Update: My data were collected in Testing Grounds, but WardenWolf has confirmed some of the data in live play. See http://www.youtube.c...?v=lJQxOdRS3Ug. Many other confirmations can be found in the replies.
Update: A section on weapon stats was added.
Update: See Paul's [first] response, which is quoted at the end of this post.
Update: Results of testing in live matches presented here: http://mwomercs.com/...-test-results/.
Test Results & Discussion
In reading the following, keep in mind that missiles do splash damage and that the amount done (as well as the components damaged) depends on the missiles' impact point, which is random. (I did make sure that all missiles hit their targets.) The inconsistent damage among the various tests is a reflection of the game mechanics. See the Methodology section for more information on testing procedures.
I conducted the first tests from a CPLT-A1 with 6 x Streak SRM 2 positioned 20 m in front of an AS7-D (921 HP) and aiming (roughly) for the torso center of mass. The choice of aimpoint appears to have no major effect on the impact point--the missiles go where they will. The results were as follows:
[The numbers in this section were originally slightly off due to arithmetic errors in computing the total HP for each 'mech. What was I doing? All incorrect originals have been crossed out and the correct numbers inserted. Sorry for any confusion this may have caused.]
50 missiles fired per test
80% - 3.7 damage/missile
80% - 3.7 damage/missile
70 missiles fired
70% - 3.9 damage/missile
weighted (by number of missiles fired) average over all tests =
Next, I used the typical RVN-3L build (dual Streak SRM 2 and 295XL engine) and fired while orbiting at close range to get results representing typical combat use. Generally, I made no effort to distribute the damage evenly and simply fired as often as possible while maneuvering around the target. The exceptions are the Commando and Jenner tests, where the missiles were fired mostly from the front, so that I could avoid prematurely blowing off limbs and ruining the data. The results are given by target chassis.
6 missiles fired -- 48% --
6 missiles fired -- 56% --
6 missiles fired -- 56% --
6 missiles fired -- 54% --
In the following test, two missiles each were fired from the left, front, and right of the target, in order to get a better feel for what uniformly distributed damage would look like:
6 missiles fired -- 63% --
Streak SRM damage is amplified by
JR7-D (253 total HP)
14 missiles fired -- 59% -- 7.4 damage/missile
20 missiles fired -- 51% -- 6.2 damage/missile
16 missiles fired -- 61% -- 6.2 damage/missile
Damage against the Jenner is between 150 and 200% higher than would be expected.
20 missiles fired -- 67% --
18 missiles fired -- 68% --
Damage is again amplified by a considerable margin--
100 missiles fired -- 47% --
This time, the overshoot is only about 30%.
CTF-1X (575 total HP)
68 missiles fired -- 54% -- 3.9 damage/missile
64 missiles fired -- 54% -- 4.1 damage/missile
About 60% too much damage.
102 missiles fired -- 61% -- 3.5 damage/missile
About 40% too much damage.
Clearly, the mental picture that most of us have had about how Streaks deal damage--that is, 2.5 total damage/missile spread over various components--is wrong. Streaks seem to be doing too much damage because this is, in fact, what is happening--not merely because it is annoying to be killed by them and we are naturally inclined to make excuses.
I conducted all tests in the Testing Grounds mode. I have assumed that all 'mechs present have their stock armor layout. This can easily be checked by firing, at a given location, some combination of non-missile weapons that does the same amount of damage as the location has armor points. The internal structure may be checked in the same way after the armor has been removed. In doing a few of these checks, I did not notice any discrepancies.
Clearly, I have also assumed that the damage system works (in its other particulars) on Testing Grounds in the same way that it does in multiplayer matches. This is harder to check, but I am aware of no reason to doubt the assumption.
The total HP for a given mech may be calculated as follows:
[total HP] = [stock armor HP]+[total structure HP],
[total structure HP] = ([max. armor HP]-18)/2+15.
This is because
- the head always has 15 structure HP and 18 maximum armor HP, and
- otherwise, the amount of internal structure HP for a given location is always half the maximum armor HP for that location.
In testing, it is important not to blow off limbs with HP remaining, because not all of the resulting drop in the health percentage is due directly to weapon damage.
Comparison with Medium Laser
In light of the following comparison, it is almost unnecessary to make an argument that the Streak SRM 2 is an unbalanced weapon.
Medium Laser + 2 DHS
- 1.25 damage/s up to 270 m
- damage tends to be spread over the target
- damage tends tends to be reduced vs. fast-moving targets
- moderately difficult aiming
- roughly 400 m effective range
- 0.72 heat/s
- 3 tons / 7 slots total
- "infinite ammo"
- no risk of ammo explosion
- roughly 2 to 9 damage/s, depending on target; 4 damage/s typical
- damage tends to be spread over the target
- damage is hugely amplified vs. small (and typically fast-moving) targets
- extremely easy aiming (only target lock required), but disabled by ECM
- 270 m max. range
- 0.57 heat/s
- 2.5 tons / 2 slots total
- ammo is good for 3 min 42 s of continuous firing
- low risk of ammo explosion (with typical ammo placement)
The splash damage system likely works (roughly) as follows:
- Determine missile hit location.
- Find the distance from this point to the center of mass (or some other reference point) of every other hit location of the target.
- Reduce the damage (from a maximum of 2.5) based on these distances, and apply the result to each location.
Clearly, compact 'mechs (not necessarily low-tonnage 'mechs)--those with many hit locations close together--would take the most damage under this system. For example, the 15 damage typical on the Commando--a very compact 'mech--is consistent with taking the full 2.5 damage per missile to six locations simultaneously.
How might it be fixed?
Remove splash damage. Every missile deals its damage to the location it hits.
Other Interesting Stuff
- Based on my tests, conventional SRMs deal damage in the same way as Streak SRMs, except that you have more control over the impact point at close range. If, in your Splat Cat, you alpha an Atlas in the CT at about 20 m range, you will do roughly 160 damage (4.3 damage/missile), not 90, as is typically believed.
- LRM damage is also higher than would be expected. Firing an LRM 5 at an Atlas from about 200 m, I observe 3.1 damage/missile (not 1.8) as a typical value. Doing the same thing versus an Awesome at about 700 m, I observe about 5.8 damage/missile. I cannot be sure that all missiles hit, so the actual damage per missile may be even higher (although only the missiles that hit are important).
- Try walking right up to one side of an Atlas in a Catapult. Fire an SRM 6 into its knee from a distance of 15 to 20 m. You will find that it takes many more shots than would be expected to destroy the leg, even though it appears that all the missiles hit. Edit: I just tested this in an A1, and it took 11 SRM 6 salvoes (66 missiles) to remove the armor from one of an Atlas's legs, each of which has 82 points of armor. That's about 1.2 damage/missile. I have only noticed this when shooting an Atlas in the knee from the side; shooting 'mechs in the torso has always done more damage than expected. The Atlas leg bug may well have nothing to do with the overall point of the post. [Edit: This is a known issue.]
Some people have calculated damage per missile numbers from their stats and found values closer to the nominal ones. For the record, here are my current missile stats:
LRM 15: 1620 fired, 179 hit (lol), 206 damage -- 1.15 damage/missile
SRM 4: 502 fired, 302 hit, 808 damage -- 2.68 damage/missile
SRM 6: 3540 fired, 1977 hit, 4739 damage -- 2.40 damage/missile
Streak SRM 2: 1612 fired, 1274 hit, 3884 damage -- 3.05 damage/missile
On Page 9, OneEyed Jack reports (for the Streak SRM 2) 502 fired, 333 hit, and 1041 damage. This yields 3.13 damage/missile.
These numbers are lower than the ones obtained by firing at stationary targets. Why?
The answer is probably that my experiments show what happens when missiles mostly strike the torso directly, but the statistics count all hits. A more detailed explanation is given in the next three paragraphs.
At the end of the Other Interesting Stuff section, I reported that shooting an Atlas in the knee results in around 1.2 damage/missile, or about half of the nominal 2.5 damage/missile. I just tried (in Testing Grounds with an SRM 6) shooting a Catapult below the knee, shooting an Atlas in the forearm, and shooting a Centurion both in the forearm and below the knee. In all cases, it took more salvoes to remove the armor than than would be expected if all missiles did at least 2.5 damage to the location struck. This is in marked contrast to what typically happens when the missiles are fired at the torso center of mass, where we get a moderate to massive amplification of damage, depending on the target 'mech.
Also, note that missiles that strike nearby terrain deal a small amount of splash damage (this is especially true of LRMs). If this happens, the missile responsible is probably counted as a hit in the stats.
Finally, note that most of the larger 'mechs saw only a modest amplification of damage in the "ideal" case of most missiles striking the torso (e.g., Catapult 20%, Atlas 40%). Hits against Commandos, Spiders, etc. will typically form only a small part of a given player's total SRM hits, with hits against larger 'mechs forming the majority (because these are both more common and easier to hit). Under these circumstances, it is not hard to see how the resulting considerable (but not ridiculous--say 50%) average amplification for torso hits could be offset by the extremity-hit and near-miss-splash-damage effects, winding up (for SRMs) in the 2.5 to 3 damage/missile range in the stats.
In summary, I suspect that we are not seeing (for SRMs) 7+ damage/missile all the time in the stats, because
- hits to extremities often deal sharply reduced damage,
- near misses that strike objects/terrain and result in minor splash damage are probably counted as "hits," and
- not every 'mech is a Commando.
- a single SRM that strikes a Commando in the torso usually deals about 15 damage,
- smaller but significant damage amplification occurs for many other 'mechs, and
- missiles that strike the extremities of many 'mechs deal significantly less damage than expected, making it more difficult to damage limbs (even when all missiles hit) than with other weapons.
Paul Inouye's response, quoted for your convenience:
This problem has 2 levels.
First is that Testing Grounds has quite a few issues when reporting damage and the numbers you're seeing are inflated quite a bit (almost double). We will be addressing this bug and others as Testing Grounds matures over time.
Second, this does NOT eliminate the findings that S-SRMs AND SRMs are doing more damage than intended. This is not due to some top secret, behind your back weapon balancing. It has to do with splash damage, how it was first implemented and the new smaller Mechs coming out.
Here is one of the scenarios described and I've turned on the debug tools to let us see exactly what is going on in terms of hits and damage being done.
The Raven 3L has just fired 1 volley of 2 x S-SRM2 at the Commando 1B. As you can see, the amount of damage done to the Commando does not make sense. There is a total of 51.5 armor being stripped off the Commando. We've been able to reproduce this repeatedly and we're getting an average damage of 12.9 per missile. Quite a bit higher than the intended 2.5 damage per missile plus splash damage.
So what has happened to cause this? Smaller Mechs and more complex geometry than what was available when the splash damage system first went into the game. When SRM splash damage went into the game, there were a total of 4 Mechs available to the playerbase. The Jenner, Hunchback, Catapult and the Atlas. These 4 Mechs have very unique targeting silhouettes and were used to calculate the radius of splash damage per missile. Now what has happened is that the splash damage across smaller Mechs or Mechs with more complex/tighter component positioning are getting hit with more splash damage than intended.
In the image below, you can see how much overlap the damage done to the Commando has and how that it is taking significantly more splash damage than it should.
We are looking at the tuning for these hit locations/splash damage and will update as soon as possible.
Edited by Amaris the Usurper, 18 March 2013 - 01:40 PM.
update from Paul Inouye