Alpha 4v68 Meta Analysis (update)

[arkhangelsk]

Let me put a ball in on BuckleUpBones' side, because I can kind of see a point. While I appreciate analysis, I can also see how it can be overdone and overshadow the overall holistic feel of the system as a whole. In other words, once you see a "bad number", it sticks to your mind like a dead pixel, and all considerations as to its contribution to the overall system, or that it might have a justification disappear. You see the bad number and your immersion is significantly reduced.

Here let me use the much-hated speed malus. On paper, assertions like speed matters not with accurate plotting (a.k.a. against a faster target I'll just pull more lead) sound convincing. Further, it can be argued on a Virtual Course plot, two ships paralleling each other at 30 knots produces the same plot as both at zero knots - a dot indicating zero relative movement. Thus, in theory the speed malus should not exist.

Okay, but historically there have been a number of instances that suggest that once speed gets above about 25 knots, the hit rate can get much lower than predicted. The speed malus plausibly reaches 90%.

It must be noted not all of them involved high relative speed. Nowaki was not that much faster than Iowa, so she cannot generate more than a few knots worth of range rate unless she charges Iowa which was clearly not what happened. On some virtual course plot she'll look just like a ~5-knot ship. In the Battle of Java or Komdandorski the ships were broadly paralleling each other for much of the fight, And by the way, this also shows that they weren't always destroyers or "violently maneuvering". The only common thread is that the ships are uh, fast. I don't think I've ever heard of thousands of shells being flung at a slow ship and missing since the advent of director firing, regardless of its maneuvers, aspect or size.

There are only so many times we can blame the poor gunnery of the ships involved (in effect, calling them "exceptions") before at least suspecting ... maybe the core problem is that they are fast, and the reasons we use to dismiss that being a factor aren't as strong in reality as they are on paper. The fact is that parallel @30 knots does not look the same as parallel @0 knots once you take your head out of the Virtual Course Plot and into the True Course Plot, and the assessment of course not always as precise as it might feel on paper.

There's also the point that historically, destroyers are not insta-killed by battleships. I really don't think they've ever been that hard to kill once competent FCS shows up in UA:D and you are willing to use the main armament, but they definitely die easier than ever in Alpha 4 (of the new missions, Modern Battleships vs 12 destroyers is by far the easiest to steamroll). Apparently, the size malus was effectively nerfed in the face of complaints of its theoretical implausibility, and we are seeing the results. We get rid of the speed malus too, and destroyers WILL start being insta-killed.

Look guys - if we don't want lots of dead destroyers, it's between the speed malus and the size malus. We already HAVE a maneuver malus too in case everyone has forgotten.

But of course, none of this seems to matter - people see "speed malus NINE-ZERO percent" and vision of all other things uh, disappear. There's not a lot of discussion of whether the game overall "feels" right in terms of how fast ships are dying overall. A disproportionate amount of discussion is on killing the NINE ZERO percent speed malus 😅

[Evil4Zerggin]

Thanks for the votes of confidence!

On 2/20/2020 at 2:47 PM, RedParadize said:

@Evil4Zerggin I like your analysis. But I believe your mathematics formula are not the ones used. At best, they only represent the result of your test, which unavoidably have a limited sampling. There is more than likely more variable than what we can see. We can doubt the accuracy of value like HP as they are most likely calculated on top of the rest. Having said that, its still useful to test and reflect on these result.

I started to do proper datamining, and it turned out my curve-fitting was exactly correct, except the CL displacement power is actually 0.603 rather than 0.6. I'll probably be taking a closer look at the source code in the coming weeks.

Edited February 23, 2020 by Evil4Zerggin

[Evil4Zerggin]

On 2/21/2020 at 1:26 AM, arkhangelsk said:

Apparently, the size malus was effectively nerfed in the face of complaints of its theoretical implausibility, and we are seeing the results. We get rid of the speed malus too, and destroyers WILL start being insta-killed.

I think size penalty is a better primary solution (though speed can help, just some of the problem right now is that it's too easy to make ultrafast BCs, and I think a quadratic increase in difficulty to hit with speed is more plausible than the capped-hyperbolic that we currently have). Right now the chance to be hit is proportional to 1 + displacement / 26500 tons*, which means that size only really matters for BBs. As far as displacement goes, a 200t TB is currently only 37% harder to hit than a 10000t cruiser based on its displacement alone.

* I haven't confirmed this by datamining yet but I don't expect it to be too far off.

On 2/21/2020 at 1:26 AM, arkhangelsk said:

On paper, assertions like speed matters not with accurate plotting (a.k.a. against a faster target I'll just pull more lead) sound convincing. Further, it can be argued on a Virtual Course plot, two ships paralleling each other at 30 knots produces the same plot as both at zero knots - a dot indicating zero relative movement. Thus, in theory the speed malus should not exist.

[...]

The fact is that parallel @30 knots does not look the same as parallel @0 knots once you take your head out of the Virtual Course Plot and into the True Course Plot, and the assessment of course not always as precise as it might feel on paper.

I agree, and in fact I'm not convinced these "zero relative movement" assertions make sense even on paper. The whole fire control problem is one of producing accurate plotting---it's circular to say "assuming we have accurate plotting, we can have accurate plotting." Or in the parallel-30-knots case: the entire problem is to determine whether both ships are moving at exactly the same speed and exactly parallel, as opposed to separating or closing at X degrees, or one ship traveling Y knots faster than the other.