Spading is all about figuring out how the game works. From discovering new mechanics such as Semi-Rare Adventures and the effects of Harold's bell, to figuring out the combat frequency in the castle, to finding the success rate of tattered scraps of paper, spading involves the use of many different methods in order to discover the basic mechanics of the game.
All spading begins with a hunch. Through regular adventuring, you realize that something interesting is happening, and want to find out if it's RNG or something deeper. People realized that something wrong was going on while adventuring with +ML and combat modifiers, and it was this hunch that led to a concentrated spading effort to solve the new mechanic. Someone realized that the random numbers on fortune cookies weren't so random after all. It is moments of inspiration such as these that leads to most of the good spading done in the game.
Dedicated spaders have established a multi-clan forum dedicated to spading.
When it comes to spading, the three most important things are: Sample size, sample size, and sample size. The obvious implication of this is that 500 turns of data will be more precise than 15 turns of data, but that's not all. If over the course of 500 turns an event happens 3 times, the margin of error is still quite large. While a sample size of 500 is good for most things, a sample size of 3 is subject to huge RNG.
Margin of error
A quick way to calculate how trustworthy your numbers are is to calculate the margin of error. This is most often done by calculating the range of 95% confidence, which in turn is done by the following equation:
Error = 2 × [r × (1-r)/N]0.5, where r is the rate that an event occurred, and N is the sample size.
For example, if an item was found to drop 9 times in 61 adventures, then the rate is said to be 14.8% +/- 9.1%. This further emphasizes the issue of sample size. Due to the very small sample of actual drops in this case, the margin of error is almost as big as the calculated average! In cases like these, much larger samples are required.
Besides sample size, the biggest difference between regular adventuring and spading involves your control over external variables. While adventuring, your Meat/item familiars will constantly be gaining weight, and you'll be swapping buffs and equipment. While spading, it is important to know everything that is affecting your account at a given time and to keep these constant. For example, it is impossible to spade Meat drops with a constantly leveling Leprechaun, unless the weight gains are explicitly stated! When your results are displayed, usually it is taken for granted that all these secondary variables are constant, and that the only variables that are changing are those that are explicitly stated throughout the spading project. Plenty of spading data has had to be discarded because of people forgetting about passive skills, or what ballroom song is set. Make sure your data is of high quality and control all your secondary variables!
Some important variables to keep track of while spading are:
Effects and wiki-reliability
Try to only use equipment and skills with explicitly stated enchantments. When using things that don't explicitly state their effects, you are trusting someone else's spading - which may not be a reliable source. In particular, avoid effects with only qualitative descriptions.
How can I help?
- Check the Needs Spading category for a list of everything on the wiki for which spading has been requested. Also see the Current Projects and Needs Content.
- ALL new content needs to be spaded to some extent.
- Post your results on one of the spading forums, such as HCO  or AFH . Take the feedback from more experienced spades and use it to refine your method.
- These forums are also great places to find projects you can help join, learning about the process of spading before you go about designing your own data collection and analysis.
Methods for spading various mechanics
A major part of spading involves discovering new and efficient ways of spading various gameplay mechanics. Current methods may be too time consuming or imprecise for the mechanic you are trying to spade, but here is a partial list of some spading methods.
Spading Meat drops involves taking advantage of the way Meat bonuses are coded in the game. Meat drops are first calculated from a continuous range of possible Meat drops, then +Meat bonuses are added, and finally the results are rounded down. As a result, Meat drops caused by +Meat bonuses will not fall into a continuous range. In fact, the precise set of Meat drops is unique to each amount of +Meat bonuses applied.
A helpful zone would be the Beanbat Chamber, which has two monsters, one of which drops 28-40 Meat. Once 9-10 unique Meat drops have been recorded, the unique set of meat drops can be compared to those possible with a wide range of meat bonuses to find the bonus in play. 15 turns of spading can often achieve results accurate to within 0.25% of the true bonus.
Combat Frequency Modifiers
The Penultimate Fantasy Airship noncombat Random Lack of an Encounter is useful for spading combat frequency modifiers, since the results for the choice "Head down to the galley" depend on your current modified noncombat rate.
- Assumption: The number of times "and on" is displayed varies according to 2 + Ceiling[ Floor(noncombat rate) / 3 ]
Using this test:
- Run a net +noncombat on your character
- Choose the "Head down to the galley" choice and count the number of times "and on" is displayed
- Adjust your noncombat rate in intervals of 1% by adjusting the weight of a hound dog, and repeat the test
- After roughly 3 data points have been obtained, the total noncombat frequency modifier on your character can be calculated
- Assumption: Beanbats have a 50.0% enchanted bean drop rate. Therefore...
- With +100% items, a bean will drop every round
- If a bean fails to drop, less than 100% items have been applied.
Using this in a test:
- Apply +X% item drop and the unknown effect
- If a bean fails to drop, the unknown effect provides less than (100-X)% item drop bonus
- Retest with more than +X% item drops
- If after 300 turns you have acquired 300 beans, the unknown effect provides at least a (100-X)% item drop bonus.
- Retest with less than X% item drops and repeat.
- Starwed has made a helpful webpage for spading simple and complicated item droprates.
DelevelingMost skills and combat items that delevel do so explicitly, displaying the message below:
|Monster attack power reduced by X|
|Monster defense reduced by Y|
Prior to the explicit deleveling message, unarmed damage was used to determine deleveling effects. However, this method is not optimal given current resources.
While most initiative modifiers are explicitly given in the game, there can still be other methods that affect initiative (such as increased ML).
Assumption: An increase or decrease in initiative will directly affect your chances of running away.
To spade an increase or decrease in initiative, unequip all and turn off all ML effects.
- Pick a zone with one monster: the Dire Warren, the Hedge Maze, or using the drum machine or create a zone with one monster (using Banishing effects).
- Find your Point of No Escape (P). This is the level of modifiers (usually negative) where
- You cannot run away from a particular monster over 5 adventures.
- You can run away if you have 5% more initiative.
- To determine this point, only use equipment that have a stated plus or minus initiative.
- Avoid antique equipment, as this equipment tends to break during testing.
- The 1-ball is very useful, as it is the only common +5% initiative equipment.
- Once you have established your Point of No Escape, apply the effect or outfit you wish to test.
- Find your new Point of No Escape (Pn).
- If you adventure and can run away, apply another -5% of initiative.
- If you can adventure 5 times and cannot run away, apply another 5% of initiative.
- P - Pn is the initiative modifier.
This information is mostly obsolete -- there may now be conditional items which can drop from yellow rays, pickpocket attempts, or rave steal
There are several types of conditional drops, all outlined on the Conditional Drops page. There is an easy way to check for conditional drops. Using a major yellow ray from the He-Boulder (or a pumpkin bomb or anything else that causes the Everything Looks Yellow effect) forces all non-conditional items to drop from the monster.
It is still possible for conditional items to drop when doing this, and they aren't distinguishable if they do drop. The way to prevent them from dropping is to reduce your item drop modifiers to at least -100%. Good items for this task are makeshift SCUBA gear and aerated diving helmet, which combine to give -150% item drops. By doing this, any item that drops is not a conditional drop, and any item that does not drop is a conditional drop.