out, and although it doesn't happen every new version, it's not a super rare event that I want to make a change that fairly drastically impacts the levels. Currently I simply throw away the old state, but preferably I'd like to somehow let them continue playing. Any ideas to solve this issue? I guess the solution will involve thinking a bit more about what it would mean to retain your save state when most of the levels in the game world change, but maybe someone has a smart idea.
I noticed a little bug today if you reset the save state. I'll investigate soon, but I suspect this won't be a show-stopper for anyone :)
The issue is: let's say you complete level E5 in one version, but I realise that E5 is actually too difficult so I decide to remove it. Because of how I generate levels I wouldn't specifically remove E5, I'd figure out the rule group and probably move it upwards (to the more difficult region). But this has knock-on effects, so a bunch of levels that you've previously completed (perhaps all of them) are now gone... The issue is that even a small change has really big knock-on effects to the world.
I don't currently lose any data and technically I could let you continue on your old save, but I'm worried about diverging save states when I really want to have them all converge (ignoring actual progress differences). I can compare the save state you have with the new world state after a change, but I don't know how to merge the two together intelligently.
So are you saying that the generation of levels is contingent on the completion of previously completed rules groups in some sort of tree structure?
It's pretty easy for me to move a particular level from one location to another and reproduce it in its exact form with the same level seed (by hardcoding an exception), but It's a little difficult to do that if I still want to procedurally generate the levels - which I do want to do because they can extend infinitely in at least one direction. So I'd rather adjust the underlying distribution of rules mentioned above, but doing this makes pretty much all the levels change.
Are things a bit more understandable now? :)
You can currently "re-enter" a completed level, but only to look at the history. Are you thinking that if node E7 is completed in the old save, then mark E7 as completed in the new save, even if it would be a different level in the new world? I could do that, but then obviously you could encounter a different E7 to me. What if E7 was a particularly easy level for me, but not for you? I'd like it if everyone was playing in the same world so they can share and compare progress, but maybe that's just not compatible with my other assumptions and goals.
So to summarise the save merging issue:
1. World is procedurally generated, not manually manipulated.
2. I want to be able to make changes to the world.
3. Changes can have knock-on effects that make the new world indistinguishable from the old.
4. I want all players to be playing in the same world.
5. I want players to keep their progress when I make changes.
I can currently satisfy all of those except #5, but it seems like maybe I need to give up on one of the others if I want to satisfy #5, perhaps #3 (by using manual exceptions for changes) or #4. Dropping #1 is quite difficult cause of the infinite levels thing, and I feel like bargaining I can drop #2 is too risky. I guess the reason why I asked about this is I'm still hopeful I've missed something and someone has an awesome solution I can steal borrow :)
This problem seems like quite a deep rabbit hole ;) Is there not any way to just simply fake it?
1. You will need to give your different generation algorithms a version number and save this with the stage once completed.
2. You will need to keep all those different generation algorithms in your code so that when a player replays his old stages, you know which algorithm was used to generate it.
This way all stages the player hasn't played yet will use your latest version of the algorithm, but once a player plays a stage it locks in that version of the algorithm and can never be anything else. This does come with a fair amount of book keeping, especially if you add additional rule sets to future versions, but I don't see how that's avoidable.
I'm assuming you were looking for something a bit more elegant, but this is the best solution I've come up with to solve this problem.
The other solution is to just release it perfect and never touch it again :D
@NickCuthbert: your DAG idea sounds pretty interesting. It might make different areas of the world have very distinct feelings. It would definitely come with its own problems, but it would certainly be quite unique. Unfortunately I don't foresee me implementing that any time soon ;)
This problem is definitely a bit of a doozy. Totally open to non-elegant faking-it solutions, I just can't think of any that satisfy my goals :(
That said, I would be more strongly motivated to *return* to the game if there's some sense of how well I'm doing, or something a little meatier for achievement stuff, so I think I'd be very pleased to see such changes made (for now, I'm satisfied with just playing the game enough to see what it's about and give feedback)
Either way, I should have more time to play it over this weekend while I'm doing the airport dance ;)
One exception is that sometimes changes can mean that old levels are replaced with something other than a level, but I guess this will be a lot more rare and then I can just discard the old levels.
@Nandrew: Cool, thanks :) I actually already store some statistics, but they're not exposed to the player. I think I should look into doing that soon (probably a good idea to update the stats for analytics anyway). I'm also keen to try an achievement system and see if maybe that helps. My hope is that the achievements will get the player into the metaphorical deep end of gameplay, and then statistics might help keep them coming back for longer.
Something I forgot to ask is: what do people think of the Haiku system? Ignoring the quality of the Haiku themselves, do you think the system adds to the game and makes you want to play a bit more to unlock the next Haiku?
This game is like crack, can't stop, it mocks me.
I kinda agree with Peter's decision, but your are not very clear on exactly what you mean with worlds merging.
Remember to save yourself of future design problems by trying to keep old content.
Why not just take both Pieter's and Dislekcia's proposals and combine them then.
To fake it you can add extra "quests" to same levels graphically,
but I suspect you will open yourself up to a whole lot of extra design work in future.
So then you keep history any playable seeds (saves) like this:
Level 1 Seed A = World A seed + Level 1 seed;
Level 1 Seed B = World B seed + Level 1 seed;
You can even change both:
Level 1 Seed C = World C seed + Level 1 seed X;
This is what I did for the PCG with many levels, since each player has different world, but needs to save progress.
I save world + level as a combined seed.
Then all you have to do is add a new button and or result per combination since you may have multiple worlds.
All you need is the seed, and like Pieter says know what algorithms to use when you generate that version of the seed,
this way you can always on the fly generate it all. Maybe I'm missing something, but I do not see the issue, you just need to save more correlated info so you can translate it correctly on load.
To clarify on what I mean by merging/converging worlds: if I started playing on version 3 and you start on version 5, then I'd like it if we're basically playing the same game in the same world on version 6 and we've had the same experience. The technical aspect of keeping old versions of levels and content is basically already solved in the code base, it's the design problem that is the issue.
The current plan (still mulling it over before I find the time to implement it) is to keep only a single previous level state (the one you were busy with), inform you that it's changed in the new version, and then give you the option to replace it with the new up-to-date version and play that instead. The distinction here is that the world save states are converging towards the current global version of the world. This is basically what @dislekcia suggested if I understand correctly.
@NickCuthbert: Not a silly idea, I have definitely thought about that. I might try a super simple meta-game, but I'm loathe to design a whole other game on top of everything. I want the focus of the game to be the core puzzle solving experience. That said, I'm aiming to have about 50 Haiku, so it wouldn't be unreasonable to add in a few Haiku that are nonsensical/encoded or such, and lead the player on a very short ARG-like meta-game. That said, why can't the Haiku be the reward in themselves? ;)
Your worry that the Haiku system might be gimmicky just brought to mind a book I have been recently reading, which suggests that play is more meaningful when the outcomes of actions are embedded into the context of the game.
When I say meta-game, I was thinking more about altering existing mechanics. So for example you could have a special level where the game doesn't provide instantaneous feedback, just tells you how many inputs you correctly classified after x cases, with some of the Haikus containing training examples. As always, take my uniformed opinion with a pinch of salt though, considering that I haven't actually released anything. :)
I've now hit the challenge wall for myself, I think. Picking at various levels to see which ones I actually get through, etc :P
I'm wondering about giving the players an option to simply reset a level. It's not the most logical behaviour, perhaps, but I find myself occaisonally wanting to remove information clutter and start with a blank slate again when I'm struggling with a puzzle. It helps me feel internally refreshed and ready to think more clearly for Round 2
Some of the changes from the previous version include:
- Revamped save system. Unfortunately old saves will be lost, but hopefully not again. Also, you can now have up to 3 save slots if you wanna show someone else the game or such.
- In-game statistics - how long does it take you to solve levels?
- Some pretty artwork by @pieter
- A lot more juice
- Some basic sounds
- Many other minor implements and bug fixes.
Thanks for all the help from the community over the years, I'm keen to see what the reception is like :)