undocumented "thorns" damage



  • Is the ATTACK body part supposed to do retaliation (thorns) damage? I noticed in several of my battles, that a melee creep attacking another melee creep (without the second one actually attacking back) receives damage which correlates to the attackee's attack power. It does not apply, when the attacker stands on ramparts though. Whether the ramparts received that damage instead, I did not investigate.

    If so, then this should be clearly documented. Also this makes RANGED_ATTACK an even stronger choice, since it is not affected by this thorns damage at all.

    I would like to see these kind of mechanics (if they are indeed intended) on the TOUGH bodybart instead, since really nobody uses this, because it's just a waste to use TOUGH - regardless it's price.

     

    See:

    https://screeps.com/a/#!/history/W19S13?t=4738050

    On tick 4738051 the melee attacker receives 450 dmg, without actively being attacked. This would match the 15 ATTACK parts of the attackee.



  • It's not documented, but afaik that is the intended behaviour (melee attack another creep with attack part will cause damage to the attacker). If the attacker is sitting on rampart, the rampart will take that damage. That's why you might not want to have melee creep being defender: the rampart will go down twice as fast. 

     

    As for TOUGH body part, couldn't you just put them at the start of the body array when creating the creep? When creep is attacked, they will lose part in the order that you specify in body array.



  • 1) still imo worth it, because you poke down attackers so quickly, that this is effectively evening out the extra damage from your creeps

    2) still useless, because you could as well just declare move parts first, and have pretty much the same effekt.

    And if it's really intended behaviour, it should be documented. Since this was not the case in the beginning, and was "introduced" without mentioning it - at least i'm not aware of, and I play this game since the beginning - it might as well be some kind of bug.


  • Dev Team

    Yes, this works as intended. This mechanic is in the game since the first prototype in November 2014. We have mentioned it in attack method docs just now.



  • To combat NhanHo's comment:

    The creep that is attacked doesn't send back retaliation against all those that attacked it right? It's just auto attacking one of the creeps that attacked it if it hasn't attacked already? Why would the rampart go down twice as fast, unless you're combining ATTACK and RANGED_ATTACK components in which case you can only do one or the other each tick without this funtionality?

    It'd be a bit strange if you could output more then 30 damage per part of ATTACK per tick.



  • @RaskVann The creep retaliates to every attacker, so if you have 8 creeps with ATTACK parts, attacking your one creep with ATTACK parts, it will retaliate to ALL of them.

     

    @Artem thanks for adding this to the documentation. However the description is still very vague. Could you please improve the description of the `ATTACK` body part at the top, stating answers to the following questions:

    1) does the ATTACK part let the creep retaliate to more than one attackers per tick (my tests confirm this)

    2) does a creep on rampart retaliate (it should not, because technically, the rampart is attacked, not the creep)

    3) what about retaliation damage against an attacker that stands*on* rampart - does it damage the rampart? (I was unwilling to test it, because sim mode does not let me set the hitpoints of a rampart)

     

    Also I think, it should retaliate to ranged attackers. (if they are in in adjacent squares only?) (Perhaps only with 50%?)



  • @chris, Thanks for the information Chris, it helps.

    @Artem The documentation helps, I'd still agree with Chris on more disclosure, we know this is a pain to stay on top of and appreciate the work.

    Chris's information also brings up a disturbing issue. New players will never have a chance against higher level units.  The only possible way to defend as a lower level room is having less space to travel allowing larger groups of smaller units to overwhelm the attackers. Since these lower levels don't have the capacity to build larger units they don't have a choice but to go for numbers as long as energy permits. 

    With this in place attacking a unit with more then one unit MULTIPLIES the damage of the outnumbered unit. You've made the unit more effective because of a choice the enemy made accidentally. A few rules surface with this in place:

    • Never attack a unit with more then one of your units, you make them more effective.  This can only be ignored if you can kill the unit quickly and then recover by healing after the fact. (Benefits larger player, why do they need more help?)
    • Make units as large as possible is always the best move since you can potentially take advantage of this effect. The only way to avoid this is to ignore melee units entirely in favor of ranged attackers. (Why are we putting so much emphasis on ranged attackers?)

    This makes melee units effective for defense which is interesting as long as the attacker has to either attack a specific point or has more units then the defender. I would argue that the extra steps this adds to combat aren't intuitive or fun simply because the extra defense isn't enough of a interesting mechanic to outweigh the other costs.

    A 'Thorn' mechanic could be very interesting for a building or power creep but I don't think it fits in the default ATTACK body.

     

    For clarification, a unit that does not attack that tick should automatically attack one of its attackers if it has the capacity. In my opinion it shouldn't attack all of them by becoming more powerful for that tick.



  • @Raskvann: right now the game is defender biased, even with the thorn attack include. Removing it will cause even more of a stalemate. 



  • As stated in other threads, this further increases the discrepancy between the bodyparts. ATTACK is now quite strong, but still significantly inferiour to RANGED_ATTACK (but the price actually might make up for this). However, with these new insights, the TOUGH bodypart seems even more  useless. At the current state, you could as well remove it from the game. It needs some serious overhaul IMO.

    Possible suggestions for the tough part:

    - remove weight

    - move the relaliation mechanics from ATTACK to TOUGH, and include retaliation to RANGED_ATTACK. It could reflect damage based on either the amount of tough parts or on the amount of attack parts of the attacker or a combination of both. It could be flat or a percentage value.

    - add damage reduction (again, might be flat/percentage)

    - something I did not think of???

     

    @RaskVann some kind of "automatic attack" is not feasible because it would totally counteract the game mechanics.



  • @chris: do you mind explaining your statement earlier about putting TOUGH part at the start still being useless because you can put MOVE at the start? What I was trying to say earlier is that you can have TOUGH and MOVE both at the start to buffer the damage before it get to other parts (in other words, TOUGH part = cheap 50 hp for your creep)



  • TOUGH = 100HP, 10 energy, adds weight

    MOVE = 100HP, 50 energy, gives move

    If you see both bodyparts next to each other, you can see what I mean. The 40 energy you save cannot make up for to the (useless) weight it adds. On RCL8, you have 12900 energy available, 25xM,25xA costs 3250 energy. You basically could build nearly 4 of those (missing 100 energy). This gives you 25 Parts of buffer before your ATTACK suffers. Now if you build something which we could say is somehow similar: 15xT,10xM,25xA you pay only 2650 energy. 600 Energy saved, but 300% movement speed lost (and even worse on swamp!!). And you would only be able to build 5 creeps for that (350 energy missing). 

    If you are defending, this movespeed might perhaps somehow work, but if you are attacking it probably won't. And *if* I defend, I'd rather go 10M,40A instead of the above.