I apologize if I come out a bit rude.

Re reading myself i think it's a bit harsh, you are still doing a great jobs! Also it's not you, you are reporting what you find interesting and it is still insightful for most.

No, I mean you have a point too. Firstly, I'm no game designer and there are many things I don't know! But hopefully that naievete helps me write the pieces a bit better

But secondly, and more importantly, there are people out there that have had these ideas already and academia may be repeating their work (I don't think they are in this case, but still) and the point of the Saturday Paper is to try and get the word out more so people can share ideas and build on each other's stuff instead of reinventing the wheel.

I'll take your suggestions onboard, and thanks for reading as always

Sorry, I might be misunderstanding your post here but to me it's not clear what your objection is to the piece. Is it a problem with the paper, or my article?

EDIT EDIT - Nevermind.

This is about building a system that is flexible enough to construct simple puzzles from a database of basically distinct and separate concepts. It's a simple idea but I like the use of planners and the potential for more complex and clever systems to be built on top of this:

http://www.gamesbyangelina.org/2013/05/the-saturday-paper-use-new-ideas-on-genre/

Cameron Browne is actually an RA in my research group! It's a nice thesis but a bit hard to summarise. I might cover something similar one week though.

You should have a maintained list of all procedural game paper

That's a nice idea, but there are simply too many! Plus there are sites that attempt to do this already, including the conferences themselves. I prefer to pick out ones myself and go more in-depth

The next paper goes up tomorrow on puzzle generation for adventure games. I hope you'll all let me know what you think.

Hi all!

Just to clarify a few things - The Saturday Paper is a regular fortnightly column I started on my site where I discuss other people's research ( ) into game design and related topics. I'm trying to highlight work you might not have heard of, but that could be useful to you right now!

So this isn't my work, but it's work that I think is cool and want to give more exposure

Hey thar! I wrote that post (but am not the author of the paper, obviously). I just wanted to chime in and say the major contribution is showing how answer set solvers can be used to solve this kind of problem. Obviously we can hard-code a system that, for instance, can place keys before locks by considering paths. But this applied answer set solvers in a way that was difficult to do previously. It didn't really come across in the piece I wrote because I couldn't go too in depth without taking too long, but I recommend the full paper as a read sometime!

I like the look of the Rogue Souls project! Good luck

Hey,

I started a few mini-game projects to practice pixel art and animation - my first time for the latter! I'm making a game based on this photograph: http://www.guardian.co.uk/pictures/image/0,8543,-10505044516,00.html

This is what it currently looks like:
http://imgur.com/a/IjhYW

The sun slowly sets, a few animations for the flamingos, etc.

I'd like to add in a few distance scenery pieces. Lake Makat is ridiculously beautiful and includes some stunning far scenery:

https://en.wikipedia.org/wiki/File:Ngorongoro_Crater_Panorama.jpg

But this is noticeably harder to draw than small things like the jackal or the flamingo. There's larger spaces of colour, less detail, and generally bigger things to draw. Does anyone have any advice on drawing larger, more natural objects in pixel art, or how to 'graduate' to larger drawings?

A game coming up with random combinations of class abilities or even different parts of abilities would never be balanced.

It seems to me quite plausible that a system could analyse classes for balance. Why not?

"Ridiculous" is very dismissive.

I develop (and love) games in my spare time, but I'm also lucky enough to have a day job that involves them too - I work as a researcher looking at procedural content generation and game design! Woo!

I want to share the cool work going on in universities around the world with the rest of the gamedev world, so I've started a series of blog posts, once a fortnight, highlighting some interesting work and giving links where you can find out more. Here's the first, about an RPG that invents new skills for its players:

http://www.gamesbyangelina.org/2013/04/the-saturday-paper-generating-new-rpg-classes/

My hope is to get new ideas into communities that might not normally see them, and get people talking and collaborating! Let me know what you think about the post and if you have any suggestions. Thanks!

*Feels compelled to make game using these*

TIGS Comp. Assemblee Part Deux: Part Deux.

Hello! My name's Michael Cook, and I make and research games. I've posted an introduction before but I've changed focus a bit since then!

Anyway. Here's what I look like holding an invisible basketball:



If you prefer me to be holding a bigger basketball, there's always this photo? My point is, people take photos of me while I am gesticulating wildly. This is not my fault!

I work as a PhD student at Imperial College in London, where I'm looking into AI techniques for autonomous game design - software that can make games without humans getting involved! I'm building an AI called ANGELINA that tries to design games with as little help as possible, but because I work solo mostly, and because I find myself coding lots of little template games to help ANGELINA publish its results, I feel a bit like an indie sometimes! Certainly, the indie community has been very kind in welcoming me in and talking about my stuff.

I do make games just on my own too, but they're not much to speak of yet.

If you'd liek to know more about ANGELINA, you can check out the website or just drop me a message!

A Puzzling Present is now available for download!

http://www.gamesbyangelina.org/downloads/app.html

In January I'll be open-sourcing it. The Google Play store is still indexing it but you can grab the APK from the downloads page too.

Getting closer to launch! I haven't updated because, frankly, I"ve been working non-stop. This is the first time I've worked on a game that's releasing 'for real', as it were, and coupled with the uncertain nature of ANGELINA and the research-side of things, I'm pretty stressed out!

We got some wonderful art through from our artist, Harriet Jones, though. Here's the app icon:



She was great to work with and produced really high-quality stuff. She's really eager to work with indies too - highly recommend her!

This looks so great, the concept is brilliant but the art is what really sold me. Good luck with it!

So happy to see this released! Had incredible fun on my first run. Ate an unidentified rune as a panic move. Polymorph! Chuckled for a good five minutes after my corpse was strewn across the dungeon.

Congratulations!

Updated the post with a non-404'ing link! That always helps.

[Download here.]

Hey all,

Last week I posted about a game called AMBEX. I said I'd made it in a couple of hours based on some suggestions of words on Twitter. I did it again this weekend and made NOSE AND TAIL, a five-level puzzle platformer in about three hours. The source and art are public domain on GitHub, but you'll need LibGDX and flixel-android to compile it. More on the site itself.

Download here.
Source on Github here.



This time I asked for two careers, a noun and an emotion. They were a little loosely inspired this time but it still works. I had a basic idea (asking for two careers because I wanted to make a game about swapping between characters) but the randomness of the suggested words gave me new ideas (like swapping costumes at particular boxes, which would've been action figure boxes could I actually draw).

I really like the idea of asking for four specific words and then making a game in one sitting (or a short period of time). It's more flexible than a group jam, and the short time periods mean you focus on making anything - which is teaching me a lot, like what code I use the most, what things I've never tried before, and so on.

Instead of creating a thread for every game I make, I'll edit this one and put a new post in. I hope to do more of these in the future. All feedback appreciated!

Today I made a game in two hours about hitting and climbing into exploding toasters to propel yourself to victory.



Made with LibGDX and flixel-android (which is a wonderful library).

Source here: https://github.com/cutgarnetgames/ambex
Zip here: http://www.cutgar.net/alphas/ambex.zip
flixel-android here: http://code.google.com/p/flixel-android/

It's really just a few levels long, and was made to clear my head from work. I asked for two nouns, a verb and an emotion. I got toaster, kebab, exploding and ambivalent. I had planned to add sound effects monotonically narrating every action ("boom") but didn't get aorund to it.

Although it was only a joke, I kind of wanted to use the mechanic for something (you can see in the third and final level that I tried to implement a resource mechanic but failed). I'm mostly posting it here because it's open sourced and therefore a working (but crap) example of LibGDX+flixel-android.

I might extend it, but probably won't.

Thanks!

But that's not new, it's call rubber banding, generally it is adhoc, but you could create an AI that watch and generate events/rules/element to counterbalance. Actually the stake space is a mirror of the flow graph everybody is referring to in game design. Progression toward losing is the difficulty axis, progression toward winning is the skills axis (But I replace skills with stimulation, sex and watching a movie can put you in flow without real skills needed).

Okay, but you're assuming the game is adapted in realtime. The aim of ANGELINA is to produce a game as a human would produce a game - releasing a piece of software that is complete, and does not change. ANGELINA stops contributing to the game the second it ships. So I can't do live adjustment.

ON party in a RPG, think in term of gameplay not character, every character have a HP bar, losing distance is the cumulative sum of these HP bar, as simply has that. Now if you have a heavily classed party, not only cumulative HP is the global losing state, but each independent HP is a distance toward losing a particular option, losing your healer is losing the ability of recovering from losing, and also you have to manage between which option to support with your healer, tank or dps? both are in need! who do you prioritize slowing enemy progression (tank) or keeping your momentum (DPS) ...

You say that losing your healer is losing the ability to recover from losing - but this is loaded foreknowledge of how an RPG works, which (again) is something that separates what I'm trying to do with ANGELINA from some of the work by people such as Joris. If we clump everyone's HP into one bar, I like that idea, but it doesn't capture how important losing the healer. Also, everyone losing 1/4 of their HP is not the same as one person losing all of their HP. We can't bring in too much of our personal knowledge about how the genre works, because that blinkers the AI and doesn't let it understand it on its own personal level.

I think there's lots of work to do. One thing I need to do is write better solvers-  even the platform one leaves a lot to be desired right now.

More screenshots!

From my screwed understanding of AI and GA, I see the fitness function as an "emotion"... Also this fitness function seems to really "care" about solvability, that's why, that's its "personality"

Yes, that's a good observation! I guess that is the extent of its personality so far. Hopefully in the future we can build in new models for things like enjoyability, or mood. The reason why I've avoided it so far is because I like to avoid hardcoding in my own concept of fun or whatever. I'd rather try and discover a way in which the AI can express or discover this itself. Obviously this is crazy and very difficult, and so requires some thinking. Solvability is simple compared to fun!

Something that would fun is to run a Dijkstra like algo to annotated the level (progression, critical path, alt path, branchyness, rhythm) and look how it evolve as the fitness function increase.

Yes, I like that idea. More metrics like this might help us understand what is considered a good level for a given mechanic. Hopefully A Puzzling PResent will include some surveys that can give feedback on level quality, so we can see which fitness functions and evolutionary setups produced the best levels.

Another cool thing would be to have an "originality" rating, for example

This is a really cool idea! It's actually a known concept in GA I think (the idea of evaluating solutions based on their distance to other solutions) to help artificially drive variation in a species. I should definitely look into this - thanks for suggesting it!

On spell:
http://forums.tigsource.com/index.php?topic=29145.msg807484#msg807484
Here is a deconstruction of battle mechanics I made.

Added to my reading list, thanks!

What your system can't find as a mechanics (I think maybe I'm wrong) is indirect mechanics.

You're absolutely right, your example of the switch for instance. I include examples of this in my paper (under review right now) and say that hopefully in future we can extend the system so it can create its own classes when looking for mechanics. So it might, for instance, produce a result that say:

> create a new class called Class1, extends FlxSprite (FlxSprite is Flixel's basic game object)
> in update method, call isTouching() with argument Player. If true, set flag F1 to true.
> in the Exit object, add constraint if(F1) before the exit logic

This is really generalised and pseudocode, but my point is: right now the system can only modify fields. However I'd added functionality for it to generate constraints, and in future hope to add things like method invocation and class generation. It's complex stuff though, as I'm sure you can imagine.

One great thing the system can do right now is find emergent mechanics. That is, I can say:

> Here is Mechanic M1. Here is Mechanic M2.
> Generate me a level that cannot be solved with JUST M1, and cannot be solved with JUST M2, but can be solved with both.

This might search for a subspace where M1 and M2 are used together.

The basic space is your HP (failure) vs the HP of the enemy (winning).

I think this is a good starting point but there may be other factors. For instance, in party-based RPGs: how many people survive? In RPGs with complex movement and positioning: what strategies do we use during simulation to determine where the AI and players move? This is particualrly hard because the new spell might open up an optimal strategy that we aren't testing for because it's not optimal unless you have this spell (if you see what I mean).

It can be even more crazy when you realize that game rules are "just" permission of movement in that space! Okay I stop right now ... but it mean your current take can already be tweak to cover all of this. Basically a complete arbitrary game is a space in which both winning and losing are possible. But a good game modulate the progression between the two to maintain "tension" (yay I won't develop this one here, flow theory and the kind yaddy yadda). And that's where Joris Dormans works on emergent design (actually internal economy) came handy, it's all about those modulation!

This is actually a very interesting way of thinking about a game. I like this a lot - really got me thinking. Thanks again!