How to kill shielded behemoth with silo

In terms of complexity of interactions, etc., static area shields have the advantage that you don't have to pay much attention to where they're deployed at all, so you can focus your APM on more interesting things than repairing your Stinger/Gauss.

If every unit were micro-intensive the game would be incredibly difficult to play.

The unintuitivity seems to come solely from tacnuke. What if tacnuke damage was just slightly above shield health so it always pierced?

just give us shield link turn on/off button already :P

Latest spring allows an interesting option [?] - Subtracting the shield level from the projectile's damage (and reducing shield to 0).

But shields are interesting.

Shields that just subtract damage are nothing more than a "DPS-Armor", i.e. anything below x DPS and y Alpha can't damage a unit. Unless they are completely drained, they'll always block a guaranteed amount of damage, even if it's a nuke. Current shields on the other side slowly lose their ability block different weapons as they deplete their charge. If you want a shield that can deflect a nuke, you'll need a bigger shield generator, the aspis just can't handle it (unless shields get changed which would make even thugs reduce a nuke's damage).

I think the current system not only adds complexity, but also depth. For example you can use an allied's artillery unit together with a well timed Eos to kill an enemy with 100000 metal in shields. With the proposed change this would be impossible, simplifying the game away from strategy towards economy(and macro/metal is already op).

quote: Shields that just subtract damage are nothing more than a "DPS-Armor"

Yes, they're consistent and intuitive. Not sure how that's a bad thing.

quote: Shields that just subtract damage are nothing more than a "DPS-Armor", i.e. anything below x DPS and y Alpha can't damage a unit. Unless they are completely drained, they'll always block a guaranteed amount of damage, even if it's a nuke. Current shields on the other side slowly lose their ability block different weapons as they deplete their charge. If you want a shield that can deflect a nuke, you'll need a bigger shield generator, the aspis just can't handle it (unless shields get changed which would make even thugs reduce a nuke's damage).

That's not exactly true. From what I understand, the way GoogleFrog implemented partial penetration is such that when a projectile hits a shield the shield only absorbs damage vs other shields. That may allow another shield to completely block the projectile, but if the projectile passes all shields then it still does full damage to whatever non-shield thing it hits. I don't know how it works when a shield completely blocks a projectile though, for example whether a shockley hitting a 3rd shield will do full damage to the shield or if it will only do its remaining damage vs shields, or what happens with aoe, etc.

quote: I think the current system not only adds complexity, but also depth. For example you can use an allied's artillery unit together with a well timed Eos to kill an enemy with 100000 metal in shields. With the proposed change this would be impossible, simplifying the game away from strategy towards economy(and macro/metal is already op).

A single tremor does enough dps to negate 7x shields worth of recharge and costs as much as 3x shields, making 2.3x cost. A single shockley can take out 3 shields and costs just a little more than 1 shield. In comparison being able to bypass 4 detris worth of shields with only a single arty and a single eos seems more like an exploit than a strategy.

quote: just give us shield link turn on/off button already :P

This is just passing on responsibility to players to micromanage their shields, rather than implementing a good design.

quote: Latest spring allows an interesting option - Subtracting the shield level from the projectile's damage (and reducing shield to 0).

I am aware of this and did not check whether Spring would actually support it. Consider beam lasers, they are always the problem. Aside from that I don't want to reduce the damage dealt by projectiles for two reasons:

• Visual consistency.
• AoE.

There is something wrong, visually, with a Glaive surviving a direct hit from a Tacnuke that just barely penetrated a shield. Also, reducing damage would imply a change to projectile destruction. Currently every projectile always deals its full damage to units within its AoE. This is why I am not worried about shields blocking nukes: no shield has large enough radius for blocking a nuke to be particularly relevant. Reduced damage partial penetration would imply that a fully blocked projectile disappears completely. I like projectiles exploding on shields because it creates intuitive soft-counters for shields which depend on your opponents unit placement. It allows large AoE weapons, such as Inferno, to be somewhat but not completely countered.

quote: I don't know how it works when a shield completely blocks a projectile though, for example whether a shockley hitting a 3rd shield will do full damage to the shield or if it will only do its remaining damage vs shields, or what happens with aoe, etc.

With partial penetration, if a projectile was blocked then the total charge spent among your shields was equal to its damage.

I made projectiles lose speed as they lose shield damage. This was intended as a bit of visual feedback but it turns out to have large effects on shield balance.

So if a shield is around 96%, the tacnuke will not explode on it, will go through, will visually make a normal explosion but with only 1-2% of damage ? That would not be intuitive at all, because the missile is there, in one piece, and it should then hit as a normal missile.

quote: quote: This is just passing on responsibility to players to micromanage their shields, rather than implementing a good design. or to automate.

A system that requires automation to not be too micro-intensive had better add something significant to the game. Otherwise you may as well skip the hassle and implement something abstract which achieves the same result.

Overdive was once implemented with a big sliderbar on every players UI. With this bar they could control how much of their energy went to overdrive. In theory players could choose to stockpile energy. However, in practice this choice added very little and we ended up with a mostly automated sliderbar. Nowdays you have very little control over your overdrive contribution because the rate is set within the overdrive gadget, beyond the reach of some theoretical choice. Overdrive is mandated to work a particular way and people have to plan with that in mind. It is better than the option of making a fairly trivial decision.

Tactical AI (the skirmishing and jinking) is automation which reduces the micro required to deal with range differences and slow projectiles. This automation continues to exist because differences in range and the ability to dodge projectiles form a significant part of intuitive unit interactions and balance. In theory you could balance a game of similarly-ranged perfectly homing weapons but I think it is more interesting to have the dodging and outranging.

So what would be the result of shield link control and shield link control AI? Firstly it would be really hard to write and configure such an AI. This AI would have some ability to decide which projectiles to block and which to let through, given a reasonable shield configuration. For example it might de-link frontline shields against small artillery to save its charge for a sudden transfer to counter Tacnukes. A powerful shield link AI would need to know whether damage on a unit is more or less desirable than less shield charge. This valuation task amounts to creating an AI for the whole game.

Tactical AI faces a similar valuation dilemma. It is possible to save or sacrifice a unit through choices in jinking and so it is vital that tactical AI be easy to override with move commands. Tactical AI does not need to be optimal because it is predicable, reasonably optimal for most situations and very easy to override. Players can use this predictability to rapidly decide when and where they need to intervene and override tactical AI. None of this would hold for shield link AI. There are many problems facing shield link AI:

• If shield link AI aims for predictability then it is going to fail in some situations.
• The factors in a shield link decision are fairly hidden so failure is going to look random.
• Shield link decisions hinge on the exact values of hard to access numbers. Players would not know when to intervene.
• A shield link UI would have to involve state toggles and state toggles are awful. The UI for a shield link override would be significantly worse than the tactical AI override.
• It is impossible for shield link AI to achieve optimality without solving ZK. Therefor it is still going to fail and the failures will look random.

We end up with something that looks similar to the current system except that shields are generally more effective at blocking, penetration is harder to predict, and there is a pressure to manually optimize your shields. Shield link control leads to few extra decisions and system gains little-to-no increase in intuitiveness. If the intention were to increase the effectiveness of shields then it would be much better to bake the change directly into their behavior.

Manual control is a doubled edged sword and often unwieldy compared to baked in behavior. Current shields cannot be said to have bad AI simply because there is nothing to optimize; they simply sit there and implement their preset behavior in a very predicable way. Players do not need to consider whether to link their shields. Instead they can take linking as given and focus on much more interesting decisions which, incidentally, come with a much better UI. Unit placement and movement is much easier to use than state toggles.