Spawning: Discovering The Weights

Cataloging the relative influencers of the Spawn Engine for Halo 4 is a dramatic turn for me to make, having witnessed endless fails and having preached the futility of doing so with Reach. But a claim that it is possible having been found true (to a point) piqued my interest to write this article. Though the tests are a bit complicated, the relationships between the influencers can be effectively reduced to a single graph. And knowing these relative weights can help forgers understand the spawning behavior and help them design their spawn layouts more knowledgeably.

Updates

20 Sep 2013 Update:

343 Industries has come out with a new game type recently called Ricochet. The CertainAffinity web site has a blog post that stated that a Respawn Zone that holds the RICO_RESPAWN label will force a team to spawn on Respawn Points within that zone. I found that imminent danger and LOS can thwart this behavior, but the fact remains that this label increases the weight of the Respawn Zone by quite a bit. I discuss this in the appropriate updates in this article below.

28 Jan 2013 Update:

I want to update this article with new information about the LOS Influencer. It turns out that the tests I conducted were concealing the true and typical influence produced by the LOS Influencer, that the test configuration I describe for most of my testing did not take into account what was known before as the LOS bug. (Honestly, I wonder if it is a bug after all, given what I learned here today.) I describe my new findings below under the LOS Influencer section, but I want to take a moment here and repeat that it can be damaging to one’s reputation to put an article like this together, since there is a heavy reliance upon empirical evidence rather than solid original source. And as I show in this particular update, I simply didn’t think that adding a third Respawn Point to the test would have altered the value of the LOS Influencer, let alone from -3 to off the charts beyond -300.

This potential for great misinformation is the nature of empirical evidence. The sample points represent only a very narrow set of configurations that we can think to try out. They don’t tell us how the Spawn Engine handles every situation, and thus when unexpected behavior occur, we wonder if the model presented (as in this article) is even close to reality, or what? That is where the author’s credibility takes a beating.

Introduction

I have read countless articles, posts, and threads attempting to catalog weights of the various influencers in Reach using nothing but empirical evidence and inductive reasoning – it was… heart breaking. The trait that they all had in common was that they lacked details of their tests (no peer review was possible) and they all contradicted the game publisher. This has lead me to preach many times on the fallacies of trying to catalog the influencers’ weights in Halo based upon empirical evidence alone. You can imagine my tepidness in contributing to this topic as well.

But I was intrigued by [audley] which suggested the weights were really so close to each other that one could actually pull off such an analysis. Could this be true? The only thing that [audley] lacked were the details of the test procedures used, preventing any peer review of his test environments from taking place – I had no way to determine why he came up with different results than I did.

I would like to stress that the results are not anything close to precisely describing what is occurring under the hood. Instead it represents a rough model that a forger can use to understand how spawning is affected by several key influencers. Models such as this one derived from empirical evidence cannot be considered technical specifications in any sense of the phrase.

This article is a lengthy analytical read. For those of you who are interested only in the results of my tests, you can skip down to the bottom of the article where the Summary section presents the weights in an easy to understand format.

The Spawn Engine

I have no cite from the game publisher to confirm or deny that the Halo 4 Spawn Engine is based off the same that was in Reach. But based upon the evidence (functionality, randomized tie breaker, and the  same rare corner case LOS bug), it would be an astronomical coincidence that the two Spawn Engines were independently developed and yet shared these exact same traits (the LOS bug was the calling card). For this reason, this entire article will assume that the Spawn Engine is indeed based off of the Spawn Engine in Reach.

Update: Certain Affinity has confirmed that the Halo 4 Spawn Engine is “an evolution of the spawning system of Halo Reach.”

I also want to point out that spawning behavior and the Spawn Engine functionality are two different things. The spawning behavior (the experience that the user perceives) might actually resemble that of Halo 3 due to the weights assigned to the zones (I personally cannot say if this is true or not, I am not as familiar with Halo 3 as I am with Reach), but the functionality of the Spawn Engine appears identical to that of Reach and foreign to Halo 3. Specifically, when selecting the best Respawn Point to spawn on, Halo 4 and Reach both share a behavior of a randomizing tie breaking algorithm that was never present in Halo 3.

The Test Format

Tests were conducted with the following map layout. There are two Respawn Points, A and B, for the test subject to spawn at. Both would fall into a teleporter which would take the subject to point C in a Kill Boundary for instant death. The point of death would be at C, equal distance from A and B. This was to ensure that no matter how large the death influence might be, it could not bias the choice of which Spawn Point would be chosen.

For Ravine, the distance between A and B was well over 60 units, and C was over 40 units from either A or B.

For Impact, the distance between A and B was well over 120 units, and C was over 70 units from either A or B.

Within the test, Respawn Point A and B would be covered with different types and quantities of zones. Each test will be described with which zone types cover A or B along with their quantities, along with the results from those tests, and the conclusion we can draw.

With the suicide penalty set to zero, the average spawn was about 5 seconds, so the test was cycling samples quite quickly. The fall of the spawning player into the teleport was long enough to clearly note if they spawned at A or B. A single 10 minute test would result in about 100 sample points.

Defining Precision

We were told by Bungie that the Reach Spawn Engine applied a small random weight to all Respawn Points to work as a tie breaker, and that this random weight was about 1% of the Weak Respawn Zone. This random weight we will call Δ (delta). If a tie still occurs, selecting the first encountered of the highest weighted Respawn Points would still produce an apparent  random selection. (Tests indicated that the Halo 3 Spawn Engine did not have this feature in it, but instead would select from amongst the highest weights the first Respawn Point it encountered within the software’s Respawn Point collection object. With no dynamic influences present, the spawning experience reduced to having just one point to spawn on.)

We do not know the range of Δ in Halo 4, but we do know what we can expect to see if it is on any scale like it was in Reach – and we see those very things during these tests.

• I have confirmed that Δ is adequately random by observing statistically similar spawning at A and B when no influences are present.
• I have confirmed that Δ does not match the strength of the Weak Respawn Zone covering A by noting the player only spawns at A.

Now at this point you would be correct to ask,”How do we know if there is a Δ? What if A and B were simply randomly picked as the two Respawn Points tied for top weight?”

Because when you introduce gradient influences, you can adjust the amount or strength of influence on a Respawn Point by incrementally moving the source of influence toward or away from the Respawn Point. In turn, you can observe statistically favored selection of one Respawn Point over another – the statistical ratio of selection between them changes. On the other hand, without Δ, the width of this window of probable spawn selection would be essentially zero (think Halo 3).

Q is a negative gradient influencer. G and H are Respawn Points. H (our control reference) is surrounded by a zone of negative static influence, and G is somewhere within the influence of Q. By moving Q toward and away from G, we can identify the distance from Q in which its influence is equivalent to our control reference. Then by changing our reference, we can repeat the test. From these results we can plot the influence of Q as a function of distance.

The red dashed line is the influencers’ weights. The orange region is the potential weight when you add in Δ. The gray box is the window in which the potential of spawning can compete with H, though more likely at the far side and less likely at the near side. Moving G out of the window on the near side, G’s potential weight will never exceed that of H and H will always be selected. Moving G out of the window on the far side, G’s potential weight will always exceed H and G will always be selected.

This window helps us identify the distance of an equivalent weight compared to the control influence placed over H. It also gives us insight into the gradient of the influence by how wide the window appears to be. Given Δ is the margin of precision available to us, two weights can only be found precisely equivalent, not exactly equivalent.

Spawn Engine Influences

There are two types of influencers – static and dynamic. We will begin with the static, which are the four zones: Respawn Zone (RZ), Anti Respawn Zone (AZ), Weak Respawn Zone (WZ), and Weak Anti Respawn Zone (WAZ).

In Reach it was impossible to determine the relative weights between these zones, because the RZ was magnitudes greater than the AZ or WZ. In other words, there were not enough WZs to counter and precisely nullify a RZ, so we could not show that RZ = nWZ. But, [audley] claimed that the zones are all within roughly one magnitude of each other. It turns out that he was correct and as a result we are able to determine their relative weights.

CertainAffinity claims that (like in Reach) each canvas can have its own weights for each of the influences, to compensate for the size of the canvas itself. This is critical to understanding why canvases may exhibit different ranges or values of influence for death, zone, proximity, LOS, etc. (For example, the Enemy Proximity influencer was configured more narrow on Haven than on Ravine, because it is a smaller map.)

For these tests, the map concept is the same on all four forge canvases. Only the canvases for forging were tested (since they are the most commonly forged canvases). The results listed in this publication at this time include mostly Ravine and Impact. (Results from Impact are not as complete as those for Ravine. Only PE Primary radii were measured on Erosion and Forge Island at this time.)

Frame of Reference

We begin by taking the lowest influence and using that as our reference. By definition:

• WZ = 1

Zone Weights

With 14 WZ on B and 1 RZ on A, spawning appeared identical between A and B. I conclude RZ = 14. (This means that you can create the same influence of one Respawn Zone by stacking 14 Weak Respawn Zones on top of each other.) With 1 WZ and 1 WAZ on B and no zone on A, spawning appeared identical between A and B. With 2 WZ and 1 AZ on B and no zone on A, spawning appeared identical between A and B.

For Ravine and Impact:

• RZ = 14
• WZ = 1
• WAZ = -1
• AZ = -2

Rico_Respawn

With the introduction of the Ricochet game type, 343i has made a change to the weight of the Respawn Zone, but only under very specific circumstances.

When the RZ has the Rico_Respawn label applied to it and the RZ is assigned to red or blue team, then it will produce a weight onto the Respawn Points within its zone of just over 200 for the team it is assigned to. If the RZ is not assigned to a red or blue team, then it did not exhibit the heavy influence on any Respawn Point. At this time I assume that this heavy weight only occurs with the Ricochet game type, but I have not confirmed this point either.

This measurement took me quite a while to take, because I had to use numerous Respawn Zone objects; and when you have too many Respawn Zones on the map, then some of the zones are simply ignored. For this reason I had to rely upon the powerful flat portion of the enemy proximity influencer (PE Primary) that I discuss in detail later in this article.

On Forge Island, I had a unique setup that I used (not depicted here) in which A was covered with an RZ labeled Rico_Respawn for Red team. B was covered with 13 RZs. The total differential between A and B without the Rico_Respawn RZ under test is 13 • 14 = 182. If I added 1 PE Primary influencer then the difference becomes 202 and the red player spawned at A. When I added a second PE Primary influencer, The difference became 222, and the red player spawned at B.

If I removed one of the RZs from B, the difference became 12 • 14 = 168. With 2 PE Primaries added to A this difference became 208, and the red player spawned at A. With a third PE Primary added to A, this difference became 228 and the red player spawned at B.

The weight exerted by a Rico_Respawn-RZ, therefore is between 208 and 222. As you read more of this analysis on influencers, you will see why it is pretty much impossible to force a player to spawn outside such a Respawn Zone. And with such a large weight to overcome, there isn’t really any purpose of trying to be so precise in determining what the weight might be. You could just as well claim that the weight exerted is -200. As a forger, that is more than enough information to forge with and to understand what you see that seems strange at first when you test out your maps.

Dynamic Zone Influence

CertainAffinity confirmed that the zones do not change their weight regardless of whether there is an enemy or ally in them.[1] To demonstrate this, both A and B are surrounded by their own RZ, but the one covering A was expanded to cover over half the test area. While it did not stretch over B, it did cover a second player standing equal distance between A and B. This meant that the second player was in the zone covering A, but not in the zone covering B. It also meant that if there was any possibility of a proximity influence from this second player, it would be equivalently applied to both A and B.

In both the case of the second player being an ally and the case of him being an enemy, the spawning was identical between A and B. This demonstrated that the weight applied to A did not change relative to B. The zone influence does not change if an enemy or ally is present in the zone. This confirms what CertainAffinity has told us.

Without knowing the correct range of the ally or enemy influence, it is possible to look for and believe one sees a dynamic zone influence, when in fact one is really witnessing nothing more than a proximity influence. I will discuss the proximity influence next and will go into detail of the complexities in measuring them. It will become clear why the proximity influence can easily be incorrectly characterized, thus inadvertently giving rise to false conclusions for other tests of influencers’ functionality.

Proximity Influences

Remember I said that when you have two equal influences over A and B you will see identical spawning between the two? And if one is more than Δ more than the other (you are outside the window), you will see spawning only on one of them?  And do you remember I said that if you were in the window but not in the middle of that window you would see statistical favoring of one point over the other?

When you cross out of the region where Δ can permit spawning at either A or B, you have absolutely no idea what influence you are experiencing. In fact, the only thing you can say is that you are NOT experiencing a specific influence ±Δ. For example, when you think you are moving out of a window of -19 influence, you could simply be in a -18 influence, or suddenly in a -2 influence, or even at 0 influence (beyond the boundary of the influencer altogether) – they all show the same result when your control influence is -19, so you really have no idea what you are witnessing.

To plot the weights of the proximity influences of players and death, you need to find points of equivalence to multiple known influences (your control influences). You begin with the source of influence at A, then move away until you find the first point of equivalence and plot that, change the control influence, and repeat. But as we will see, it won’t tell us the entire picture, but will only give us a few sample points. From these points we “connect the dots” the best we can, but we have no certainty what shape the curve really looks like.

This is where inductive reasoning really shines, the author relying on it begins to lose all credibility by making outrageous claims, and I begin to cry as I question if I really want to go on reading any more. To save my own credibility, I will connect the dots only to paint an “effective” model to forge with, but do not make any claims I know for certain what the curves actually look like under the hood. And I will base those curves in part on what I learned from Bungie (e.g., that some dynamic influencers in Reach can have two tier weights).

Ally Proximity

With the ally at 0.5 units from A (standing right next to A), 4 WZ was applied to B and the player spawned only at B; 3 WZ was applied to B and the player spawned only at A.

• @ 0.5 units (spawning right next to an ally), 4 > PA > 3.

I found a point of equivalent influence for +3 at about 5 units from A, +2 at about 12 units from A, and +1 at about 18 units from A.

Do you see a pattern? It appears that the slope is roughly 1/6 (e.g., 1WZ/6 units). That would make the strongest point of influence nearest the ally look more like 4 than 3, with an overall radius of about 24. While there may be a curved slope, it may also be my test was too rough and my measurements may be off a bit. Ultimately, a straight line with a slope of 1/6 is more than adequate to characterize PA for a forger’s use.

From Ravine:

• @ 0.5, 4 > PA > 3
• @ 5, PA = 3
• @ 12, PA = 2
• @ 18, PA = 1
• radius is about 24?

Enemy Proximity

The same type of test was conducted for enemy proximity, using multiple WAZ over B. I found that the enemy proximity was strongest at his side between -19 and -20. I found that this influence remained between -19 and -20 until about 9.5 units from A, then most of this influence simply dropped away at 9.5 units from A. I then repeated with weights as low as -4 over B and still the influence looked like it dropped entirely away. It was not until I tried -2 over B that I learned that the influence of PE was not zero after 9.5 units from A, and that there was a gradient component to measure for quite some distance more.

I found a -2WZ point of equivalence at about 12 units from A, and a -1WZ at about 17 units from A.

Do you see a pattern? It appears that the influence of the Enemy Proximity is two tier – a near negative equivalent of the Ally Proximity, combined with a very strong 9.5 radius flat weight between -19WZ and -20WZ.

For Impact, I noticed a point of equivalence for -20 at 1 unit from A and -3 at about 10.5 units from A. I do not remember if I did not test for these on Ravine of if they were not present. These gradient tests are roughly measured and could be off by about a whole unit in distance.

Later I found this -20 value for Forge Island and Erosion as well, but also found that all I had to do was be within the radius of the PE Primary to see it, rather than only being within 1 unit of the Respawn Point. But I also found that I was just on the edge of -20 (remember the discussion of precision?).

From Ravine and Impact:

• @ out to 9.5, PE = -20 (Impact, Erosion, and Forge Island)
• @ out to 9.5, -20 < PE < -19 (measured on Ravine)
• @ 10.5, PE = -3 (measured on Impact only)
• @ 12, PE = -2 (measured on Ravine)
• @ 17, PE = -1 (measured on Ravine)
• radius is about 22?

If you do not take into account the full range of PE, you might place the enemy too close when searching for the behavior known as dynamic zone influence. Then seeing the total influence over A change from the mere static influence of the zone, you would erroneously conclude you are witnessing a dynamic zone influence at work, when in fact you are witnessing a proximity influence.

LOS Influence

28 Jan 2013 – I am updating this section with new information on the LOS Influencer.

My results for LOS Influencer are very close to what [Audely] presented, which tends to suggest that his test was based upon the same setup that my test had used – two Respawn Points, one as a control and the other as the subject of the test. The information I presented here only occurs in narrow situations that are simply not found in playlist maps. For completeness I wish to keep the original test documented here along with the conclusions. But I want to stress that you will probably never see these kinds of results as an affect upon your map. For a deeper understanding of the LOS Influencer, read the new information I present at the bottom of this section. Also, be sure to read up near the top of this article where I lament that I ever bothered to venture into these murky waters of relying upon empirical evidence to characterize the Spawn Engine.

The problem with trying to measure a relative influence of the LOS influencer is that the proximity influencer is mixed into the equation unless you move clear across the map (in this case in the center equal distance between A and B).

I will define the LOS the influence upon a Respawn Point that is seen when scoped, even along the outer edge. However, the influence along the most outer edge appears to be slightly weaker than in the center, clear visible area. This slight difference may simply be my perception, but if it really exists it is well within Δ – essentially zero, and not worth mentioning.

With +24 influence (zones) over A, LOS (scoped sniper) at near point blank (PE), player did not stop spawning at A. That is, the total positive influence of +24 was greater than PE + LOS + Δ.

With +23 influence (zones) over A, LOS (scoped sniper or scoped binary rifle) at near point blank (PE), player spawned evenly between A and B. DMR and Rockets resulted in spawning only at A, indicating less LOS influence by these weapons.

With +4 influence (zones) over A, LOS (scoped sniper, scoped binary rifle) at about 17 units from A (PE=-1), player spawned about evenly between A and B. DMR scoped required closer range at 12 units (PE=-2) indicating a -2 influence. Rockets scoped required closer range at about 15 units, indicating about -2.5 influence.

From Ravine:

• EnemyLOS.sniper and LOS.binary = -3
• EnemyLOS.dmr = -2
• EnemyLOS.rockets = -2.5

With -3 influence (zones) over A, ally was positioned near +3 influence distance from A, determined by equal spawning between A and B. Also, With -1 influence (zones) over A, ally was positioned near +1 influence distance from A, determined by equal spawning between A and B. In all cases, spawning of player did not appear to be impacted in any way by LOS of ally.

From Ravine and Impact:

• AllyLOS.any = 0

Update:

Now I want to present to you newly discovered information on the LOS Influencer, information I stumbled across only because the model I presented above proved to be faulty. How did I know it was faulty? Because when I was trying to analyze the first six maps going into the forge test playlist for spawn traps, my analysis showed that on Simplex that a team playing a 2v2 match on the map could prosecute a solid spawn trap. Well, the problem was that the LOS prevented the trap from being as solid as I had hoped. But the real question was why?

The LOS Influencer was pushing the spawning player away from the Trap Point. But if its influence was truly at most -3, then this should not have occurred – ever! Something was wrong, and I was mystified for hours. Then it dawned on me – I was witnessing the LOS bug, where the LOS Influencer can appear to overcome the Respawn Zone, even on Reach. So let’s begin by discussing what triggers this bug, because it turns out that it wasn’t even what I had thought.

In Reach, the Respawn Zone was so powerful that no combination of negative influencers could overcome it – you were going to spawn in your team’s Respawn Zone no matter what. But I found and JonnyOThan confirmed that you can actually overcome the Respawn Zone with the LOS Influencer IF you have more compatible Respawn Points outside the Respawn Zone than inside.

When I first discovered this bug, I reported it to JonnyOThan that I had one Respawn Point inside the zone and two outside behind my player. When I scoped the Respawn Point in the zone, the spawning player spawned outside the Respawn Zone.

I then wondered if it was simply more Respawn Points outside the zone than inside. So my next test was to put 5 Respawn Points in a zone very close together so I could overcome them with one scoped sniper, and 6 Respawn Points outside the zone behind me. The LOS overcame the zone again.

Do you see the problem with my test? It never was about how many Respawn Points were in a Respawn Zone, but how many were in the LOS Influencer cone region.

Take a look at the picture above. The PE from the enemy in the center covers A and C and drives their weights down to -20, but it doesn’t cover B, whose weight should be -3 from the PE secondary (drift) and -3 from the sniper. The spawning player should always spawn at B. But he spawns evenly between A and C. Why? Because the LOS Influencer in this configuration is extreme, and greater than any other influencer, even imminent danger.

With B covered in +300, the EnemyLOS drove the spawning player to A or C evenly, meaning EnemyLOS < -320. This makes the LOS Influencer the strongest on the map.

From Ravine:

• EnemyLOS < -300 (off the charts!)

I don’t want to end there, I want to finish describing this phenomena.

If you cover A and C with -60 (every Anti Respawn Zone and Weak Anti Respawn Zone), then the LOS.Enemy will still push a player to A or C, which indicates that the LOS was not simply turning off Respawn Zones over B, but truly overcoming B with enormous influence.

If you have no zones of any kind, and add a second Respawn Point at B such that both Respawn Points at B are within the LOS, then the spawning player only spawns at B, because the EnemyLOS only applies a -3 influence (for sniper for example).

How does this impact spawning analysis? That’s for another article for another day.

The other thing to note is that for AllyLOS, in the typical map where LOS only covers very few Respawn Points, the influence is like EnemyLOS, it is off the charts and will prevent a player from spawning in the LOS.

The only thing left to learn, and I suspect only from the game publishers, is why the dual behavior with these two influencers? Is this a bug or is there a purpose for these dual behaviors?

Ally Death Influence

I am not certain how well we can study the death influence, because unlike other influences, it is not just a function of distance, but of time as well[2]. We know that the death influencer will go away after a period of time, but it isn’t clear if it decays over time or suddenly disappears. Add to this the random selection of Respawn Points in our tests, measuring the effects of the death influence becomes very complicated.

We also do not know if the means of kill impacts the influencer’s weight (e.g., sniping versus assassination). For this reason, I will just sample some weights using sniper equal distance between A and B and call it a night.

A has +2 influences by 2 WZs. Red 1 jumps over the side. Blue snipes Red 2 on A. If Red 2 dies 2 seconds before Red 1 spawns, then Red 1 usually spawns on B, but sometimes on A. This strongly suggests that the death influence is directly centered over A will have a weight of about -2 for at least 2 seconds. (It could be more a fraction of second after death, it could be less 4 seconds after death, etc.)

The next test is to move the death influence (incrementally) from A until we detect a point of equivalence with only +1 influence over A – that is, the death influence is -1 at a given distance from the origin of death. I found such at distance at about 10 units from A.

For Ravine:

@ origin, Death.ally = -2, for unknown amount of time

@ 10 units, Death.ally = -1, for unknown amount of time

radius = 20 units ?

Suicide Influence

The suicide influence was tested using only one person, so it will only reflect the influence after 3 seconds. It appears to demonstrate a weight of -2 at the point of suicide. It probably is the same as an Ally Death Influence.

For Ravine and Impact:

@ origin, Death.suicide = -2, for unknown amount of time

Suicide influencer is probably the same as Ally Death Influencer.

Enemy Death Influence

Red1 standing midway between A and B snipes Blue next to A two seconds before Red2 spawns. Red2 spawns about evenly between A and B, indicating the Enemy Death Influencer yields a zero weight.

For Ravine:

@ origin, Death.enemy = 0

Odd Ball Influence

Odd Ball Influence is defined as a dynamic influence that moves with and has an origin at the odd ball.

The test map was modified with ball spawns at A and 1 unit increments away from A. With +5 over A, I observed equal spawning between A and B while the ball spawned anywhere from next to A out to 12 units from A. Beginning at 13 units from A, the player only spawned at A, suggesting -5 < BALL after 12 units from A.

I then decreased the weight over A by 1 for each successive test run until I reached 0.

For Ravine:

@ out to 12, BALL = -5

@13, -4 < BALL < -3

@14, -2 < BALL < -1

@15, 0 = BALL

To see if the ball in motion had an influence, the ball was spawned midway between A and B to eliminate its initial spawn influence. The enemy picked up the ball and moved to about 12 units from A, where its influence would be -2 (PE) + -5 (BALL). With +7 over A, I observed a very narrow window resulting in spawning at A or B. This suggests that BALL maintains its influence as expected. However, I did not sample other locations due to the complexities of PE.

Imminent Danger

Imminent Danger Influencer is defined as a dynamic influence created as a grenade, round, or rocket approaching a Respawn Point. You can imagine how tricky it would be to time the rocket approaching the spawn. I also used a Scorpion round in my testing, since it is slow enough to get a sense if it is far greater than a rocket or grenade.

With +140 over A, a volley of rockets from a Rocket Hog overcame A within one second of impact. Scorpion and Wraith did not. It may be that each rocket contributed to the total influence.

With +56 over A, Plasma Grenade within 1 unit of A, and both the Scorpion and Wraith overcame A within one second of impact.

These influences exist within a very narrow time frame and exhibit extreme influence over what is likely to be just a few units of distance. It is my opinion that understanding this type of influence is of little value to the forger, since there is nothing we can do to compensate. (You won’t lay 10 Respawn Zones over a group of Respawn Points would you?) And the scope of the influence is simply to avoid spawning at all costs into areas of immediate and imminent danger.

Ordnance Drop Influence

Ordnance Drop Influence is defined as a dynamic influence that is present when an ordnance drops at an Ordnance Drop marker, and as a static influence that is present around the Ordnance Drop marker. Since this Influencer is new to Halo, the goal of these tests are to first demonstrate if any influence even exhibits itself from an Ordnance Drop. And if it does, then characterize it as best as possible.

Given the complexities of trying to time a spawn as a drop is taking place (falling from the sky), I won’t bother trying to determine that aspect. It may be that it presents something of akin to Imminent Danger, since players can be hurt if they stand too close to the drop as it occurs.

With no zones over A or B and a drop next to A, I observed equal spawning before a drop, and after a drop.

For Ravine: OD = 0

Summary

Having ventured into waters that I think one should never wade into, I present to you my tests and my conclusions for peer review. As I said from the beginning, the results will be within an unknown margin of precision. And since I never want to present numbers with the deceptive appearance of precision, I find it more meaningful to say that the enemy proximity influencer is -20 than to say it is -19.5 or to say it is between -19 and -20. (The value -20 gets the point across, doesn’t it?) I highly recommend you shelve the entire article, print the summary picture below, and pin it up over your XBOX for when you forge. Because when it all comes down to it, the precision is over rated – all you really need are the general shapes and sizes of the influencers and how they all relate to each other and canvas distances to get a sense of what you can expect to see by the spawn layout you forge onto your maps.

The influences covered by this article include:

• AllyLOS(in the typical case) and EnemyLOS(in the typical case) off the charts < -300
• Enemy Proximity: roughly -20 out to about 9 units away, then mimics negative Ally Influencer to about 25 units away.
• OddBall: -5 out to 12 units, then drops to 0 for a total radius of about 15 units.
• EnemyLOS(in the special case): around -2 to -3, depending upon the weapon
• Ally Death (and probably Suicide as well): -2 at center of death, drops off to 0 after 12 units away.
• Anti Respawn Zone: -2
• Weak Anti Respawn Zone: -1
• Weak Respawn Zone: +1
• Ally Proximity: a little more than +3, slow drop off to 0 around 25 units.
• Respawn Zone: +14
• Respawn Zone assigned to Red or Blue team with Rico_Respawn label in Ricochet game type: +200
• Enemy Death, Ordnance Drops yields no influence.
• AllyLOS(in the special case) = 0
• Imminent Danger is extremely strong and extremely short lived.
• There has never been a Halo title that has demonstrated dynamic zone influence.

_______________________________

[1] Bungie confirmed that the zones in Reach did not change their influence when enemies or allies entered them. Tests conclusively prove that zones in Halo 3 did not change their influence when enemies or allies entered them either. In other words, contrary to forge lore, there has never been a Halo title in which a zone would dynamically change its influence for any reason. The closest one could point to is the zone enabling or disabling by the flag status when their game type labels were flag state specific, but even this is not technically considered the same as “changing weights”, and this specific functionality has nothing to do with whether a player is in the zone or not.

[2] The Death Influencer goes away over time. How fast and whether it decays or simply goes away entirely is not known. However, the fact that it does go away over time explains why immediate spawning results in safer spawns away from the action, while normal spawning demonstrates more spawning near allies even near the action.