Specialization

Version 0.34: This article is up to date for the latest stable release of Dungeon Crawl Stone Soup.

Specialization is a choice to invest XP in a few specific skills, rather than attempting to learn many skills at once. It is often essential, at least in one form or another.

Useful Info

A higher skill level means that skill is more powerful. Weapons will hit harder, attack faster, and be more accurate. Spells will be more powerful and less likely to fail. Note that end-game weapons and spells usually require a lot of skill investment. Defensive skills will make your character more durable.

Specializing is easy. Press the m key to access your Skills screen and see all the skills you currently know. Pressing the key associated with a skill will toggle its training Off (dark grey), On (light gray, +), or "Focus" it (white, *), causing more experience to be allotted.

By default, the skilling system will use Automatic Mode, which automatically adjusts experience division between skills based on use. You can also press / to swap into Manual Mode for more control. In Manual Mode, experience is divided evenly between enabled skills, with "Focus" doubling a skill's percentage.

Analysis

In Crawl, the goal of the skill system is "to get the most benefit out of your XP as possible". Sounds simple, and it really is. Specialization is a tool used to achieve this goal. So, there isn't a true Pros and Cons with specialization - more so Pros and Caveats.

Pros

1. Specializing in a skill typically leads to less "wasted" XP.

Usually, but not always, "wasting the least XP" is an efficient method of skill training.

Let's say you could either have {6 Polearms, 6 Conjurations, 6 Ranged Weapons} or {10 <one skill>}. The former may sound more appealing, as you can take advantage of 3 skills at once. However...

• This can lead to skill waste directly. If you blast things with Conjurations before enemies get into melee range, then Polearms and Ranged Weapons skills aren't doing much. If you are wearing heavy armour, to make your melee more effective, then Conjurations skill isn't doing anything (it is very difficult to cast in plate armour!).
• There are more than three skills in Crawl. For example, a Polearms user could go {8 Polearms, 5 Armour, 5 Shields}, and a caster could go {8 Conjurations, 6 Dodging, 2 Spellcasting}. Both will be more effective than the original 6/6/6 distribution, which invests no XP into defenses.

2. There exist skill thresholds, and specializing allows you to reach these thresholds faster.

The game has a few skill thresholds:

• For weapons and destructive spells, this is when you can "comfortably kill things at <a certain level>".
• For casting in general, this is when a spell goes from "unreliably castable" to "castable" (around 10% failure rate, as a rule of thumb).
• For defensive skills, breakpoints exist at each full point of AC / EV / SH. From the skill menu (m), press ?, then the skill to see how much each skill point will improve defenses.

Training skills too liberally will delay the threshold. E.g., going for {6 Polearms, 6 Conjurations, 6 Ranged Weapons} may prevent you from 'comfortably' killing things at all.

In other words, having one reliable way of killing things is better than 3 unreliable ways of killing things. If you want to go for a "hybrid" build, it is best to train one skill to a competent level, and then the next. This is not to dissuade "hybrid" builds entirely, but to put them into perspective. Of course, when things are 'competent' or 'comfortable' is very arbitrary - it depends on both the character and the player.

Caveats

1. Higher levels of a skill are more expensive. / The first few levels of a skill are cheap.

For each level of a skill, the next level is more expensive. Is it really worth the XP to train from 25 Fighting -> 27 Fighting? No matter how much you value extra HP, the benefit of 2 Fighting is small. It can be wiser to invest in Armour, Dodging, etc., which could all provide more relative benefit.

This concept isn't exclusive to the late game. For instance, training 14 <weapon skill> to reach mindelay may not be the wisest choice. You could, as an example, get {10 Weapon, 8 Fighting, 8 Armour} instead, which would be a more well-rounded investment. Investing in a skill you're already good at gives a more immediate benefit. A Polearms-based Fighter may need 4 levels of Conjurations before they can cast their first spell. Before you reach 4 Conjurations, you aren't getting much from that skill. Meanwhile, having 4 more levels of Polearms would help for a longer period of time - but this is exponentially more expensive the higher the skill level.

Judge whether a skill would help in the short- or medium-term, and make sure whatever build you're currently using is good enough to make it out of the short-term in the first place!

Other considerations:

• Early on, you could invest a few levels into Evocations or Throwing. After all, 3 levels of Evocations can be more beneficial than 1 level of a weapon skill. This ultimately depends on character build, and what Wands or other Evocable items you've found.
• If you are in heavy armour for the entire game, then the Dodging skill might not seem appealing. But it's often worth it to train a few levels into Dodging, since they are so cheap. For the same reasons, light armour characters should consider investing a few levels into Armour skill.
  

2. "Avoiding wasted experience" is not always the most beneficial method of skill training.

Alchemy is a good example of this. Spells like Olgreb's Toxic Radiance are great in the early game. But as enemy HP increases, and as more enemies have rPois, poison starts to wane. If you then find a great weapon and wear heavy armour, then you'd have "wasted" all that Poison Magic training.

Yet, this doesn't change that melee is (could be) the most beneficial thing you could train right now. Plus, all that Alchemy training wasn't useless - it got you past a good chunk of the game. And if you decide to stay in light armour, then OTR and other poison spells can remain useful. Transitioning to other schools of magic is even easier: you've already trained some Spellcasting, and taken Intelligence on your level-ups, so finding another good spell on the ground may be reason enough for a few cheap skill levels in a new school.

Even in this example, you should be following the guideline of "train a killing skill until you can comfortably kill things, then train other things". At a certain point, you don't need to invest any more XP into Alchemy and can afford to train other skills (such as defenses and/or an alternate killing skill).

3. You need some way to deal with Orbs of Appropriateness.

The endgame Orbs have a variety of immunities and high defenses overall.

You don't need to kill everything in Crawl; you can walk up to the Orb of Zot, Apport it, and get out without killing a single Orb of fire. But there are other lategame enemies that will be immune or heavily resistant to certain elements; Hell knights for fire, Titans for electricity, etc.

For casters, Conjurations is a good bet. Many schools have Conjuration spells to take advantage of. Later on, you can branch out to other schools with the Conjurations training. Both Orb of Destruction and Iron Shot deal non-elemental damage, which can be resisted only by AC. Or, if you have enough magical aptitude, then you can cast level 9 spells, which can all pierce through resistances in some manner. Allies-based spells from Summoning, Forgecraft, and Necromancy also solve this problem for casters; because, just as weapon users will find, nothing in the game is immune to physical attacks.

Strategy

As you have full control over your skilling, it's best to specialize selectively. What does this mean for your character?

• For melee and ranged characters, it means focusing most, if not all, of your experience on one weapon at first. Unless you find an absolutely amazing weapon, once you specialize, you shouldn't switch weapon class. Afterwards, you can start investing in defensive skills, Invocations, Throwing, etc. There are two main skill breakpoints associated with weapons:

• The first is when you can attack at 1.0 delay. This prevents a regular-speed enemy from hitting you twice in 1 turn. Death yaks become much scarier when they can deal 60 damage instead of 30 damage per turn.
• The second is the Minimum delay of a weapon. After training to mindelay, you cannot attack faster by training weapon skill. Thus, investing more into said weapon skill will give you less benefit. It can still be valuable to train a skill past mindelay for the damage bonus, however.

• For casters, it means getting your main killing spells reliably castable (<= 10% failure) as soon as possible. Whether it be Stone Arrow or Ensorcelled Hibernation, you will want to kill before you get close to being killed. Additional levels might help increase spellpower, but this is rarely more important than getting more useful spells to a castable level. When you're able, you sticking with those initial schools provides a head start on later developments; however, don't get caught in a sunk-cost fallacy. Often, the best approach is branching out for spells that complement your main spell package rather than hoping the best spells in your school come eventually. Focusing on schools with more multi-school spells, such as Conjurations and Forgecraft, will also allow for more flexibility.
  • Some characters, especially those not worshipping a spell-gifting god, might not find any lategame-worthy spell of their school. Make sure that your currently useful spells are reliable, then dedicate everything else, somewhere else. Don't consider yourself fully committed until after the Orcish Mines - a book shop might just provide you with a clear direction. You can also bide your time by investing in defense skills and Evocations that will always be useful, or worse case, swapping to melee if RNG doesn't provide you with a suitable spellcasting direction.
  • It's worth pointing out that all spellbook-starts have had their initial spell package dialed back in recent years. While it may have been suitable to "force" your starting school for the whole game in the past, this change, along with the Parchments change to spell acquisition, has made spellcasting much more reliant on "playing with what the dungeon provides". Knowing when to pivot is one of the harder skills in Crawl. Stay flexible, keep your options open, and let the dungeon guide your long-term build rather than trying to force one that never materializes.

History

• 0.34, fewer spellbook starts have long-term plans from Turn 1, and paired with the addition of Parchments, this has changed how often spellcasters can reliably stick to the starting school all game.
• Prior to 0.9, experience was gathered and spent gradually as you performed actions related to each Skill. It was much harder to avoid unintentionally training certain things.