Strange Design

What is the goal of hidden content?

I "finished" Super Mario 3D Land a few days ago, making my way through all 8 worlds and defeating Bowser at the end.  Each world had 5-6 levels in it, ending with a boss fight.  But in world 8, after defeating Bowser, I was taken to a special "extra" world, where I was able to play two more levels and then defeat him again.

Secret levels!  Awesome.  The game had shown me from the very start that there were 8 worlds, so getting a few extra levels in a 9th world felt like a bonus.

So then, after defeating Bowes a second time, I won the game properly.  Credits rolled.

And then... 8 MORE worlds popped up!  I'm only half way through the game's content!  But this time, instead of feeling excited, I felt mostly cheated.  I thought I was going to be done with the game - able to pass it off to my kids.  I thought that the game was a bit easy, and short, but I was willing to accept that and just feel as though those impressions were a result of my extraordinary skill.

One sort of secret levels got me excited, and another caused me disappointment.  What was the difference?

Waging Expectations

The most significant difference, obviously, was that minor adjustments to my expectations help to keep experiences fresh & exciting, while huge revisions to my expectations make me feel as if I have been lied to.  

It can be really hard to anticipate what a player's expectations will be - is it "expected" that Dr. Wily will have extra levels within his castle?  Is it "expected" that Mario will have the ability to warp between levels?  This is an important role that maps serve - they create expectations of how much time / distance separates us from our goals.  Adding things on the map can be exciting, but changing the map itself is probably a no-no.

I'm struggling with this same issue in my professional life - how much content should I give away for free, and how much should I give to my paying customers?  In this case, I'm actually trying to create the opposite feeling - I want free players to feel as if they have experienced most of the game - but then surprise them with the revelation that there is a lot more for them to learn and experience.

GEB

Discussions of hidden content often lead to a discussion of books - because it's hard to have "extra content" in a book.  You can see how many pages a book has when you pick it up, and you can see the typeface & how many pictures it has at a glance.  One book has successfully pulled off this trick however - the fantastic "Godel, Escher, Bach" abruptly ends about 60% of the way through, and then starts a new book completely.  You "finish" the main discussion, with the characters and goals you are introduced to, long before you run out of pages.  The book even hints at this experience during the discussions of incompleteness, twining, and complexity.  But that's probably just the exception to prove the rule.

Defining Game Balance - Part 2:

Two days ago, I put forth the following definition of Game Balance:

1 - To be balanced, a game must rely on one or more economic systems for which both positive and negative transactions have significant Entropic Force.

2 - The resulting game system must be complex enough that a majority of the transactions lie on one or more of several shortest distance paths between start conditions and victory conditions.

Reactions have been remarkably cogent, which I appreciate.  I promised that today I would discuss the consequences of accepting this definition, but I'm going to start by explaining why, exactly, I'm taking the time to explore this issue.

For 14 years I have been a professional game developer, specializing in systems design.  I play a wide variety of games, and I play them frequently, but when I talk about games I am almost always talking as a developer, not as a player.  So when I talk about game balance I am not looking for a way to examine the potential flaws in an existing game - I am seeking a way to ensure that my next game will be balanced.  Before release it is possible (and a very good idea!) to get feedback on the product, but you simply cannot get the same breadth of feedback which the commercial release will receive.  For this reason, I am not interested in balance discussions which can only take place after release - except insofar as they offer insights into how one might avoid similar problems in future products.

Balance is Not About Winning or Losing

You will notice that in my definition I specifically do NOT say anything about how often a given player will beat another, (or beat the AI) - or how successful players of equal skill might be when playing against one another.  It is very possible (in some odd circumstances even common) for strategy A to beat strategy B consistently, even in a well-balanced game.  All I require is that there be at least one strategy which can frequently beat strategy A, at least one strategy which can frequently beat THAT strategy, and so on until you eventually reach strategy B.

To make this point another way - Just because you love to choose Rock every throw doesn't mean you should win 1/2 the time.

Balance is a property inherent to the System

By my definition, game balance does not relate to empirical data.  This means that local metagames may skew systems out of balance by imposing house rules, or applying social stigma to some strategic choices.  I personally have applied some of these balance - altering local rules, and I probably griped about how the game didn't work very well the way I was trying to play it.

Those of you who read my blog regularly will know that I'm a big fan of catering to your players, rather than resisting them.  So I don't mean to suggest that players are foolish and designers are clever!  It might very well be that the elegant balance you create for your game does not translate into happy customers, or brisk sales.  That's a hard pill to swallow, I can say from experience.

Symmetry is not Balance

One might suspect that perfectly symmetrical games would be the epitome of balance, but this is not at all the case.  Perfectly symmetrical games offer all players the same choices, but it does nothing to guarantee that those choices are reasonable.  Many well-balanced games are symmetrical, but the relationship does not obtain in reverse.  Perfect symmetry does tend to lead to even chances of victory for all players, which in some modes of thinking is equivalent to balance.

Balance is not defined by the Strategy you Talk About

Now let's look at the inverse situation - games which frequently end up in one of several well-understood strategic situations.  If I can make accurate predictions about how a game will develop, does that mean that player choices are dictated by mechanics, and thus a source of imbalance?

Maybe.

The real question is - are the predictions / situations which you predict an overwhelming factor in the overall game systems?  If they are not, then the frequent appearance of those situations does not indicate a break down in the game's balance.  For example, excellent Age of Empires 2 players will reach the Castle Age at right around thirteen minutes.  Does this similarity between games mean that all games are the same?  Certainly not.  If sports fans predict the particular play their team will attempt does that mean that the success or failure of that play is already determined?  No.  In both those cases there are dozens or hundreds of other factors in play.

If I can predict that my opponent will always choose to throw Rock?  That's a different story.

Building a Balanced Game

The most important question begged by my definition of Game Balance is, in my opinion : "How can we build a game such that we avoid imbalance?"  I have for your digestion several steps:

1 - Base your game around an economy, and describe your systems in economic terms.

Most games can easily be described in economic terms - but many designers don't think that way during the development process, and thus miss lots of easy opportunities to impose balance.  What are the currencies which matter?  What actions allow conversion between currencies?  How do those conversions compare?  Your design team needs to have a firm grasp on the economic underpinnings of their own game in order to anticipate and eliminate imbalance.

2 - Exhaustively playtest ALL of your positive and negative transactions.

Designers are players too, and players often gravitate towards a particular style in a given game.  Similarly, I've seen plenty of development teams overlook some obviously imbalance economic transactions, simply because they didn't fit the playstyle of any of the developers.  Finding an exploit you never tested is almost impossible.  Once you have defined your economic currencies, you can identify all transaction points, and make sure they get individual attention from your team.

3 - Measure Entropic Force for each transaction in economic terms.

As many people have noted - the concept of Entropic Force is central to my notion of balance.  In most cases, the true cost of a decision is not the price you pay to make it, but the price you pay to change your mind later.  The magnitude of the entropic force your game systems exert can vary wildly, but it is in examination of these forces that many potentially unbalanced systems will reveal themselves.

Let me give a specific example here which is near to my heart - the skill tree in Titan Quest.  The individual skills were carefully balanced (we'll assume) such that each one added a reasonably equal amount of power to your player.  That sounds like a good basis for game balance.  But the Entropic Force for some skills was incredibly high - specifically those skills which could be used only if you built your entire playstyle around them.  Selecting one of those skill sets meant that you could never change your mind, or employ new strategic options based on the game situation - your skills focused you in a single direction.  Furthermore, since your strategy was now defined, future skill / equipment selections were rigidly determined.  Eventually, this meant that all specialists were functionally identical to one another, and thus many skill specialists became too strategically predictable to fight competitively.

Were the skills too weak or too strong?  No.  Were they unfun or uninteresting?  No.  Their flaw was that they exerted an Entropic Force on players far in excess of other skills, eventually creating a system-wide imbalance.

4 - Complexity aids balance, rather than undermining it.

For some people, simplicity is the best test of balance.  I specifically embrace complexity as a prerequisite for balance because I believe that complexity is a necessary precondition for strategic depth, and that there is no need for concepts like "game balance" without that strategic depth.  Finding and tuning balance is all about creating unpredictable, interesting design spaces for players to engage in.  Exhaustively testing every possible interaction is not only unreasonable - it defeats the purpose of balance.  If I want to prove that gravity is constant everywhere on the earth, I should NOT have to physically measure the gravitational constant at every individual point, despite what formal logic might suggest.

My greatest source of inspiration in this matter is human social interactions.  Social interactions are incredibly (perhaps infinitely) complex, and yet stable patterns emerge from them all the time.  That is my model for what an elegantly balanced game system can accomplish - it creates interesting, self-sustaining patterns without collapsing.

Defining Game Balance

My proposal for this year's (2012) GDC was titled "Six Equations to Balance Every Game."  I had a pretty hot list of equations, which covered a pretty broad spectrum of systems.  I think many people are afraid to talk about game balance, because the term is used so broadly that it often seems pretty devoid of meaning.

But I believe that without any sort of complex wordplay we can squeeze a really useful, sensible meaning out of the word "balance" and then use that term to ask some meaningful, answerable questions about individual games.

Some Necessary Conditions

Not all games are even candidates for "balance."  Rock Paper Scissors or a coin flip, for example.  Before it makes sense to talk about balance we need a few pre conditions to be true:

• There is a victory condition.  (The game can be won)
• Success rate is something which can be improved over time.  (The game is skill-based)
• The number of possible actions is always interestingly large (three or more)
• Consequences follow directly from actions (the game is largely deterministic, not random)
• At least a few unexpected "emergent" behaviors are identified.

Some Partial Answers

Some common definitions of balance resonate with me, though these examples are all rather narrow in their use:

• The game has no dominant strategy.
• All strategic options (characters / races / civilizations) are potentially valid.
• The identify of the victor remains in doubt until the game is nearly over.

My definition in two parts

1 - To be balanced, a game must rely on one or more economic systems for which both positive and negative transactions have significant Entropic Force.

2 - The resulting game system must be complex enough that a majority of the transactions lie on one or more of several shortest distance paths between start conditions and victory conditions.

There are lots of terms in there which could probably stand some explanation, but I'll resist the temptation to be overly verbose.  I believe my second point is fairly clear - it's mostly a rephrasing of the "no dominant strategy" point made above.  If we imagine all possible games as a series of decision trees, my #2 statement says that a majority of the transactions must lie on one or more of those winning games.  If there is some choice which can be made - a specific character, technology, card, unit, whatever - and that choice is not part of any winning strategy, then that is essentially a dead choice.  A few dead choices is acceptable, but a clear majority of the strategic options in the game need to be part of a potentially winning strategy.

My first statement, however, requires a bit more examination.

"Economic Systems", "Positive/Negative Transactions", "Entropic Force"

An economic system is one in which different units of value may be exchanged.  In most games one key resource exchanged by players is time - either real time or turns.  It is essential that there be multiple units of value, because otherwise the choices for exchange are too limited.  If workers can use their time to gather minerals or gas, that represents a strategic choice.  If there is no choice - then you simply have steps along a fixed path, and not an economic system.

"Transactions" are exchanges of one unit for another within the economic system.  "Positive" and "Negative" are not value statements within this context - they indicate whether a transaction increases or decreases economic potential.  To take a real-life example, spending money to buy a house increases your economic potential (because a house can be exchanged later - and provides benefits in the interim) but buying food decreases your economic potential (because you eat it, and it is lost).

In an RTS, military units are an negative transaction (economic loss), while investment in infrastructure is a positive transaction (economic gains).  In a shooter expending ammunition is an negative transaction, and collecting ammo is a positive transaction.  And so forth.

So what the heck is Entropic Force?

Entropic Force is a harsh reality of the physical world - entropy means that whenever you convert energy from one form to another you lose some of it.  It's like the tiny transaction fee you pay when you exchange money from Euros to Dinar, or Yuan to Pesos.  Simply exchanging one currency for another will eventually bankrupt you - which means that you have a strong incentive to minimize your number of transactions.  If both positive and negative economic transactions have "strong Entropic Force" that means that there is some amount of inefficiency in every type of investment, choice, or transaction.  Since every decision (including doing nothing!) has an entropic penalty, every decision matters.

And that's the key to balance.

In my next post I explore the ramifications of this definition a bit more closely, and answer the inevitable challenges.

Considering: Puzzle Games

I've designed more than a few games over the past 13 years, but I've never released something which I would call "a puzzle game."  That's set to change soon, (in the next month or two) and as a result I've spent the last few months immersing myself in puzzle game designs.  As a card-carrying geek I have, of course, done my share of puzzles in my time.  When I play a professor Layton game, I'm happy if 30% of the puzzles are novel.  (Curious Village: 22%, Diabolical Box: 34%, Unwound Future: 28% thus far...)  So I thought I'd dip into two puzzle-related themes: a discussion of what elements make a puzzle game really shine for me, and a list of my favorite puzzle games.

Clear Actions or Clear Goals... Not Both

Here's an interesting first premise: a compelling puzzle game should have either a clear goal (Tetris) or a clear action to take (Triple Town) but NOT both.  This comes from something I heard a few years back which really resonated with me

"there are only two questions man can asks himself that mean anything. Why did God create the world? And what do I do next?"

(This particular phrasing comes from Uncanny X-Men #45, but I believe I first heard it elsewhere... not sure where.)

We (players) need motivation to take action, and motivation comes in two broad flavors: to advance a goal, or to fulfill an immediate need.  Both types of motivation are powerful, and lots of games employ them both.  Combat-based RPGs, for example, do a great job of blending world-saving ambition with a need to horde health potions and upgrade armor.  But puzzle games are different, because of the way they manipulate uncertainty and the unknown.

Puzzles are all about illuminating hidden information, and therefore you cannot build a puzzle without creating something which is initially obfuscated.  Consider a classic wood-block puzzle, in which some pieces, but not all, can be initially removed.  The end goal is clear (disassemble the puzzle) but exactly how that goal is achieved is unclear.  Conversely, some puzzles give you very clear methods for advancing (any Tower Defense Game) but obscure what, exactly, you should be doing with those methods (since the composition, timing, abilities of each creep wave is a surprise.)

Content which grows Geometrically

If a puzzle game doesn't get too hard for me to beat - then it probably won't hold my attention.  But if I can't beat the game I'll stop playing.  So for a game to really hold my attention, it has to show me challenges I can't beat (yet), and then give me choices about which challenges to tackle next.  I'll keep trying the "too hard" bits every so often, and then eventually I'll master them.  That's an awesome feeling, but it doesn't come unless a game offers both puzzles of wildly varying difficulty & non-linear progression through those puzzles.

Situations which Encourage Community

A game like Loops of Zen is pretty awesome, but it doesn't give you any situations which can be expressed in words to other human beings.  To really ignite my puzzle-game fires, a game needs to have something about it which can be expressed in words, so that I can talk about it with my friends.  Sometimes this takes the form of a strategy-swapping session, and sometimes it's all about me expressing how awesome the game is to people who haven't played it yet.  I'm a firm believer in the idea that video games should encourage interactions between people, and so this is a key concern for me when exploring a fundamentally solo activity like a puzzle game.

My Favorite Puzzle Games

Heaven & Earth.  This is one of the most amazing, most pure puzzle games ever.  EVER.  The slightly date graphics can be hard to push through for some people - for an updated taste check out the lovely Figure Ground web version.  Play through the first 12 puzzles, and feel your mind expand.  Then your head spins at puzzles 13, 25, and 37 - when new mechanics emerge.  So, SO worth your time.  Then grab the full download for the other 95% of the game.

Adventures of Lolo (1,2 &3)  Some classic puzzle games on the NES.  Dexterity played a small part in the solutions, but the real challenges were all mental.  Some surprising and unexpected depth comes to the fore as you play these games.  I occasionally got stuck for days on a particular puzzle, before I could wrap my head around the mechanics of the solution.

Lode Runner  There have been many versions of this game - the one I played to death back in 1985 was Lode Runner II on my Apple IIe.  I started at level 1, and played as far as I could before running out of lives.  My personal best was somewhere in the mid-80s (I can't recall the level number exactly, much to my chagrin.)  If the iOS version is anywhere as faithful to the original as the video makes it appear, I'll be all over it.  (Note - I was unable to find a link to a real, classic version of Lode Runner to play.  I found literally hundreds of remakes, youtube videos, etc - but I would appreciate a link to something I could actually pass around.)

The Lost Vikings 2  An early game by Blizzard, The Lost Vikings 2 was an amazing action-puzzle game, in which you asynchronously controlled 3 characters with orthogonal abilities to navigate increasingly complex spaces.  This game beats out the Gobliiins franchise, in my opinion, though both games have significant merit.

Picross 3D Did you know they are still supporting this DSi-era gem?  All the fun of classic picross puzzles, plus sweet, sweet victory animations.  My girlfriend played all the way through this one, and monopolized my DS for months in the process.

Pandora's Box  This game doesn't seem to be playable in any form now, which is a shame.  It's designed by Alexi Pajitnov, who also created Tetris.  This was Microsoft's first attempt to take control of his talents - Hexic HD was the second.

Plants vs. Zombies  This game is not only fantastic fun, it's a joy to play.  My favorite tower defense-style game by far (though I like Vector TDX for a more "pure" TD experience).  This is also one of only two "modern" games on my puzzle game list, which means you probably have several ways you could buy yourself a copy right now.

Mass Victory - Development Diary #3

It's been quite a few months since my first post about Mass Victory, when the game was finally starting to come together.  My second post came in May, when I was finishing up my first round of serious playtests.  I'm doing some final clean-up on the game now in anticipation of bringing it to the Board Game Convention in Dallas in two weeks ("BGG.con") so I thought it was a good time to talk about some of the final  rules revisions I've put in place.

The Motivation for Change

By this past June or so, I was very happy with how the game played.  I could play it with people and have a really good time.  I had lots of innovative ideas at work, people picked up the game quickly, and overall I felt like I had a slam-dunk.  I've continued to play the game with family and friends since that June, and nothing has really stood out to me as broken.  But I still want to improve the game, and I think I can do that without really "changing" anything.

You see, I built the mechanics of the game such that players move along two parallel curves of development.  (Man, I'm referencing myself a lot today!)  The first curve is upgrading their personal ship - fun!  More power!  Winning the game!  The second curve is repairing the worldship, and thus bringing the game to an end state.  A really nice bit of inter-player dynamics is the realization that controlling when the game will end is, generally, more important than being in the lead right now.  But the point is that players need to move along that curve, and moving along the curve together makes the game more interesting.

So I'm looking to adjust the start and end points on that curve a bit.  I'm letting player ships start with slightly higher stats and/or slightly stronger abilities.  That essentially means players begin the game on turn 3 - which just serves to shorten the game a bit (not a bad thing) and guarantees that all players reach that "turn 3 power level" at exactly the same time.

I'm also going to remove a few choices from player turns - choices which were seldom-used.  Eliminating choices doesn't weaken anybody relative to the other players, and it gives me a good pool of rules for expansions.  So, without further ado, let me talk about the five modest changes I've made to my final rulebook.

Change #1 - Actions after you move, not before!

Since the very first version of my first prototype, I've given players the choice of taking their 1 Action either before or after they move their ship.  It's a small choice, and it makes sense to me from a story point of view.  But I'm cutting it.  Very few people used it, and it cuts down on quite a bit of text in the manual and player reference card.  It also solves a counter-attack problem which I had to patch in a rather messy way previously - ending your movement near another ship was dangerous, because if you didn't destroy them they could attack you on their turn, before moving away.  I also had a rule that ships couldn't attack on the Outpost, to stop silly first-turn battles.

Nobody will miss this rule.  I can even put it on a ship as a ship power.  Win/win.

Change #2 - Sensors, not Cargo Space, drive Card Limit.

The five attributes of your ship (Thrusters, Mining Robots, Cannons, Field Sensors, Cargo Space) have been pretty locked down for a year or more ("Cannons" used to be "Lasers" - but that's just chrome.)  Attributes generally range from 1 to 5, and Cargo ranges from 10 to 50.  I had an idea that your Card Limit (the maximum number of Research Cards you can have in your hand) should be driven by your Cargo Space.  I came up with the formula "2 + Cargo Space/10" - which meant most people started with 3 cards, and could eventually get up to 7.  That worked well, and it meant that people with more mass tended to have more cards.  Perfect.

Except that some players ("kids") didn't really understand "Cargo Space/10" - especially for oddball ships which had cargo ratings of 8, 16, 22, 30, 36, or similar.  So, after much deliberation, I've decided to let Field Sensors drive your Card Limit instead.  Formula is now "2 + Field Sensors."  This works out to be essentially the same, and so much simpler for everyone!

Change #3 - Static Mining Bonus.

There are two main ways to gain resources (mass) in the game - Mining, and Excavation.  These are supposed to be similar, but not interchangeable.  Here's the old breakdown comparison:

Mining:

• Reduces speed
• Gains minerals equal to die roll.
• Gains bonus of mass + sensors for white gems.

Excavation:

• Costs 1 Blue mass
• Gains minerals equal to die roll.
• Gains bonus of mass for white gems.

Excavation has an up-front cost, which is big in the beginning of the game but pretty trivial by the mid game.  Cannons can also be used to blow up ships, so it's important that Excavation be strictly worse than Mining.  But the only real bonus to Mining was the "+ sensors" mineral bonus when you rolled a white gem.  That was statistically significant - but not in a way which was obvious to new players.

So I settled on a new, statistically equivalent bonus for mining: +2 Red Mass.  If you mine, you just get 2 more Red Mass than you roll, for free, guaranteed.  This makes Mining way more awesome in the early game for players looking at Research Card costs in their hands, and helps level out the fact that the Red Planet is far away from the starting action.  (Red Mass is, by far, the easiest sort of mass to accumulate, but most of it is hidden on moons, which are face-down at game start.)

Change #4 - Teams of Two.

Mass Victory has always been a team-based game, even though it's not uncommon for one or more players to play the entire "team" by themselves.  I've worked hard to ensure that the game can be played by two teams of three, three teams of two, four teams of two, three teams of three... lots of viable combinations.

But this is a bit of a nightmare to explain, especially since it's something players need to know right away when setting up a game.  So I've decided that all teams will have exactly two ships - period.  I'll put something in the end about "Hey, maybe try teams of three." to indicate to comfortable players that the option is viable.  This shouldn't really bother anyone - five players will still need someone to double-up.  But it makes the game language much simpler when I can assume that teams all have exactly two members.

Change #5 - How Mass Accumulates on Planets & Moons.

This has been the biggest, and most volatile system in the game.  I've always wanted the game to simulate mass moving around between bodies - moons bloating up from debris and falling into planets, black holes being formed, etc.  I started with a no-loss system, where games started with a certain amount of mass, and none was ever created or destroyed.  That was awesome, but it became too much bookkeeping.  My most recent system had players placing spent mass into orbit in spaces adjacent to their ships, and then allowing all such mass to "drift" during movement phases.  That added some nice semi-random mass accumulation to the game, but it was still a lot of bit to move around.

It was also possible under that system for some mass to "fall into the sun" - where it was removed from the game.  This made mass accumulation on planets and moons very slow, depending upon how aggressive players were with their drift decisions.  I didn't like this, but I thought I was stuck with it.  Mass accumulation was just too important - too central to the game.

Of course, I eventually came up with a simpler solution which was, frankly, also much easier to understand.  Instead of placing mass into the orbital track, and over some number of turns randomly distributing it across the solar system, why not simply roll a die when it was spent, and place it directly onto some planet or moon?  All at once, turns were sped up, bookkeeping was eliminated from the movement phase, and mass accumulation finally matched up with my projections during games populated by timid players.

Incidentally - that image at the top of the post?  That's my new card-back image.  I got a set of ships and cards printed via The Game Crafter, and I'm very happy with the results.

Don't let game fiction dictate your mechanics: improving Dungeons of Dredmor

It's official - people are finding my Dredmor strategy guide useful.  This is only going to encourage me to write even more this excellent game - I have my upcoming skill builds article slated for Thursday.  But today I'm going to at least try to talk about generalities, although I'll be using Dungeons of Dredmor as my primary example.

Game Mechanics

When I talk about game mechanics I'm talking about the underlying rules which govern the simulation.  In Dungeons of Dredmor, these mechanics are mostly about damage - dealing it, absorbing it, avoiding it, transforming objects into sources of damage, etc.  It's common for a game's mechanics to come out of some game fiction - the situation you are simulating or discussing with your players.  Burliness lets you swing your axe harder, Nimbleness allows you to dodge more attacks, etc.  It would be just as simple, from an implementation standpoint, to have the length of your hair determine how hard you swing that axe, and determine your dodge rate by scaling your body odor - but those relationships would probably be hard for your players to wrap their heads around.

But allowing your game fiction to dictate mechanical relationships can cause lots of problems - because relationships which work in the real world are often too complex to model well in-game, and simplified versions can become mechanically uninteresting.  Let me give an example from a fighting game I developed with Pipeworks in 2004:

An Example of how Mechanics driven by Fiction can be... dumb

I had decided that this game would give a rock-paper-scissors style priority system to attack directions.  The three directions were Straight (thrusting straight forward) Hook (coming in from the side) and Cut (uppercut or overhead attack).  I needed to decide which type would trump which - so I went to have a discussion with one of our designers who was a boxing enthusiast.  Here is a transcript of that conversation:

Me: "Hey Rich, I need to make a loop of attack strengths for different punch directions."

Rich: "Ok, that sounds cool."

Me: "My directions are straight, hook, and cut.  What do you think should beat straight?"

Rich: "Well, let's see" (the two of us pantomime some slow-motion punch exchanges) - "I think hooks should beat straight.  You step out of the way and nail your opponent.  Bam!"

Me: "Ok, then uppercuts should beat hooks?"

Rich: "Yeah... (more pantomime) that seems good.  Uppercuts get you in close, inside the hook."

Me: "Great.  Then straights beat cuts, and we're all set."

Rich: "Hold on... (even more pantomime) no, I think uppercuts should beat straight attacks too."

Me: ... "but then there's no loop.  If straight attacks lose to everything, and uppercuts beat everything, then nobody would use straight attacks and everybody would use uppercuts."

Rich: "Well, I guess that's just the strategy!"

Now I don't mean to imply that Rich was dumb - it might be the case that, in boxing, uppercuts beat everything.  But boxing is very complicated - it's about endurance, stamina, positioning, speed - dozens of systems which we weren't using in our game.  So in real life uppercuts might be too tiring to throw for every punch - but our fighting game didn't have that limit.  So in this case allowing reality to influence our mechanics would just lead to an unbalanced experience for the players.

Dredmor Combat

Combat in Dungeons of Dredmor is relatively straightforward - players can block or dodge attacks, and armor absorbs some amount of the damage that gets through.  In order to deal damage, players need to hit their opponent, and then the opponent's armor absorbs some of the damage.

If you deal 100 damage per hit, but hit only half the time, you have an average damage of 50.  If you deal 80 damage per hit, but have a 90% hit rate, you have an average damage of 72.  So damage and accuracy multiply together to determine your average damage output.

The nice thing about this system is that damage is determined by your Warrior attributes (melee power, warrior skills) and accuracy is determined by your Rogue attributes (nimbleness, dodge, perception, rogue skills).  So going 100% Warrior is less effective than mixing Warrior and Rogue skills.  You need both elements, at least in some measure.

Similarly, it's pretty much impossible to play a 100% Rogue character - because you will eventually need to engage in at least a bit of hand-to-hand / magical combat.  Rogues can do a lot of fighting at range by using bolts, thrown weapons, and traps - but they can't do everything that way.  Rogues can either dabble in Warrior skills, or use some Magic to fill out their ranged combat potential.

Dredmor Magics

Magic relies on three numbers: the damage you can deal with spells, the number of Mana Points you have, and the cost (in MP) of that spell.  Dealing twice as much damage with a spell + casting the spell at half cost means you are 4x more effective with that spell than another character would be.  (This is not quite true because of the "turn cost" - but so long as the damage from spells is relatively close to melee damage the approximation is good enough.)  Having more MP allows you to cast more spells - which improves exponentially as you also decrease the cost of those spells.

So how are those three numbers calculated?  They all derive directly from Sagacity - the Wizard stat.

• Magic Power = Sagacity / 2
• MP = 5 + (Sagacity * 2) 
• Spell Cost = (initial cost - scalefactor*Sagacity)

You gain 2 points to your Sagacity with each Wizard level you gain - and 1 for Warrior or Rogue levels.

This all seems very sensible - because it follows some rules we expect.  But, in fact, we have allowed the game fiction to paint us into an unfortunate corner.  By allowing our Magic system to be driven entirely by this one stat - we are saying that this one stat can replace ALL OTHER STATS in the game.  Dodge doesn't matter.  Strength doesn't matter.  Armor doesn't matter.  The only thing that matters is Sagacity.  Wizards focus 100% of their efforts on increasing Sagacity (or Magic Power directly), instead of being forced to spread their efforts around the way every other character must.

This might not seem like a big problem - but it is.  If Wizards are not overpowered, then that means the upper limit of the exponential growth must be in line with the upper limit of Warrior / Rogue power - which means that initial Wizard power must be tremendously low - so low that "dabbling" in magic is not worthwhile for most characters.

The Fix

So how could we fix this?  The simplest approach would be to separate "increased magical damage" from "decreased spell cost."  That would give Wizards linear power increases within their own skills, and reserve exponential increases for cross-class characters.  This would be identical to the power vs accuracy example we started with for Warriors / Rogues.  There is a stat which Wizards do not get - "Caddishness."  If spell cost were reduced in proportion to Caddishness, then Wizards would have to dabble in either Warrior or Rogue skill trees to get the cost of their spells down.  In addition, Rogues and Warriors would have cheaper spells - albeit their Magic Power would still be quite low, so damage would be limited.

This fix would reduce the upper limits of Wizard Power - but that just means a re-balancing, to increase the damage of low-level spells.  That would actually help all characters, since spell would be a viable early-game resource instead of just a stepping -stone towards eventual awesome Wizard-hood.

Dungeons of Dredmor Strategy Guide Part 3: More Skills in Depth

Part 1 of this guide discussed character archetypes, primary & secondary stat progression and general mechanics.  Part 2 analyzed the first 18 skills in the game.  This final part looks at the last 16 skills, including the complex "Support" and "Crafting" skill sets.

Update: It's not the final part any more!  I've added a fourth part, detailing some character skill builds.

Update2!: All skill descriptions are now up-to-date with v1.0.10 of the game, including YHTNTEP

Necronomiconomics

This magic tree has never quite been ready for prime-time.  The essence of the tree is that you have access to very high-damage spells, at the expenses of causing yourself great pain.  Unfortunately, these "great pain" effects scale up (dealing more damage) the higher your Magic Power is.  So strong Wizards with 30+ Magic Power cannot realistically cast these spells without killing themselves.  One option is to use the Level 3 "Mark of Chthon" - which grants 4 resistance to Necro damage.  That can take much of the sting off.  Pact of Fleeting Life, at level 4 grants a temporary Vampirism effect, which can also help offset the effects.  However, each of these effects cause their own, stacking, penalty effects.  Is it worthwhile?  I don't think it is, in general.  For experts only.

Viking Wizardry

Viking Wizardry is just plain fun.  It works well for melee-types who want to dabble in magic, but it has enough beef to make sense for dedicated Wizards to use it as well.  Level 1 grants an electrical damage buff, level 2 & 3 grant decent Damage-over-time effects which wear down heavy enemies before you crush them.  Level 4 is a huge area-effect debuff which makes your enemies take 3 extra damage on every hit.  Level 5 is a +4 fire damage buff to your attacks, plus a chance to rain down fire whenever you deal or take damage.  Level 6 is Thor's Fulminaric Bolt, which deals HUGE damage to one target, and moderate damage to anyone (even you!) next to them.

Viking Wizardry is a really fine magic school for warriors who just want a bit of extra options - the damage from the first skill is lovely, and the root effect from the second skill is handy even if the damage is low.

Astrology

Astrology has been much improved since initial release - it's now a very reasonable self-buff school for melee hybrids, or for some very specific Wizard-only effects.  Radiant Aura at level 1 grants some serious Magic Resistance, plus it causes blinding flashes to stun and damage melee opponents.  How good is blinding flash?  It's 4 + 50% of your Magic Power, plus a 4-turn stun!  So... that's amazingly good.  Worthwhile at the very beginning of the game, and free later on, so why not use it?

Solar Inscription does similary damage, and dishes out an 8 turn stun - so it won't kill things outright but it does a fine job of blocking monster zoos.  Syzygy at level 4 is a watershed - it can easily become a permanent buff to your character, granting them +2 damage, +12 mana, +12 health, +1 Magic Power, +5 Dodge.  Nice!

Celestial Aegis at level 5 is another big buff.  2 Armor all the time is pretty awesome, and +2 resistance to Necro damage can help take the bite out of a Necronomiconomics build.  15 Magic Defense is also pretty handy.  This now lasts for 8 hits, which makes it worth keeping up most of the time.

The Stars Aligned at level 6 is sort of amazing - but it costs a ton, and is best when you are surrounded.  You didn't survive this long by letting yourself get surrounded, did you?

Promethian Magic

The final Magic School is all about direct damage.  If you want to play a "pure" Wizard, this is probably where you want to go.  Each level provides Fire Resistance as a bonus, which is handy because you will probably be walking through your own flames on accident.  Level 1 is Dragon's Breath, which hits 3 targets in a row in front of you for modest damage.  Level 2 is the Rune of Exploding, which works like a high-powered version of the Blade Golem.  Level 3 is Summon Wyrmling, a pet similar to the Mustache Golem.  Level 4 is the crucial level - where you are rewarded with the AoE Obvious Fireball.  This is the spell you can, in practice, clear out most of the game with.

Level 5 gives Gog's tactical Pyre, which sets a burning spot on the ground.  This is like the inverse of the Unliving Wall - you want to move people into it and keep them there while they burn.  Level 6 is the Infernal Torus - a blasting spell which deals huge damage to everyone around you.  It also leaves fires everywhere, so be careful where you step!

Magic Training

The first blatant "Make Wizards Stronger" tree is, actually, pretty good.  Remember that, as a Wizard archetype skill, you're getting 2 Sagacity each time you put a point into this tree - which means +1 Magic Power and +4 Mana.  This tree gives big bonuses to Savvy - so it will really cut your spell costs in a hurry.  You'll also earn significant bonuses to Magic Power (always nice) and level 2 and 4 give some interesting utility skills.  Level 2 lets you cast anti-magic field on an enemy - essentially cancelling their ranged attacks for a few turns.  Extra-planar-Concentration gives you a huge mana regen effect, at the cost of not being able to block, dodge, or counter while it's in effect.  Totally worth it, in most cases.

If you're playing a mage, you almost certainly want to take Magic Training.

Blood Magic

This is the best way to get Mana back, if you want to use your Mana to kill enemies.  A dedicated Wizard will probably spend 5-8 Mana killing an enemy, and after getting level 2 in the Blood Magic tree, you'll earn that much for each enemy you kill.  At level 3 you not only gain more Mana, but you also get a temporary buff to Mana regen and Magic Power.  At level 5 you get a longer-duration buff which does essentially the same thing.

Level 4 allows you to tap your own life-force to create awesome healing potions.  You can only do this every 100 turns, and the penalties are significant.  But a stack of these will really help you survive anything the dungeon has to throw at you.

Update!  Blood Magic's ability to steal mana has been toned down significantly.  It's still quite good, but Ley Walker and Magic Training now give it a bit of a run for its money.  Experiment with all three to see which work best for you.

Ley Walker

To understand this skill tree, you need to know how Mana regenerates.  On Easy difficulty, you get one Mana back every 4 turns.  On normal, it's every 7 turns.  On Hard, it's every 8 turns.  Each +1 to your Mana Regeneration ability reduces this timer by 1 turn.  Ley Walker's level 1,2, and 5 increases your Mana Regeneration by +1, so at level 4 Hard feels like Easy - at least in terms of mana regeneration.

On Easy difficulty, this skill tree is fantastically good - you start regenerating 1 mana every single turn.  On other difficulties it's a 30-40% increase to your mana regeneration.  Isn't that good?

Well, as it turns out, it's not really that good.  You have plenty of time to regen mana in between battles - it's the mid-battle regeneration which is key.

Update!  This skill tree has been given 2 additional skills - at level 2 and level 5.  Level 2 gives you an incredible mana regen skill which you can activate at will to get 24 mana back.  If you get just a tiny bit of time - this actually covers your mid-battle mana regen needs.  Not shabby.  Level 5 gives you a teleport skill, which also drains away your mana.  If you don't have Mathemagic or Burglary, Level 5 might be worthwhile as a panic button.  If you have either, you absolutely don't need level 5.

If you take this, get level 2 ASAP.  And enjoy.

Perception

Perception is a very interesting Hodgepodge of skills.  Level 1 grants +5 Enemy Dodge Reduction, which is solid by not remarkable.  It also grants +1 sight range, which doesn't really come into play until you start to engage with invisible enemies around floor 5.  Level 2 grants +1 bonuses to sight range, trap vision, and a +5 to Dodge.  +1 trap vision is actually pretty good, but the rest is unremarkable.  Level 3 grants another round of bonuses identical to level 2, but with the notable addition of +2 to Traps skill.  Traps skill allows you to disarm and collect traps much more easily - so it's a source of XP as well as a good way to stay alive.  Level 4 Grants +5 Enemy Dodge Reduction, +1 sight range, and the Second Sight ability - which grants huge bonuses to trap vision, sight range, dodge, and +15 (!) to Enemy Dodge Reduction.

Update!  Two more skills in Perception line:  Level 5 now reveals invisible enemies - which is pretty awesome at those moments when you need it.  Level 6 now gives you LASER EYEBEAMS.  Yes, really.  Most awesome of all, the damage from your eye lasers scales up with your perception rating.  So grab some glasses if you can!  One of the best damage skills in the game, and it costs no mana.

Update Again!  As if Perception were not already pretty good - the dodge bonuses have been upped to +5 each, and you have a 5% chance of spawning a useful item every time you kill an enemy.  These are essential bits like iron ingots, coal - all the crafting items you never quite got enough of.  You find more, and better, items as you increase your Perception skill.

Burglary

This is one of my personal favorite skill trees - it has goodies at every tier.  I find it very hard to avoid taking this on every one of my characters, because it fills so many niches.  Level 1 gives you the minor ability to steal an item from every vending machine.  Not bad.  But level 2 has "Lucky Pick" - the ability to create lockpicks.  If you become accustomed to using these periodically, you have an infinite supply of lockpicks, which means you get tons of easy XP from locked doors and chests.  On top of that you get +1 trap vision and +1 Nimbleness?  Amazing.  

Level 3 ups the ante further, by throwing in a +1 Traps skill - which is enough for a few hundred XP worth of trap disarming, easily, by the second floor.  The actual lockdown ability is handy as well.

Levels 4 and 6, however, are the reasons you'll have a hard time believing you ever went without Burglary.  Level 4 allows you to become invisible at will, and level 6 grants you a perfect teleport ability to any visible space.  Those two skills will absolutely save your life over and over.  Level 5 is the wonderfully thematic five-finger discount, which gives you a free item the first time you enter a shop.  Not an item from the shop itself - but an item of appropriate level.  It's like the vendng steal skill, but sometimes much better.  I've never found it especially useful - but the potential is there.

Update! Burglary has been stretched across a few extra levels - which makes it less of a "every character must take this" tree - but all the goodness (plus some new goodness!) is still there.  Luck Pick is now the level 2 skill - so you'll want that ASAP.  Level 1 is a new ability which gives you a free item from each vending machine.  This can be trivial, or it can be a huge source of early funds, depending on your luck.  Move in Mysterious Ways has been moved to the end of the tree - so it requires 6 skill points instead of 4.  Still totally worth it.

Assassination

This is a Rogue Archetype skill, but it only works if you are fighting in melee range.  Skills at level 1, 4,5 are similar - each provides a bonus to Critical Chance, and each gives you a ~25% chance to deal extra damage + stun your opponent.  With all three levels under your belt, your attacks will stun enemies more than half the time - that means no counter-attacks!  Interestingly, this sort of anti-combos with high Counter Chance, since opponents don't attack you much.  Even one level of this skill tree is useful, and the benefits increase with each additional skill.  If you aren't focused on Counters, and you aren't using Dual Wielding, this is probably a good place to find some damage mitigation.

Either put no points into this skill (still very useful!) or get 5 into it.

Fungal Arts

Oh my, Fungal Arts is a huge grab-bag of strange and awesome skills.  At each level you get an increased chance to "bloom" an enemy you kill into a random mushroom.  There are 12 different types of fungus in the game, which you should learn about.  Their effects are consistent from game to game.  All you really need to know is that Mud Wen are poisonous - though not immediately deadly.  Getting levels in Fungal Arts means you will be creating more and more mushrooms - which is pretty awesome.

Levels 2 and 5 give you summonable fungus pets, which are fairly comparable to the Moustache and Mortal Machine from Golemancy.  They cost no mana, but have long recharge timers.  Level 3 gives you the power to turn mushroom stacks into a random type of mushroom - useful but not awe-inspiring.  Level 4 gives you a 15% chance to deal extra poison damage and create spore clouds on attacks.  Every level also grants +1 Poison resistance, which can be very handy later on.  In most cases, you will really want 2 levels of this - more if you run out of critical skills to improve.

Alchemy

Update! Alchemy is probably the most-changed crafting skill, and that's a good thing.  Alchemy used to give a multiplicative effect on creating booze (used to regen mana) so 1 apple -> 3 weak booze, each of which -> 3 strong booze, each of which -> 2 Aqua Vitae.  That was 18 AV for each fruit, each of which could be changed into a potion.  Ridiculous!

As of v1.0.10, this relationship is reversed.  It requires 2 fruit to make 1 weak booze, and 2 weak booze (of the same type) to make 1 stronge booze.  1 Strong Booze -> 1 AV.  This means that there are two types of alchemists - those who use the skill to make booze as a source of mana, and those who horde booze as a source of potion reagents.  Both types probably want to get level 5 pretty quickly, to craft the top-level staff weapons, including the Fruitful staff, which gives you a huge supply of fruit to feed to your distillary.

The other thing Alchemist do is make potions, and while potions are awesome they aren't really something you'll use frequently.  Invisibility potions are excellent, as are potions of purity.  Health potions are easy to come by, and you should have 50+ stockpiled by dungeon level 4.  Mana potions are pretty useful if you're not focused on brewing booze.  Elemental resistance potions are nice in specific instances - keep a few gems around so you can make them as necessary.  Alchemy also has access to some fantastic throwing weapons - they deal heavy AoE damage, and they don't care how strong your throwing arm is, so Wizards get lots of use out of them.

Alchemy really runs on grinding iron objects into rust powder for healing potions, so it doesn't synergize well with Smithing, in general.

Wand Lore

Update!  Wand Lore has become a 4th crafting skill, and the jury's still out on how useful it is.  You'll need to collect burnt-out wands to craft new wands out of.  Wands have been adjusted a bit to be more consistent damage-dealing tools, which more charges.  They also tend to scale strongly with Magic Power, so your wizards will get SIGNIFICANTLY more use out of them.

Archeology

I'm not in love with Archeology, but many people feel about it the way I feel about Burglary - that it's an essential part of every character.  It provides Nimbleness, Traps (awesome) and Trap vision bonuses, so it can be a good all-round boost to early characters.  Level 2 gives you an ability to transform artifacts (enchanted items) into free XP - another excellent boost.  Level 4 gives you the ability to transform useless artifacts into good artifacts - not a bad trick to have in your kit, though I'd usually prefer the XP.  Level 5, however, gives the amazing Ancient Kronian Ritual.  This allows you to use anvils of Krong Twice - effectively doubling the number of sweet, sweet bonuses you can confer upon your best equipment.  Once you have this, you can use your transmute skill to (eventually) ensure that your bad enchantments become positive.

Smithing

Smithing grants +1 Burliness at each level - a throwback to the time when it was a Rogue archetype skill.  It is now a Warrior skill, so the Burliness is probably too good.  Smithing provides no skills, though level 2 does add +1 Traps.  (Have I mentioned how excellent +1 Traps is in the early game?).  Smithing allows you to create armor & weapons out of ingots, which is a pretty great trick.  Steel is often your limiting factor here - be very careful with your chalk, coal, and pearls.  Thrown Weapons can get away with only taking level 3 Smithing, but anyone else will want to get to level 5, for the best possible weapons & armor.  Smithing is a really excellent skill tree, and highly recommended when you want to explore your first crafting skill.

Tinkering

Tinkering is the final skill tree, and the final (and most complex!) of the crafting trees.  Tinkering provides Savvy, and Traps skill at each level - a total of +8 Traps skill if you reach the maximum level of 5.  Tinkering has a strange catch-all of recipes - it has the amazing clockwork chainsaw, power limbs, Tirker Goggles, Three-pronged sword - but it focuses on three primary types of objects: Traps, Crossbows, and Crossbow Bolts.

Building Traps is, simply, not a very good idea.  You will be picking up hundreds of traps during your play, so making more is a fairly ridiculous proposition.  That means that Tinkering builds crossbows and bolts - so it is the defacto skill for characters who want to fight at range without spells.  Tinkering also makes some Thrown weapons (it's interesting how Thrown weapons are distributed between Tinkering and Smithing...) but those are limited to variations on Makeshift Bombs.  Thrown weapons actually are a nice choice if you focus on making Clockwork Limbs to increase your strength.  The crossbows made with Tinkering are worlds better than anything you might find in the game - they provide between +10 and +14 Pierce damage, which means that your crossbow is generally much more important than your bolt.  Archer skill level 3 (max) provides +4 Pierce damage, which is why Tinkering is much better than Archery, if you have to choose between them.  The best bolts you'll create in quantity deal 8 damage, so it's really all about your crossbow.

Unfortunately, even with max level Tinkering, you simply won't be able to produce / buy / salvage enough bolts to kill everything with your crossbow.  Keep a melee skill or a strong mage-focus build around, and try using Tinkering as just a quick way for mana-free ranged attacks.

Dungeons of Dredmor Strategy Guide Part 2: Skills in Depth

This is Part 2 of my Dungeons of Dredmor strategy guide, focused on analysis of all the skills in version 1.0.6 of the game.  The second half of the skill list is in Part 3.  See Part 1 of this guide if you have any interest in learning about character archetypes, primary & secondary stat progression, or general mechanics questions.

Update: I've added a fourth part, detailing some character skill builds.

Update2! : All skill discussion is now consistent with v1.0.10.  A lot has changed, so pay close attention!

Swords

The best thing about the Swords skill is that it gives you access to swords - which are the most varied and most ubiquitous weapon type around.  The Swords tree doesn't give you much damage - but it does raise your accuracy quite a bit, and the 6th skill gives +4 counter attack chance, which is very good.  It also has a 20% chance to give you +16 counter chance when you are hit in combat!  Skills 2, 3, and 4 each provide +damage, and level 4 gives you an interesting attack which hits three spaces in front of you.  Swords are the weapon of choice for most Rogue-style characters, who are looking to maximize their critical chance.

If you take this you want to leave it at level 1, increase it to level 4, or get it all the way to level 6.

Axes

Axes deal more damage than swords in general, and the Axes skill line has a strong focus on critical hits.  Level 4 is an area-attack which hits every square around you - this is interesting, but generally you do NOT want to be surrounded, so it is rare that you will make such good use of this skill.  If you take axes, you will almost certainly want to build it up to level 6, to get the awesome +3 damage, +14 accuracy, +5 critical chance, and the "Cogito Ergo Splat" ability which deal existential damage in a 2-space radius all around you!  If you are not willing to really focus on a melee weapon, axes are probably not the weapon you are looking for.  For a really weird build, you can focus on Axes, Smithing AND Tinkering in an attempt to make the dreaded Clockwork Chainsaw!

Maces

Maces are a great melee option for characters who expect to use melee attacks frequently, but not exclusively.  The second-level skill gives you the Dwarven Handshake, which is a knockback attack.  Knockback gives you lots of breathing room, which increases your survivability significantly!  Level 4 provides a 2x3 damage area in front of you (significantly more useful than the AoE in Axes) which ALSO deals knockback - your best tool for getting out of tight jams.  Maces are a bit hard to find compared to swords, but Smithing can give you lots of great Maces pretty quickly.  Level 5 and 6 is rarely worth taking on its own- it is a minor bonus.

Update! Mace level 6 now has the amazing "Pedantic Beating" ability - which gives you an 15% chance of paralyzing an enemy when you hit them!  This cements Maces as the best defensive weapon for warriors, and makes level 6 a must-have.

Staves

Staves are really interesting weapons, because they provide lots of good abilities and defensive buffs.  Level 4 gives you an area attack like Maces, Level 2 gives you a stunning attack (not as good as knockback, but quite good).  But if you study staves, it is level 6 where you get your amazing bonuses: +3 magic power, +5 block, +4 Haywire, +5 Magic Defense.  If you dual-wield your staves, you can have a total of +6 magic power and +18 block, ON TOP of whatever bonuses the two staves themselves provide.  That's better than any shield in the game, plus you get the benefits of dual-wielding.

The downside of Staves, however, is that they don't do as much damage as other weapon types.  Also, the Staves skill tree is a Warrior Archetype tree, so it isn't boosting your magic power an mana as much as simply taking more Wizard archetype skills.  Wizards often take this skill for the initial bonuses, but then put no points into it.  I give an example of a great dual-wielding staff wizard in part 4...

Archery

Archery is an interesting skill, but a rather hard lifestyle to stick with.  Crossbows are everywhere in this game, but bolts (the ammunition you shoot) are pretty hard to come by.  Unlike melee skills, you don't get any penalty for shooting a crossbow without this skill, so it's often better to take Tinkering, which gives you better Crossbows, instead of Archery.  The one real benefit of the Archery line is the fact that it increases the rate at which you recover ammunition - a 30% recovery rate means you are getting significantly more shots out of your bolts than other people are.  Even then, Tinkering is almost a must, to give you bolts to fire.

Update!  Ranged weapons have been given quite a bit of love, and they are much more viable these days.  Archery is now upgraded with 2 extra damage attributes at level 2 and 4.  Level 2 gives extra damage which scales with melee power (odd...) and level 4 gives a significant armor reduction and extra damage which scales with Magic Power (ok...).  Ammo recovery rate is still king, but all the new bolt types will give you more options.

Thrown Weapons

If Archery is a questionable skill, Thrown Weapons is almost always a "Do Not Take."  Thrown weapons are tremendously rare, and don't do much more damage than bolts, which you'll find more of.  Thrown Weapons take damage bonuses from Burliness, which means they might work better for a Warrior type - but any real warrior wouldn't need thrown weapons.  Worst of all, however, is the fact that there are some bolt types which deal massive damage - any player can horde a small stockpile of these until late in the game.  There are no equivalently useful Thrown Weapons.  You'll also need to invest in Tinkering to build more Thrown Weapons.  Stick with Archery - or nothing! - for dealing ranged damage.

Update!  I'm tempted to delete the entire above paragraph, because I'm totally in love with Throwing weapons these days.  The damage still scales with Melee Damage, but all the new crafting recipes for throw objects - especially spears and throwing axes - means that a Throwing Smith can kick ass all kinds of ways.  Throwing also now has the really fun "Monster Toss" ability at level 2, which is a lifesaver when facing tough melee boss characters.  This is still not something new players should focus on, but I think it's now very viable for skilled players.  And a TON of fun.

Unarmed Combat

Unarmed Combat is a pretty awesome skill set - it gives huge bonuses to damage, block, dodge, accuracy, and counter-attack.  The weakness, of course, is that you never have any weapons to boost your base damage, so Unarmed Combat can falter quite a bit in the later game.  The semi-hidden bonus, however, is that "unarmed" fighters can hold a shield in each hand - increasing their survivability significantly.  Unarmed Combat gets knockback attacks like Maces, which is awesome.  If you take this skill tree, you almost certainly want to build it up to level 6.  You also really need to have another source of damage - ranged weapons or magic, specifically, to deal with the later half of the dungeon.

Update!  Essence cleverly added a "Nice Combo!" ability to level 6, which means you deal double damage or better 25% of the time when you hit in melee.  This means that Unarmed still has damage trouble in the late game, but can totally get huge hits through sometimes without any additional tricks.

Dual Wielding

Dual Wielding is a very, very powerful skill.  Any melee-focused character will want to consider it strongly.  Simply selecting the skill tree means that you can use a weapon in both hands - probably doubling (or nearly) your melee output.  Taking levels 2, 3, 4 and will provide you with decent bonuses to Counter Chance, which is nice but hardly worthwhile unless you are really focused on maximizing counter chance.  Level 5 now grants +10 (!) to counter chance, making it pretty awesome for any melee character.  Even with all these bonuses, 90% of the advantage given by Dual Wield are available with no extra points invested - So while many, many melee characters take Duel Wielding, very few put skill points into the tree.

Shield Bearer

Shield Bearer, though not obvious about it, is one of the very best defensive skill trees in the game.  It provides some passive block bonus at every level, which is nice.  The bonus persists even if you are not holding a shield - so it is not a worthless skill even if you choose to dual-wield.  The flip side of that is that the bonuses do not double when you hold two shields, unlike weapon bonuses.

The second level gives a very nice (+10 block, +2 armor, +2 health regen) bonus when you are hit, which happens a lot for melee characters.  The third level gives a very nice knockback & stun attack, which is worth its weight in gold.  Level 4 provides an activated defense bonus nearly as good as level 2 (and they stack!).  Level 5 simply grants +4 block, +1 armor, +1 Stubborness.  That's not amazing, but all together the shield tree can provide you +25 block - which is pretty amazing.  You absolutely want to get to level 3, at least, for the stun & knockback attack.

Update!  Defensive Bash (level 3 skill) used to be a bit tricky - because the damage it deals scaled with Magic Power, which warriors often have none of.  This has been changed to scale with Critical Chance, and has been given a 100% chance of paralysis!  So... that's pretty amazing.  Makes Shield Bearer + Axe wield characters especially awesome.  (Remember that you still don't actually need a shield to get these benefits - so dual-wield axe berserkers with shield bearer would be amazing!)

Berserker Rage

You might assume that Berserker Rage is an offensive-oriented skill tree - but you would be wrong.  This tree gives you heavy defense bonuses (and extra damage) when you are fighting toe-to-toe with enemies.  The bonuses do not last very long, but if you are wading into a large group of enemies you can easily be looking at a permanent +10 damage, +15 armor bonus.  Level 1 is decent, but doesn't really provide a reliable attack or defense bonus.  Level 2 allows you to start gaining bonuses on attack AND defense.  Level 4 is where you really start to shine.  If you take this tree, you really want to get it to level 4.  Interestingly, the power of this tree only comes out when you are not doing insane damage - so it doesn't pair will with dual-wielding.  Vampirism for a Hybrid Wizard - Warrior character is a much better fit.

Update! This skill has been given 2 more levels, so it now provides a nice resistance stat-boost at level 3, and a guaranteed critical attack every 40 turns at level 5.  Can you resist getting level 5?  I've been unable to resist it thus far.  Critical attacks can't be blocked or countered!  Plus double damage!

Artful Dodger

Dodging is cool, but the dodge bonus provided by this skill tree is pretty minimal.  However, the dodge bonus isn't the real reason you take this skill - you take it for Knightly Leap - the skill earned at level 3.  This gives you a 2-by-3 teleport skill (like a Knight's move in chess) which can be used all the time.  This is a lifesaver, and tremendously worthwhile for almost any character.  Level 5 gives a +8 to dodge, which is fine but not always worth the 2 skill points.  If you really want to be a master of the dodge, just focus on Rogue archetype skills exclusively, and you'll be just as well off.

Update! - Most level of Dodge have been given a few more points in dodge - making the tree much more effective as a source of overall defense.  Knightly Leap is still the key ability - but +8 dodge at level 5 now brings the total dodge bonus to +19 - pretty good.

Master of Arms

Here is another skill tree which, like Berserker Rage, doesn't quite work as advertised - though it works well.  Master of Arms is nominally about heavy armor - but actually it provides the same sort of protections heavy armor provides - making the armor sort of unnecessary.  Master of Arms is a skill tree which substitutes for armor - allow your character to wear Wizard or Rogue-style armor and still have a high block + armor value.  The big winner skill here is Walk It Off at level 3 - which increases your Life and Regeneration significantly when you are hit.  Along with your other skills, this cuts your regeneration time tremendously, which helps to keep you in the fight.  Note that the Vampirism skill tree eliminates health regeneration - so these two skill trees do NOT play well together.

The other big, big advantage of Master of Arms is that it starts your character with heavy armor - so you have some defense right away.  If you are finding the first level of the dungeon tricky, this will increase your survivability tremendously at the start of the game.

Deadshot

Deadshot looks like an accuracy boost for ranged characters - but it isn't.  Dead shot is a Warrior-archetype skill tree, which increases accuracy for ALL types of weapons, and also provides passive debuffs to your attacks.  Deadshot increases your Critical Chance, and can inflict Bleeding or Crippling wounds on your opponents with any attack - melee or ranged.  If you want to make a Hybrid crossbow / melee fighter, this skill tree is much better than Archery.  With only 3 levels, it's easy to max out as well - though it's not something you'd want to max early.

Update! Deadshot has been removed from the game.  Sorry. :(

Vampirism

Vampirism is a very interesting skill tree - it is a Wizard Archetype, which means that it does not increase your melee power, but it requires melee attacks to drain life for you.  It adds 1 damage + 1/4 Magic Power to your attack, and returns that much health to you.  That's amazing in the early game, but in order for it to matter in the late game you NEED a very high Magic Power - which means a Wizard-archetype build, rather than a heavy-hitting melee powerhouse.  The flip side is that you cannot use food at all - which is deadly if you don't know how to manage your resources properly.

Level 2 gives you a spell which turns a corpse into 4 health - that doesn't scale well in the end-game no matter what your Magic Power is.  Level 3 gives you a "Sparkly" stun spell, which is pretty useful and drains more health from a large area.  VERY USEFUL.  Level 4 gives you a 25% chance to confuse enemies - and does some minor mana drain as well.  Level 5 turns you into a bat - which lets you fly AND gives you +50 Dodge.  Run away!  If you want to use Vampirism, you need to build your entire strategy around it.  Not for beginners.

Golemancy

At last, we come to the first real magic tree - Golemancy.  Magic skill trees grant a specific spell at each level, so they provide a lot of strategic depth by allowing for specific spell combinations at different levels.  Blade Golem is a heavy-damage spell which requires a bit of foresight - you may only cast it on an empty square.  Any enemy which enters that square takes damage, so you generally cast it between yourself an a foe.  Not too hard to get the hang of.  Mustache Golem is a favorite of many - he takes a beating (via very high dodge!) and can kill most enemies on the first 2-3 levels without breaking a sweat.  Unliving Wall blocks off corridors and turns unwinnable battles into easy pickings.  The Mortal Machine at level 5 is an industrial-strength version of the Mustache Golem - he can do your fighting well into floor 5.  Level 6 gives the Digging Ray, which allows you to delete sections of wall, and re-form the dungeon as you see fit.

The most interesting spell, however, is Thaumite Swarm at level 4.  If you have a very high Magic Power, this is the game's most deadly spell.  It will deal huge damage to one target, and then re-cast itself on the space the target last occupied.  If a bunch of monsters are running around an enclosed space, a single Thaumite Swarm can kill them all in relatively short order.  This is interesting because the rest of the Golemancy tree works well regardless of how much Magic Power you have - making it an attractive choice for Wizard Hybrids.  Thaumite Swarm is the opposite - it demands a heavy investment in Magic Power to really shine.

Fleshsmithing

Fleshsmithing is a fairly versatile tree, especially with the most recent patch, which radically adjusted the order of its spels.  The tree also gives you bonus health at each level, and some damage resists - so it's more than just the spells, which is a unique attribute among magic schools.

Zomby at level 1 is the very first summon creature spell in the game.  Zomby is just terrible as a summon - but I wouldn't complain about getting him for free ;).  Knit Tissue is one of the very best heals in the game at level 2.  Many warriors take Fleshsmithing just for this spell.  Fleshbore is a lovely single-target damage spell at level 3, which also reduces block and armor significantly - making monsters easier to beat to a pulp for a few turns.  Meat Shield is now the level 4 skill - it gives solid regen and health boosts, which are of limited use for mages, and is now too "deep" in the tree for warriors to really go for.  Corpus Burst at level 5 is a huge bomb - though it can be hard to set up properly.  You need to kill an enemy, then move away from its corpse before you set it off!  If you can manage it, the explosion is very satisfying.

Level 6 grants the spell Miasmatic Putrefaction, which is similar to Corpus Burst, but may be cast anywhere for a 3x3 cloud of destruction.  This is a wonderful zoo-clearing spell, and does a great job of weakening monsters before you finish them off.  Everyone with this tree needs to take it to level 2 for the heal - Wizards will want to get to 6 for the lovely Miasmatic Cloud.

Mathemagic

Mathemagic is probably the worst magic skill tree you can take if you need an all-in-one solution, but it's one of the very best to take in supplement to another magic tree.  The level 1 spell is a random teleport.  I don't care for it, but it can probably save your life on occasion.  It's cheap too, and works for non-Wizards just fine.  Not a bad spell if you need to take a Magic Tree, but don't plan on putting any skill points into it.  Level 2 gives Beklam's Diminishing Calculus - a terrible spell which does slight damage, and reduces a monster's damage output modestly.  Avoid.  Level 3 gives us the wonderful Curse of the Golden Ratio.  This does moderate damage over time to one monster, but it turns that damage into gold coins - which drop at the monster's feet.  Deadly and Profitable!  The spell is expensive (twice as much as the similarly-powered Fleshbore) but not overly expensive for a dedicated Wizard.

Level 4 gives the confusingly -name Zenzizenzizenzic.  This spell is an amazing force-multiplier for the dedicated Wizard.  It increases your Magic Power by, essentially, 5.  Plus it adds 4 Mana.  You may cast it over and over.  If you have 35+ Magic Power this spell will quickly push you to 50.  Win.

Level 5 gives a perfect Teleport spell - arguably the best escape mechanism in the game.  Level 7 gives the Recursive Curse, a combination of the level 2 and 3 spell which works wonders on even the strongest monsters in the game.  It deals reasonable damage, but also weakens them against damage - so where the Curse of the Golden Ratio will often fail to hit heavily armored monsters, The Recursive Curse will eventually break down their defenses and kill them.

Psionics

Psionics is one of the best magic schools for the non-Wizard to take, as it offers a wide variety of abilities which do not require high Magic Power to be effective.  Level 1 gives a sleep spell, which is quite handy for thinning the ranks of an oncoming horde.  Level 2 gives a knockback spell, which hits a wide area.  Level 3 is a modest healing spell - it heals 36 Life, but you need to move around to take advantage of it properly.  Level 4 is the awesome Nerve Staple.  Nerve Staple does big damage to one target, and paralyzes them.  It also buffs them against Nerve Staple damage - so you can't use it more than once to deal damage.  High Magic-Power characters will use it to kill enemies outright, while low Magic-Power characters will use it more for the Paralyze.

Pyrokinesis at level 5 sets a target on fire.  It deals damage equal to your Magic Power, plus a burn effect.  Again, high Magic-Power users will kill enemies outright with the spell, while low Magic-Power users will use the burning damage to kill enemies.  Level 6 turns an enemy into an ally for 16 turns - a pretty awesome (and cheap!) effect.

I've talked about 18 of the 34 skills - look for the remainder in Part 3 out this Thursday.

Considering: Raph Koster's Theory of Fun

As part of my due diligence, I have taken it upon myself to read Raph Koster's "A Theory of Fun for Game Design."  I've read parts of it in the past, but I figured it was past time to read it and really give it a thorough digesting.

Long story short - this book reminds me very much of "Origin of Species."  It is incredibly insightful, and incredibly important - and we should overlook the not-infrequent facepalm moments in the book because of it.

On Structure

The book neatly organizes itself into two sections.  The first is what most people come to the book for - a definition of "fun" which speaks to the strengths of the video game medium - and ties notions of cognitive theory together with learning reward systems.  That part is, mostly, pretty great.  The second part is looser - it mostly deals with exploring the consequences of the definition presented in part 1.  Unfortunately the second part also fails to really stand up to any sort of serious scrutiny, and contains more than a few questionable assertions about gender roles.

The Theory

Raph's "Theory" is very pleasing in its elegance: he says that learning is fun, and games are fun because they are clean systems which encourage our brains to learn (and eventually master) them.

I'm paraphrasing quite a lot - it takes about 60 pages for the entire picture to be sketched out in long form.  There are interesting digressions into how mental "chunking" allows us to simplify tasks, and how rife our lives are with assumptions instead of direct experience.  I'm skimming this essential part not because it's boring - but because it's really worth reading on your own.

The only place where I felt Part 1 really went in the wrong direction was around page 60 - "Many games... can be directly traced back to the needs of primitive humans to survive..." Koster is speculating here about why some simple themes (shooting, racing, advancement) are so prevalent in play.  His conclusion is that our play-brain craves the rudimentary, instinctive activities of a primitive lifestyle.  I think a much more sensible answer is that simple games are easier to make, and easier to teach.

The Consequences

Part 2, unfortunately, starts poorly and never gains any significant traction.  Page 79 shows how shoot-em-up games have "evolved" over 20 years or so.  This accompanies some assertion that games continue to improve and add new features - and that game archetypes which stop "evolving" die off.

Rubbish.

There are two bits of fallacy here: first, that games must be different from other, similar games.  That's true for someone who plays lots of games of a specific type, but many customers will always be new players - who don't have a body of knowledge to draw upon.  A tweak to the power-up system might be a selling point to a 20-year veteran of SHMUPs, but it matters not at all to the 12-year-old who is playing that type of game for the first time.

Second, the notion that adding features makes games somehow "better" over time, and that this gradual improvement is a key factor in an archetype's longevity.  I don't think games are changing because they are improving - I think they vary because they are created by wildly disparate groups of people, under wildly disparate circumstances.  Koster isn't listing every SHMUP ever created - he's cherry-picking the ones which he feels added something new.  But a good game (or book, or movie) today doesn't mean that games as a whole have somehow improved - some are just better than others.

Who is Playing?

The last third of the book sees Koster really getting into trouble with his assertions - he essentially declares (after defining fun as learning, remember!) that games are fundamentally math, that the "dressing" of games does not matter, and that only math-oriented people can really play games properly.

Now, I personally am very much a math-focused person, and I suspect that Raph Koster is as well.  But to simply dismiss other types of engagements as invalid is a gross display of hubris.

Koster goes further a bit later on, when he defines "designeritis" - the ability for skilled designers to learn and absorb all of the "useful" information from a game very rapidly, thus proving their usefulness. I recognize this dangerous notion - the idea that people with a knack for something don't really have to try - they just understand it without effort.  I hear people make judgements on things almost every day which they haven't tried.  Jumping to conclusions isn't a natural consequence of being a good designer.

So Should You Read It?

I'm comfortable dissing on the second half of this book, because it is the first half which is (rightfully) remembered and respected within the design community.  I think it actually speaks well of the status of the design community that important books like this one can be subject to partial criticism.

So yes, I think you should probably read it.  The book was available at my local library - so reading it won't even cost you anything.

Dungeons of Dredmor Strategy Guide Part 1: Skills & Mechanics

Oh my, I've been loving me some Dungeons of Dredmor.  I regularly visit the Wiki, which is a great reference guide for skills points & crafting recipes.  But I haven't found anything like a real strategy article, so I'm attempting to correct that oversight.

In this first part of my guide, I'll be going over the core mechanics of the game, primary archetypes, stats, and general guidelines for building a character.  In the second part and the third part, I'll provide a detailed analysis of each skill tree's merits.

Update: I've added a fourth part, detailing some character skill builds.

Update 2: I originally wrote this guide for version 1.0.3.  Now that the game is in 1.0.10 and two expansions have come I thought it was time for an update!  I've tried to leave my original text in place where possible, but a significant number of systems have changed..

Primary Stats

The six primary stats in Dredmor are: Burliness, Sagacity, Nimbleness, Caddishness, Savvy, Stubborness.  These each drive the value of some number of secondary stats, including health and mana values.  Like most of the words in Dredmor, their names are 50% informative and 50% humorous.  So let's look past the names and make sense of what these stats mean.

The core mechanic in Dredmor is that every stat, action, and skill is associated with either a Warrior, Wizard, or Rogue.  Every skill you gain is associated with one of these three archetypes.  Each skill point grants three primary stat points, as laid out below:

Warrior: +1 Burliness, +1 Caddishness, +1 Stubborness.

Wizard: +1 Sagacity, +1 Savvy, +1 Stubborness.

Rogue: +1 Nimbleness, +1 Caddishness, +1 Savvy.

Burliness, Sagacity, and Nimbleness (Red, Blue, Yellow) are the exclusive purview of Warriors, Wizards, and Rogues respectively.  Caddishness is shared by Warriors and Rogues (Red and Yellow), and is colored Orange.  Savvy is shared by Wizards and Rogues (Blue and Yellow), and is colored Green.  Stubborness is shared by Warriors and Wizards (Red and Blue), and is colored Purple.

So we can see that, although the names are written to obfuscate, the color coding is here to help us.

Leveling Up

Each time you level-up (by selecting a new skill) you gain +1 to all six base attributes and +1 to the three attributes associated with the archetype of your skill.  If you grab a Warrior skill, you get +2 Burliness, +1 Sagacity, +1 Nimbleness, +2 Caddishness, +1 Savvy, +2 Stubborness.  It is important to note that your rate of increase doubles in your primary stats - so while all of your secondary stats will increase as your gain in levels, secondary stats associated with your archetype will increase at double the base rate.

Your seven initial skills follow this same pattern - so it becomes trivial to calculate your stats at any level, for any given combination of skills.  If you want to be a nimble character who dodges attacks, selecting "Dodge" as a skill choice will, in general, be less effective than simply picking all Rogue-archetype skills, which will grow your dodge attribute at 2x the base rate for the entire game.  If you want to be a heavy-hitting warrior, simply taking all Warrior-archetype skills will do more for your damage and block rate than anything else you can do.

Secondary Stats

Your Life (HP) is calculated as 5 + Burliness + Caddishness.  This means Warriors gain 4 Life per level, Rogues gain 3, and Wizards gain 2.

Your Mana is calculated as 5 + Sagacity*2.  Wizards gain 4 Mana each level, everyone else gains 2.

Critical Chance (extra damage for attacks) is Caddishness / 2.  Haywire Chance (extra damage for magical attacks) is Savvy / 2.

Counter Chance (hit an enemy back when they hit you) is (Nimbleness + Caddishness) / 6.  Rogues get + 0.66 per level, Warriors get + 0.5, and Wizards get + 0.33.

Block is (Burliness + Stubborness) / 6.  Warriors get +0.66 per level, Wizards get +0.5 per level, and Rogues get +0.33.

Sneakiness is (Nimbleness + Savvy)*3 / 4.

Magic Power is simply Sagacity / 2.

Magic Resistance is simply Stubborness / 2

Dodge Chance is Nimbleness / 2.  Enemy Dodge Reduction is Nimbleness / 3.

The biggest takeaway from this is the fact that Wizard skills, and the stats they drive, are totally uninvolved in anything except casting spells.  If you don't plan to cast spells, you don't want or need a single Wizard skill!  Conversely, Warrior and Rogue skills have some excellent synergy to them - trade-offs between block, accuracy, critical damage, dodge are very interesting, and lead to lots of viable character types.

Magic Power

Magic Power is calculated as half your Sagacity.  It influences Wizard characters a great deal, and it has a few other odd effects which are worth talking about explicitly.  First, Magic Power increases the damage of every spell in the game.  It also reduces the cost (in mana) of every spell in the game.  So clearly, if you want to be a wizard you need a very high Magic Power!

----

Update for v1.0.10!  Spell costs are now reduced as a function of Savvy, not Magic Power.  This means that Rogues get very low spell costs over time, just like wizards.  It also means that + Magic Power gear doesn't reduce your spell costs.  Magic Power is still just as important to wizards, but now it's worthwhile to try to increase your Savvy skill as well.

----

On the other hand, there is a lot of skill in selecting and using spells which do not depend upon a high Magic Power for effectiveness.  Golemancy, for example, summons the same Golem regardless of Magic Power.  Teleport, Sleep, and Shove do great things regardless of your Magic Power.

But the biggest oddity is the fact that some non-spells are still considered, by the game logic, to be "spells" - and thus key off your Magic Power.  For example, many traps deal more damage to you if you have a high Magic Power.  This makes Wizard characters especially prone to trap death, and means that Warrior characters can actually protect themselves by wearing armor which reduces their Magic Power!

Skill Selection

The most important aspect of your character - and the one which will drive all of your primary & secondary stats - is your skill selection.  Your skill distribution determines your archetype (Warrior, Wizard, Rogue, or Hybrid) and it also provides you most of the essential tools you will need to survive in the dungeons.  Equipment and components you randomly find will be important as well - but you can't control them.

Warrior Skills: Swords, Axes, Maces, Staves, Unarmed Combat, Dual Wielding, Shield Bearer, Berserker Rage, Master of Arms, Smithing, Big Game Hunter(RotDG), Killer Vegan (RotDG). Clockwork Knight(YHTNTEP), Battle Geology (YHTNTEP) (14 skills)

Wizard Skills: Vampirism, Golemancy, Fleshmithing, Mathemagic, Psionics, Necronomiconomics, Viking Wizardry, Astrology, Promethean Magic, Magic Training, Blood Mage, Ley Walker, Alchemy, Wand Lore, Emomancy (RotDG), Warlockery (YHTNTEP) (16 skills)

Rogue Skills: Archery, Thrown Weapons, Artful Dodger, Perception, Burglary, Assassination, Fungal Arts, Archeology, Tinkering, Werediggle Curse (RotDG), Piracy (RotDG), Demonology (RotDG), Rogue Scientist (YHTNTEP) (13 skills)

The first seven skill selections are your weapon skills - they give you mastery over different combat styles.  Archery and Thrown Weapons are pretty terrible as primary methods of dealing damage, so there are really just five combat skills to choose from.  Every character needs to take one (and only one) of these skills.  If you want to be a pure mage, you can safely skip these options - but you better have a really good plan!

Update!  With the removal of the "bad weapon penalty" - it is now very viable to make a character without any weapon proficiency!  You still really want weapon proficiency skills if you are looking to focus on a specific defense mechanism (counter chance for swords, criticals for axes, etc) and weapon skills are probably the best way to increase your warrior stats, so if you want to melee you want them just for that.  Warriors might even want to mix 2 or 3 weapon skills - but don't try to mix "unarmed" with any of the others!

The next seven skills are mostly defense skills which improve your ability to survive combat.  Dual Wielding is the exceptions - it is the "kill enemies quickly to survive longer" skill.

The next eight skills are the eight schools of magic.  Every character should really take at least one of these - to make use of your mana.  Two or three can be a good idea, but more than three is almost certainly a waste of your time.

The final twelve skills are "support" skills - which improve the function of your character, but don't really provide much direct benefit on their own.  More Mana, stunning melee attacks, free mushrooms, better wand use, a sweet hat...  The most important support skills are the three "crafting" skills - Smithing, Alchemy, Tinkering.  These skills allow you to convert various items you find into other, more useful, items.  Taking one of these skills can be fun, but taking more than one is usually an exercise in frustration, due to the amount of crap you'll be carrying around.

Update! Wand Lore is now a "crafting" skill - but it doesn't really fulfill the same role as the other crafting skills.  Consider it for healing + ranged damage.

Each skill set has between 5 and 8 skills within it.  You begin knowing the first skill from each tree you selected, and with each level you may unlock one additional skill.  You won't be able to unlock every skill from every tree by the end of the game, so you will mostly focus on about 3 skill trees, and use the others only for their first or second-tier skills.  Good candidates for "ignored skill trees" are those trees which give you a good bonus as their first skill - Artful Dodger, Master of Arms, Deadshot, Perception, and Tinkering are the stand-outs here.  Other good candidates are those trees which give you a useful ability from the outset - Fungal Arts, Assassination, and Burglary.

Don't overlook the stat bonuses as potential gains, either!  Selecting even 1 Warrior skill will start you with +1 melee power, which is a BIG help in the early game.  Every wizard skill you select is +2 mana - even if you never take another Wizard skill again.  Rogue skills improve your ability to hit enemies a bunch, even if you don't use the skills associated with them.

In Part 2, I'll go over each skill set explicitly.