A Simple Approach
The simplest way to determine how pools of fish appear is to fish all the existing pools and then wait for new ones to appear. Initially every pool in the coastal part of a zone is fished. Then, every ten minutes, one rides up or down the coast recording (but not fishing) the types of pool found.
Although inherently simple, this method is easily undermined by other people fishing. This has been controlled by:
- Fishing at quiet times of day, typically mid-morning.
- Forming a mental map of the pools found on each ride, so that if any existing pool has been fished, it will be obvious.
- Abandoning attempts once other fishermen or women are spotted in the area.
- ((And fishing on a relatively new, low-population realm, where most people lack the skill to fish.))
Two locations were tested, each selected for the ease of riding a route that checked all possible pool "spawning" locations:
Both contain an almost identical number of possible pool locations. The skill requirement and type of pools varies.
Although a popular zone, hostile creatures along the coast and a equal number of Horde and Alliance, makes the Hillsbrad Foothills a rather dangerous for casual fishermen and women. The most frequently fished pools are those that appear in Southshore itself. Some pools can go untouched for hours at a time.
It took just under 60 minutes to fish the 23 pools found on the initial run along the coast:
Then every ten minutes, the number and type of pools found along the coast were recorded (but not fished). These pools are shown in the graph below. Each pool type is shown as a shaded block. The blocks stack to show the total number of pools. Time is shown from left to right.
The most important finding is that the number of pools is not constant. In 60 minutes during the initial fishing, the number of pools reduced from over twenty to just three. In the 20 minutes after the initial catches were made (60-80 minutes after the first catch) a rapid increase can be seen in the number of pools to 15, then slower growth to 20 pools within an hour. The proportion of each of the three pool types is much more even than before, with no bias towards Oily Blackmouth pools.
Two patterns were observed that are not clear from the graph:
- The first pools to appear were all in the part of the coast that had been fished first. It is possible that pools do not reappear until about 60 minutes after being fished.
- Once pools have appeared, they remain until fished. They do not change over time unless fished.
Of course this is just one trial in a process with a lot of variation. Let's move on to Feralas.
Feralas is a quiet zone. The appearance of Stonescale Eel pools and high-value wreckage tends to attract fishing activity. The pools are patrolled by a variety of Elementals and Giants, making the area difficult for less experienced individuals.
It took just under 60 minutes to fish the 23 pools found on the initial run along the coast:
No Stonescale Eel Swarm were found, in spite of this type of pool often appearing in the area.
As before, every ten minutes the number and type of pools found along the coast were recorded (but not fished). These are shown in the graph below.
The patterns are quite similar to those in the Hillsbrad Foothills. There is a fairly rapid increase for twenty minutes after the initial catches. This is followed by a slightly slower increase for the rest of the hour. The rate of pool re-appearance in Feralas is slightly slower than in the Hillsbrad Foothills.
There is a very dramatic reduction in the number of pools following initial fishing, this time from over twenty to just one. Although there is a slight bias towards Oily Blackmouth pools, biases are not as extreme as at the outset (17 Firefin Snapper pools with no Stonescale Eel pools).
In both Feralas and Hillsbrad there is one curiosity: The balance of different pool types after our test is far more even than before the test. The reason is "cherry picking".
Cherry picking involves fishing just one or two types of pool, and leaving all the others. It is common because most people have no use for Oily Blackmouth or Firefin Snapper, preferring to fish wreckage or Stonescale Eel pools.
To examine how this affects the appearance of pools, I have continued the previous Feralas test. All the Stonescale Eel and wreckage pools were fished, leaving the Oily Blackmouth or Firefin Snapper pools untouched.
Over 30 minutes, ten pools were fished: 3 Stonescale Eel pools and 7 wreckage pools. Then every ten minutes runs were made along the coast, recording (but not fishing) the pools found. The results are shown in the graph below. Timings are from the first catch in the initial Feralas test above.
Pools continue to appear gradually over time, but the new pools are not restricted to the two types that were fished. Instead, any type of pool may appear. So, the result of cherry picking is that fewer and fewer Stonescale Eel and wreckage pools appear in the area. This explains why, when we first arrived in Feralas, we couldn't find any Stonescale Eel pools at all: They had been cherry-picked out of existence, and nobody was fishing Oily Blackmouth or Firefin Snapper pools to compensate.
Compare this test to the previous test. The rate at which pools reappear is slower when only half have been fished, than when all have been fished. It seems that the rate of re-appearance changes depending on how many pools already exist.
Rules are There to be Broken
Unfortunately, the concepts described above cannot be applied to all pools.
For example, certain Sagefish School behave quite differently. In Mirkfallon Lake, in the Stonetalon Mountains, there always seem to be two Sagefish pools in an area. Sometimes instead of a Sagefish pool, a wreckage pool, but (in my experience) not more or less than two pools. Fishing one pool to extinction merely creates a replacement pool somewhere else in the lake.
A similar pattern seems to apply to Ashenvale - but with a twist. The possible pool locations are split between various different lakes in Ashenvale, including the lake at Astranaar and Mystral Lake (near the Stonetalon tunnel). For example, there may be one pool in each lake, or two in one and none in the other. Since the lakes are several minutes walk apart, this can be dreadfully confusing for anyone trying to fish just one lake.
However more than two Sagefish pools can be found in inland areas of the Hillsbrad Foothills, so such a simple pattern cannot be applied there.
Another unproven theory I call "there is always one": In any area there will always be at least one pool at any one time. This should be easy to test in Faldir's Cove, in the Arathi Highlands. Here there are only five or six possible pool locations, all very close to one another. For the first six pools, each pooled I fished out was immediately replaced by another pool at a different location in the cove. Unfortunately after six, I couldn't find a seventh pool: So either the theory is wrong, and eventually it is possible for no pools to exist, or I simply missed a possible pool spawning location.
I have credited a theory called "double pools" to Mizzy. There is a rare chance that a new pool will appear directly under the one that you are fishing. This can result in far more fish than normal from one pool, or a pool that changes type while you are fishing from it. I recall Mizzy originally stated that this applied to any catch over four, and therefore in extreme cases (12 catches from the "same" pool), "triple pools" were possible. However, I am sure six catches is the maximum for all normal pool types except Sagefish. The (pre-Cataclysm) official documentation said "4-6". This makes these double pools somewhat less common. Personally I've only experienced double pools in Stranglethorn Vale. They may be more common there due to the zone's popularity for pool fishing, or the fact the zone has more pools than any other.
Charming writes: "Not all the nodes are active at all times of day. For instance, one node may be active during the 12pm/6am period, but NOT the next timeframe. I have noticed that certain times of the day it appears that more or less nodes are active." I don't have any evidence to support this. It is quite an attractive theory, since it matches how catches from open-water fishing vary by time of day (see the Variation by Time topic). Unfortunately there is a much easier explanation for seeing different patterns of "nodes" at different times of day: People are more likely to fish at certain times of day.
In Outland pools are the prime source for all useful fish. I have spent a long time fishing these pools: You can read further details in The Burning Crusade, Catching Mr. Pinchy, and Fishing Pure Water topics.
Extensive experimentation, particularly with Highland Mixed Schools, established three core principles:
- There are a fixed number of pool spawning locations in the set, any of which may spawn.
- Each pool will typically respawn 60-65 minutes after it was last being fished (emptied). It is very unlikely to respawn in less than 40 minutes, but will sometimes take as long as 90 minutes.
- There is a maximum number of pools that will be visible at one time across the set, regardless of time since last catch. This is typically between 30-50% of the all possible pool spawning locations.
The first rule probably applies everywhere, however it is possible that in some zones there are several sets of pools (not just a separate set for inland and coastal areas). Identifying sets is not terribly easy.
The second rule probably applies selectively: In some zones I believe the minimum respawn time is zero - typically low-level areas, such as School of Deviate Fish in the Northern Barrens. In most zones the "about an hour" value will apply.
With those two caveats, the three core principles seem to be able to explain all pool appearances.
- Pool Spawning - The rules governing school appearance. Why you a particular pool cannot be found.
- Pools and Schools - Describes how to fish pools such as schools or floating debris, where to find pools, and how to efficiently fish them.
Comments about Pool Appearance
Below are readers' comments about "Pool Appearance":
Gurko, March 2008:
I can confirm Mizzy's double pool theory. If you are tracking fish, the mini-map will show two entries when you mouse over the blip. I've even had this result in a mixed pool (blackmouth and snapper). I can only recall this happening with pools in STV, but I have noted equivalent effects with mineral and herb nodes in other zones.
el, March 2008:
Agreed. With pool tracking it is very obvious, but I've only seen it in Stranglethorn Vale, specifically on the northern part of the coastline. I reported it as a bug during patch 2.3 testing. But then I reported the Nagrand pools that were too far from land during TBC testing, and 18 months later we get an Elixir of Water Walking ;-) .
Gihelle, March 2008:
You can't track the Zulian Mudskunk pools in ZG too.
Unit, October 2009:
I was just wondering if anyone has spoken with a GM or has insider info. about how nodes actually work, fishing nodes in particular.
Are items "preloaded" on a node when they spawn?
If not, when does your random roll for the items in the loot table take place - at the splash or when you actually loot the bobber after the splash?
Grank, October 2009:
If I understand correctly, the GMs would not know the answer, but the Devs would and from what I've seen they don't give out much of the specific game mechanics. Usually it takes players with a good programming background and a fair bit of intuition to figure some things out.
Since pool catches are always random and one can't repeat the experiment (go back and fish the same pool twice) we may never know. If the contents are not created with the pool, then I would guess that the item roll would happen when you click to loot, since there is no advantage to doing it before and slightly lower overhead to delaying it.
el, October 2009:
My initial reaction was also, interesting, but is there any point in knowing? A slightly random response follows:
Pools are held in the realm server's memory, and generated as the realm server is started. If the server crashes they completely reset (including the chance that they will reappear in different places).
Quite a lot of world information is held this way. Until recently, if the realm crashed during the Sunday Extravaganza, non-contest pools would re-generate on re-start, because even the event itself was held in the memory. The state of mobs (creatures) is similar: Probably the most fantastic sight in the game is the mass-submerging of all the dwarf mining machines in Storm Peaks, at the moment the server starts up - an entire field of neatly lined-up machines disappears into the earth. (Unfortunately you need a very unstable test server to plan to see that.)
Loot. The wisdom of EJ suggests that all loot is held in cascading tables - it's an efficient way of creating the illusion of complexity, while also being very hard for us to untangle. Same logic of "illusion of complexity" was applied often in the original WoW - for example, originally all of Azeroth's fish were determined by a 2x6 matrix, with a 4-stage time variation applied to 2 of the 6 zone groups (see Gazetteer) - yet that pattern wasn't broken until I started looking, almost 2 years after the game's release.
The best understanding I have of this comes from rearranging "bugged" stacks of herbs (it should work with ore to): There is a rare chance that a stack of herbs will simply not mill (via Inscription). However, if you gradually move 1 herb at a time out of the stack, into a new stack, there normally comes a point at which one of the stacks can be milled. This suggests that the outcome of the milling is determined by a calculation based on the entire contents of the stack, not simple that "you have herbs". That in turn implies that each herb carries unseen data, because the outcome of the process must always be the same (the initial failure of milling is consistent until the stack composition changes).
Before WotLK, there was an official comment revealing that creature loot was created at the time the creature was created. The specific example was for instances (dungeons), but the same technique would logically apply to everything.
However, WotLK introduced something I call progressive de-randomization of quest drops: The longer you farm for a quest drop, the higher the probability of the drop from each mob killed. After a certain number of kills, the probability becomes 100%. It ended those really annoying ("I killed 100 mobs - this quest sucks") grinds.
During WotLK beta a few people noticed that some of the really annoying quest drops had become less annoying, but I don't think anyone realized the method of determining drops had changed - that was revealed by Tigole some months later.
The change initially caused fish quest items to "drop" with 100% chance - something which I'd found in testing, and remained during the first few weeks of release. So fish were also altered by that change, but possibly in a slightly unusual way.
(It's worth adding that pre-TBC quest fish and TBC quest fish had previously been coded in different ways - one as a separate catch; the other as a catch alongside another fish, like we now have. Fish quest drops were programmed inconsistently, so more likely to get broken.)
While progressive de-randomization only applies to quest items, it is possible that all fixed loot tables are no longer as fixed as we might think. For example, the Dalaran fountain loot table changes based on the presence of a player-specific buff. In theory Blizzard can change the catch rate depending on what hat you have on, whether your faction controls Wintergrasp, or... any spell (even an internal "invisible" buff). And because all the fish drops are decided by the server, there is no way for data-miners to know. Fortunately that doesn't fit Blizzard's overall design, which is increasingly very clear about the factors influencing anything - which is probably why we don't see Variation by Time of day for fish in TBC and WotLK.
Why are loot mechanics changing?
It's probably the worse kept secret that Blizzard had serious problems with databases in the early years. As you might imagine, the volume of constantly changing data floating around WoW is immense. And write-intensive databases don't shard easily.
However, in the 5 years since, server capability has got 3 or so times better - very crudely, you can attach more memory to individual server machines. Which is probably why we're now seeing the evolution of things like drop rates - and more obviously achievements/statistics, or larger numbers of players supported on each realm: A game originally designed just outside the boundaries of the technology, is now comfortably within it.
So, where does that leave pools? It doesn't. I still don't know. But be wary, because what might have been the case a year ago, may not now be the case.