Riddles of Difficulty
Fireball: Structure (2)

Fireball: Structure

FireballtitleOne of my many minor goals for Fireball (our first 'verb game'; an abstract and original budget market PS2 puzzle/platform game) is to avoid using a separate front end and associated loading times and instead have the entire game constructed using only the game engine. This probably won't happen for a number of reasons up to and including possible future interference from Sony, but for the time being it remains on the roadmap of the game. Pictured here, for instance, is a placeholder title screen for the game. Just behind the word in English are the kanji for the game's Japanese title (Hidama).

Fireballtitleburning_1We are currently receiving our offers for the game from the various budget market publishers, and deciding which way to go - although one publisher is a firm favourite. The actual development of the game is on hold while we discover the scale of the budget, since we can't know how much we can get away with until we know what sort of advance we can get for it. However, level design is very much an ongoing process.

As regular readers will know, we have invited all and sundry to get involved as an External Level Designer, earning a game credit and a potential share of royalties. Everything you need to know about this can be found in an earlier post (follow this link!) except the minimum spec for the tools which is:

  • CPU: P4 2.8GHz
  • GPU: nVidia FX5200 with 64Mb RAM or comparable card
  • SYSTEM RAM: 256 Mb
  • OS: Windows 2000/XP

I'm delighted to report that we have had our first two levels from the external pool.

Fireballmousetrap_1Mousetrap is a classic fuse race, in which the player must negotiate a rat's maze of burning green 'Leaf' blocks before a stone cage traps the exit. It's one of those levels which looks hard, but turns out to have several solutions which make it very easy, which is exactly the kind of puzzle I'm happy to see in this game. It was the first level submitted by anyone in the external pool.

FireballhighaltarHigh Altar is a more cerebral puzzle, based in part around the long floaty jumps that your fireball can take in the game. It has a much more exploratory feel than the levels I have been building, as you must experiment with jumping from a number of different places in order to crack the mystery of the level. It requires some lateral thinking, but the player in principle has as much time as they need to work out the solution.

As delighted as I am by both of these levels, it's clear that we could use some more level designers. In particular, I could use some people working on levels that are simply fun things to set fire too and contain no significant degree of challenge. If you are interested in getting involved and meet the minimum spec above, follow this link and check the section at the bottom of the post for more information.

The original structure we had planned for the game was to have a Spine consisting of about 60 levels which anybody could complete, but which have to be completed in sequence, and also have a Collection of levels which would unlock automatically as the player hits certain 'Ash' totals (number of blocks burned) - so that new levels will be unlocked over time, as well as through success in the Spine.

Recently, I've been thinking this isn't a very sensible structure - making the Spine easy and hiding harder levels in the Collection is all well and good, but it makes more sense to make the Spine harder and the Collection easier. After all, some players will be fiero seekers (thriving on challenge) - they will not (in principle) want to advance without winning, so the Spine structure (linear sequence; beat to advance) makes more sense for fiero-seekers (Type 1 Conqueror by DGD1). Conversely, experience-seekers (Type 3 Wanderer by DGD1) should be able to get to new levels without having to struggle - so it makes more sense to have the easy levels in the Collection, which any player can unlock by playing the levels available over and over again to score more Ash.

FireballhexagonThis has lead me to a new idea for the game's structure, as shown in this illustration. The player selects the level they are going to play on a field with a black hexagon drawn on the ground. The player starts inside the hexagon and sees three 'paths' stretching away from them consisting of linear sequence of objects which can be burned. Setting fire to an object on one of the paths is how the player begins playing a set of levels.

When the player goes outside the hexagon, they will necessarily be in one of three different sections, according to which 'path' is closer. At the top of the screen will be displayed the name of the path they are about to choose from: Challenge, Puzzle or Fun.

The levels in the Challenge section will be constructed with a bias towards fiero (i.e. more challenging, more tightly balanced towards challenge) - and the player must beat each set of levels to advance to the next one on this path. This would be like the old Spine structure.

The levels in the Fun section would be much less difficult, and focus primarily on just giving the player fun things to burn. There may be some puzzles, but they'll be very simple. As a guideline, your grandma should be able to beat all the levels in this section, in the best case. These would use the progression rule from the old Collection structure - that is, new levels are unlocked as the player acquires Ash, which they can get from playing any level (and, incidentally, they get more Ash for doing particular well in levels, so the player can work on earning medals in levels they've already unlocked to earn Ash).

The levels in the Puzzle section will be focused more around problem solving (Type 2 Manager style puzzles) - perhaps even the objects that are burned to access these levels can be 'micro-puzzles'. I don't yet know if this should have a Spine type structure (beat previous levels to advance) a Collection-type structure (you will eventually advance if you keep playing) or a hybrid structure that can advance either way. I'm swaying towards the latter.

In this way, we can provide about 60 different levels in three different 'game modes', all selected from this single area which in itself is a game level. I expect there will be some reuse of levels - so that each of the three paths consists of at least 60% original material, and 40% levels which recur in one or other of the other paths. (Minimum number of levels required would therefore be 36x3 = 108).

The idea, of course, is that we are then providing separate game paths for three different types of players; three of the four DGD1 archetypes. The fourth archetype, Type 4 Participant, should be catered for by a pad-passing Versus mode, accessed from a special object inside the hexagon. And of course, players are free to go on and try other modes after they've completed their own, if they like. Indeed, the Fun path will be unlocking whichever path the player is actually following.

I welcome people's input on this idea, and would like to know whether people think the Puzzle path should be Spine-like, Collection-like or governed by some other kind of progression mechanic.

Incidentally, if any game journalists would like to cover the game in any fashion, please get in touch with me by email.

I'll be posting more about Fireball as the project progresses.


Feed You can follow this conversation by subscribing to the comment feed for this post.

Forgive my ignorance, but what would be the difference between the Challenge and Puzzle modes? What type of "challenge" are you planning to give the player, if not a cerebral one? A challenge for the reflexes?

The whole proposed "something-for-everyone" system sounds thoroughly nonsensical to me. This is more than likely due to sheer ignorance. In any case, maybe you could clear up my confusion:

How can you expect any player to know which path is optimal for him?
Moreover, how can you expect any player to be such a perfect representation of the DGD1 classification system that he wants to follow one type of progression, and only one type of progression, for the entire game?

How can you create a successful difficulty curve with "collection-like" progression?

I imagine that a really hardcore player would play through the challenge mode, and then want to finish off the other 48+ levels in order to have truly completed the game. If he finishes the hardest levels in the game, and only then plays through all those levels which are back on the lowest rung of difficulty, would that not leave him with a sour taste? (especially considering he may need to pass through many levels he's already played if Puzzle mode is Spire-like)

In Fun mode, "new levels are unlocked as the player acquires Ash". Are you suggesting that the casual gamer will have to play the same levels over and over again to start unlocking anything?

If 60 levels, many repeated in the other two play modes, are easy enough for my grandmother to beat, and the other 48+ levels are split between only Puzzle and Challenge, then how can Challenge be challenging at all?

What, exactly, is "a pad-passing Versus mode"?

I have other questions, but chances are your answers to these questions will clear up all my misunderstandings.

Or do I sound like a pointless whiner? Really, I'm just curious- it's been a long time since I've heard of such an out-of-the-box structure.

Note to those wanting to work on Fireball: small levels can run in the editor and alpha game with a lower-end computer. My laptop is a 1.4ghz with half a gig of ram, and it can handle levels with less than 200 blocks. Grandiouse levels with hundreds and even thousands of blocks (like some I've got planned) will require a high end machine with specs meeting or exceeding the above numbers.

As for structure, how about this approach: the fun spine is made up of six to ten level lists consisting of six fields, these levels are all available from the start. The Puzzle levels will be assorted into field lists as well, with the lists containing more complex content having a higher ash total to unlock. The Challenge levels will follow suit, with the successively harder lists requiring a higher medal count to access (you could have silver medals worth 1 point and gold medals worth 3 or 5).

The above scheme imposes linear structure on the player only at the local level, requiring feilds to be played in batches of six. The player will begin with a host of fun levels available, and will readily total high ash counts as they raze levels like Chris' representations of forests and towns and parks. Maybe three or four of the Fun feild lists could have high ash total requirements, to stagger the play of the Puzzle and Fun lists. In the Puzzle lists in particular the player will have the opportunity to figure out the highest possible combo counts, if they so choose, enabling the challenge levels. Of course a handful of silver medals will happen for even the most casual player, so a taste of the Challenge fields will present itself to the lower end of the skill pool, possibly incurring improvement and further investment. Whats cool about this scheme is that it tends to adapt to the player in terms of sequence and difficulty.

Questions questions...

In answer to the difference between Challenge and Puzzle, cerebral puzzles (which you can take your time solving) can provide fiero, but a time-critical jumping challenge is a more likely source. The Type 2-Puzzle path is for people who don't particularly want the time pressures.

"How can you expect any player to know which path is optimal for him?"

There may be a misunderstanding. All three paths exists simultaneously: you don't pick one and then end up stuck with it. So the answer to this question is that I expect the player to experiment and find what they most want.

"How can you create a successful difficulty curve with 'collection-like' progression?"

Very easily. The game design contains 7 blocks. (Possibly 8 in the final version). The addition of each new block increases the complexity of play, and these blocks are paced throughout the paths. Therefore, there is an inherent difficulty curve to the play of all paths.

"I imagine that a really hardcore player would play through the challenge mode, and then want to finish off the other 48+ levels in order to have truly completed the game."

Yes, definately. I forgot to mention that the medals the player earns from doing well in levels may also unlock levels - so such a dedicated Type 1 Conqueror style player would still have challenges to find. I anticipate, however, they would probably not work only on the Challenge path, but rather complete all three paths in parallel to some degree.

"If he finishes the hardest levels in the game, and only then plays through all those levels which are back on the lowest rung of difficulty, would that not leave him with a sour taste?"

The fun levels will be easy to complete, but they might be quite hard to earn medals from. (I appreciate the whole medal issue was not adequately discussed in the original post!)

"In Fun mode, 'new levels are unlocked as the player acquires Ash'. Are you suggesting that the casual gamer will have to play the same levels over and over again to start unlocking anything?"

Well, yes and no. I am suggesting that if the Casual gamer becomes stuck after a while, they can *still* progress by playing levels they've already played. The gearing of the Ash targets will be critical.

This also encourages the player who is focussing on the Fun path to dabble in the other paths, in order to score some extra Ash.

"If 60 levels, many repeated in the other two play modes, are easy enough for my grandmother to beat, and the other 48+ levels are split between only Puzzle and Challenge, then how can Challenge be challenging at all?"

Try looking at the numbers this way. Challenge mode consists of 60 levels, 36 of which are unique to Challenge and 24 are reused from Puzzle and Fun. Let's suppose there are 10 from fun and 14 from Puzzle. This means that each block of 6 levels is allowed 1 relaxing "reward" level for each 5 challenging levels. This won't undercut the feel of challenge significantly.

However, I should stress those numbers represent the *minimum* number of levels we'd need to complete the game. Best case, all three paths will have 60 original levels.


I'm not sure about giving the player 6 of the 10 field lists in each path 'for free'. But I do think that setting Medal targets for the later Challenge field lists is a great idea. Or do you mean that they are unlocked sequentially? On reflection, I think this is what you mean.

You're advocating that the final Puzzle path field lists be unlocked from Ash values... I'm not sure about this. If both Fun and Puzzle unlock from Ash, either:

1) They have the same gearing, and are unlocked at the same time
2) Puzzle has higher gearing than Fun, so that Fun levels are unlocked before Puzzle.

There doesn't seem much point in Puzzle and Fun being unlocked at the same time. But should the Puzzle player be given a longer game than the Fun player? I'm not sure myself!

Thanks for the input - I definitely think using the Medals to gear the Challenge mode is better than a seperate batch of bonus levels unlocked by Medal.

Thanks for clearing that up. Yes, it was just a lot of misunderstandings.

I don't like the idea of making Puzzle Collection-like. You wouldn't want the player giving up halfway through: "Rats- I can't figure this out; no matter, I'll give up, and get past it by collecting Ash." Better to give him a dead end, to make him stop and find the solution. He'll thank you later.

As for the statues, I'm not sure whether it would be a good idea either, but if used they would have to be for the late Puzzle levels. They'd have to be hidden somewher in the other two modes, to give the Puzzle players an incentive to play them. Of course, they'd have to be placed in the most devious spots, so that only the best at lateral thinking could find them. Say, there could be a portion of one level in Fun which is nearly identical to a portion of a different level in Challenge, and only by comparing the two would the player figure out where it is. Stuff like that. But why make it a statue? Why not a good, old-fashioned warp?

I agree with you that the Puzzle path should not be mediated by Ash totals. I think the most consistent solution is to have the player complete each field list to get to the next, as you suggest.

Why a statue? First noun that came into my head, to be honest. :) Anything would do. I'm not sure I want to go down this road, though.

Oh, wait, of course that would allow the players to break the order! No good, no good...

A colored key! Not very original, sure, but straightforward, and if you get it early you'll be intrigued enough to hang around to see what it's for. Say there are five keys, corresponding to five doors which you find after 54 levels of Puzzle. Each door has a cryptic hint next to it. You are only allowed into the final level once you've unlocked and beaten those five levels, and of course the final level would be the most puzzling of the entire game. The only question is how to devise a puzzle so tricky that it is still impressive after a lead-up like that.

I'm excited by your recognition of and desire to appeal to all different types of players. It seems to me that it should be possible for the game itself to determine roughly what kind of player is playing Fireball by keeping track of a few statistics: how much time they spend in a challenging level, how many blocks they can get burning in a fun level, how many tries it takes to complete a management level, whether the player prefers to make several attempts at a difficult level or plan it out and execute it successfully the first time, etc. If the player gets a lot of blocks burning in a fun level, but gives up quickly in a challenging level, for example, then clearly that player just wants to burn something down. By analyzing the player's patterns as the game is being played, the game should get an idea of that player's playing style.

Why is that useful? Because then you have your progression dilemma solved. Give the player what they want, but increase the reward for the levels they aren't so comfortable with. For the above player who is good at burning things down quickly, give them quick access to the fun levels, and reward them only a little for a good performance, but jack the rewards way up for a timing or management challenge. That way the player has two simultaneous reasons to play: to satisfy their basic strengths, and to get the satisfaction of stretching their abilities for greater reward.

This model can be tweaked to apply to as many different playing styles as you can identify. And the more stats you keep track of, in game, the more styles you can identify. Ideally, after the player plays a few levels, the game should know what their strengths and weaknesses are and be able to adjust the reward systems accordingly.


What you're talking about, Sam, is the basis of a player metric system used for pattern adaptation control, which combined with a robust AI model and recombinatory content, would allow for a game that tailors itself direclty to the user based on early-game cues. The elegance of Fireball's path design is that it enables this, to a degree, to happen automatically. I imagine a portion of the audience for this game won't make it past the third or fourth challenge list, or that some people will ignore the later puzzle lists, ect. Instead of using a lot of complexity and advanced modeling of the design to customize the experience (this is an approach more useful in games with HIGH verb counts, instead of like, two) Fireball lets the player to the customization everytime they choose what list to play. Difficulty balancing, instead of happening in real-time with an AI happens by the player's choice, so if you want an easy time, roll over the Fun path, if you want a fairly easy, thought provoking time, roll down the Puzzle path, ect.

Basically, the reason Chris went with the simpler form of PAC, the player mediated one, is that Fireball is being produced on a comparitively short development cycle (roughly 8 months) with a VERY low budget. As such non-adaptive content produced for royalty share by a disparate band of rag tags (me included) is the order of the day.

Down the line however, I definetly intend on applying what you're talking about to more complex interactive environments, like storyworlds for instance.

Patrick pretty much covered my answer here. :) What you're proposing, Sam, is something that the games industry as a whole could use a lot more research on. Sadly, this game has too tight a budget to afford to explore it. My experience is that player metrics such as these are terribly difficult to assess in situ, though. I'd stop short of saying it's not possible - just that it's suprisingly difficult to achieve in practice!

Thanks for your comment!

"What you're talking about, Sam, is the basis of a player metric system used for pattern adaptation control, which combined with a robust AI model and recombinatory content, would allow for a game that tailors itself direclty to the user based on early-game cues."
Could you please give me more information on what these terms mean? I'd like to learn more about this area of game design and eventually attempt some related experiments, but I don't have a lot of technical background and I'm not up to speed on current developments.

Patrick could use to add a glossary to his blogsite. He uses a fascinating array of terms, but their meaning is not always immediately apparent. :)

Yeah, since I figured out how to code sidebars in blogger (which is really easy, go figure) I've been thinking of ways to make King Lud IC more accesible to the theory I've been developing. I'm planning on writing my Senior thesis on this very topic by the end of th year, but in the meantime a few explantory articles would do.

Thanks for the tip, both of you.

Verify your Comment

Previewing your Comment

This is only a preview. Your comment has not yet been posted.

Your comment could not be posted. Error type:
Your comment has been posted. Post another comment

The letters and numbers you entered did not match the image. Please try again.

As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments.

Having trouble reading this image? View an alternate.


Post a comment

Your Information

(Name is required. Email address will not be displayed with the comment.)