Editing v0.34 Talk:Bolt

== Inconsistency? ==
"Tests show that where even masterwork wooden bolts bounce harmlessly off copper armor, iron bolts pierce without any issue."

"Tests show that even wooden bolts are effective against armored targets with the new release. Copper bolts have been known to pierce adamantine consistently."

These two statements flatly contradict each other, is either true? [[User:OrangePikmin|OrangePikmin]] 21:31, 7 April 2012 (UTC)

The problem is the assumption that adamantine is the best armor against bolts.  I have done arena testing which clearly shows it is not.  Steel and copper armor will both deflect 98% of pine bolts, while adamantine only deflects 58%.  This is quite strange, and I am not sure why this is happening.  My best guess is that, since adamantine is completely rigid and very light, it essentially converts the edged attack of a bolt into a blunt attack with the same force and contact area.  

Summary of Arena Testing:
* Dwarves armored with breastplate, helm, greaves, 2 gauntlets, and 2 low boots of indicated material.
* Armed with adamantine crossbow, and 100 bolts of indicated material.
* Great archer/marksdwarf, grand master armor user.
* Usually 20 dwaves from each side positioned on either side of the 3 tile wide trenches in the standard arena.
* Used a perl script to count the indicated words within gamelog.txt
{| class="wikitable"
|-
! bolt material
! armor material
! hits ("flying")
! deflections ("deflect")
! through armor ("through")
! headshots ("head")
! fatal headshots ("killed")
|-
| pine
| steel
| 2011	
| 1814	
| 29	
| 93	
| 0	
|-
| copper	
| steel	
| 856	
| 0	
| 772	
| 38	
| 26
|-
| iron	
| steel	
| 302	
| 0	
| 267	
| 22	
| 13
|-
| steel	
| steel	
| 319	
| 0	
| 295	
| 21	
| 13
|-
| pine	
| adamantine	
| 996	
| 533	
| 386	
| 51	
| 10
|-
| copper
| adamantine
| 708
| 0
| 649
| 36
| 26
|-
| bronze
| adamantine
| 569
| 0
| 522
| 31
| 23
|-
| iron
| adamantine
| 609
| 0
| 551
| 32
| 20
|-
| steel
| adamantine
| 663
| 0
| 594
| 36
| 17
|-
| pine	
| copper	
| 1390	
| 1241	
| 25	
| 61	
| 0	
|-
| copper	
| copper
| 579
| 0
| 522
| 26
| 11
|-
| bronze	
| copper
| 552
| 0
| 510
| 25
| 18
|-
| iron	
| copper
| 496
| 0
| 438
| 23
| 14
|-
| steel	
| copper
| 512
| 0
| 459
| 22
| 11
|-
| pine	
| pine
| 739
| 0
| 684
| 43
| 31
|}
I am continuing testing, and will try to add a more complete table of results to the bolt wiki page when I am finished.--[[User:Pirate Bob|Pirate Bob]] 12:37, 15 August 2012 (UTC)

I now have a fairly large sample of results, and it doesn't look like any armor does anything to stop metal bolts.  I kind of feel like going forward is a waste of time, since none of the armor actually, you know, does anything... --[[User:Pirate Bob|Pirate Bob]] 03:11, 17 August 2012 (UTC)

While it would be interesting, I unfortunately cannot test bone bolts in the arena, as they are not available.  I did test wood on wood (useless), and when I get the chance I will test pine and copper vs. leather armor (I strongly suspect leather is also useless).  I might also try testing other types (i.e. featherwood and whatever wood is densest) to get some idea of how bolt effectiveness varies with density, but doing more metal bolt tests when nothing stops them is not worth the several hours it would take.  

I think instead I will redesign my test: On one side I will place a fixed number of unarmed, but armored, dwarves, and on the other side of the moat dwarves with the appropriate bolts for ammo.  I will then record the average number of hits to kill the dwarf targets.  It would be nice to also record the number of hits until each dwarf first falls unconscious, since this is usually what really matters in combat, but this will require a much fancier script to analyze.  I am saving all of my logs, so I can always go back and re-analyze them for unconscious dwarves.  

I will of course also get the same data already in my table from these tests.  However, collecting data that says no armor deflects metal bolts is not interesting.  I want to be able to grade the relative (in)effectiveness of each armor, and also compare survival with armor to without armor.  If metal armors really do nothing to help survival compared to no armor, I would say that qualifies as a moderately serious bug, and I would proceed with trying to figure out how to modify bolt and/or crossbow  raws to get more sensible behavior.  --[[User:Pirate Bob|Pirate Bob]] 12:19, 17 August 2012 (UTC)

:Side note: it's "adamanti'''ne'''", not "adamanti'''um'''" - this is Dwarf Fortress, not Marvel Comics. --[[User:Quietust|Quietust]] 12:50, 15 August 2012 (UTC)

: This is great research, Bob. Please continue testing. Ideally test every combination of arrow material and every armor. try to include wooden, leather and bone armor. And Bone arrows, too, please. If I find the time, I may try to verify your results. I did not know that you have a gamelog.txt of the fights in the arena. I can write a Python script analyzing the figthts, then. --[[User:Nagidal|Nagidal]] 13:16, 15 August 2012 (UTC)

: I found that I need to slightly improve my script so that it only counts "deflect" and "through" in lines that also contain "flying *** bolt".  I thought that by putting the dorfs on opposite sides of a moat so there can't be melee combat I would prevent all non-bolt hits, but I forgot to account for them dodging into the moat and falling to their deaths. Also, in case it's not obvious to anyone, the number of deflections plus the number of armor penetrations does not add up to the number of hits because some parts of the dwarves (upper arms, neck) are not armored.  I could add chainmail to fix this, but I was more interested in the effectiveness of bolts against plate armor. I also was considering some of the tests with leather clothing underneath to see if this has any effect.  Especially for adamantine armor.  Theoretically it should help cushion the impact (real knights would always wear padding under metal armor), but I don't know if DF models that. --[[User:Pirate Bob|Pirate Bob]] 00:14, 17 August 2012 (UTC)
: I have updated my perl script to only count statistics for lines containing "flying", and I am working on filling out the table.  I wrote some macros to help with arena setup.  --[[User:Pirate Bob|Pirate Bob]] 01:45, 17 August 2012 (UTC)

== Decorated bolts ==

There used to be an exploit wherein encrusting bolts with gems gave them enormous value boost because decoration value is counted for every bolt in stack. As I get it, Toady fixed it by setting value multiplier for bolt decorations to 1/3 instead of 10. Should be this mentioned somewhere? Also, if you manage to get a stack of more than 30 bolts, this is still an issue - e.g., for bolts[60] decorations value is doubled. --04:17, 21 May 2012 (UTC)

== This armor is useless! ==

I did some arena testing of the number of hits required to kill unarmed dwarf "targets".  In all tests, I placed 224 targets, filling all squares adjacent to the 3-wide moat, and the opposite sides I place 242 shooters (I placed a few extra shooters by the "bridge" on the right side).  I then let the arena run until all targets were dead, and counted up the total number of hits (counted the word "flying" in gamelog.txt). 

{| class="wikitable"
|-
! Ammo
! Armor
! Dwarf Targets
! Hits
! Hits/kill
|-
| copper
| adamantine
| 224
| 3985
| 17.8
|-
| copper
| copper
| 224
| 3819
| 17.1
|-
| copper
| none
| 224
| 3894
| 17.4
|}

The numbers speak for themselves - armor is completely useless against even copper bolts.  There is no significant difference between any of the results - I think the small differences are just due to random statistics.  I will try to test this with a few more combinations, and maybe repeat a few of the tests for more statistics, but my preliminary results have pretty much confirmed my suspicion that armor does nothing against metal bolts.  This just leaves the question, [http://www.youtube.com/watch?v=aqoDL4KlxvY why do we even wear it] :P ? --[[User:Pirate Bob|Pirate Bob]] 20:40, 17 August 2012 (UTC)
For more science you could mod a little and test again. For example mod the bolts to blunt attack and see if it has any differences in piercing adamantine. Or make the contact area larger. Also I recall an old [http://www.bay12forums.com/smf/index.php?topic=89997.msg2488474#msg2488474 experiment]about weight affecting bolts, the result was no significant difference. So there is a paradox: if we accept that weight has no effect,  the pine bolts show meterial sharpness has some effect, while the variant metal bolts show that sharpness has little to no effect. So I advice do the experiment again with silver and adamantine bolts. --[[User:Lcy03406|Lcy03406]] 04:45, 18 August 2012 (UTC)

I have tried silver and adamantine bolts, and neither are ever deflected by copper armor.  I will post details if I have time, but posting to the wiki is rather cumbersome - I think I may start a forum post for more involved discussion, as this is proving more complicated that I initially expected.  I believe that the main thing that determines armor penetration and damage is kinetic energy.  A bow/crossbow applies a force of 1000 (units?), over some fixed distance (which is not indicated in the raws), such that the energy applied does not depend on mass.  However, there is also a maximum bolt speed of 1000 (units?), so if a bolt is very light (i.e. wood) its kinetic energy will just be its mass times 1000^2.  I also found that copper blowdarts are never deflected by steel armor, and was unable to find any manipulation of the raws which would allow projectiles to be defected regularly but still do modest damage to unarmored targets.  In particular, increasing the contact area to 20 just makes bolts *more* deadly, as they start to sever limbs etc, and are still never deflected by armor.  I am unsure if what Toady changed about bolts has anything to do with the raws, but I noticed that on the DF2010 wiki page is says the raws changed in 0.31.09, and it gives different values for the raws than used currently, so I'll have to try the old values.  Anyway, I have lots of data, but still far less than I would like, and I need to figure out a better way to discuss it, and perhaps get more people involved in collecting/analyzing it. --[[User:Pirate Bob|Pirate Bob]] 17:45, 28 August 2012 (UTC)
At the suggestion of a friend, I wrote a script which automatically tests every armor/ammo combination, so after I finish analyzing the results of that I will post complete results.  Only surprise was that adamantine deflects adamantine 100% of the time, but no other armor/ammo combination produced any deflections. --[[User:Pirate Bob|Pirate Bob]] 12:23, 29 August 2012 (UTC)

My results have proven somewhat complex, and I felt a forum thread might be a better place for detailed discussion.  I was also pleased to find that Zivilin had just posted similar results, so I added mine to his thread: [http://www.bay12forums.com/smf/index.php?topic=116151.0]--[[User:Pirate Bob|Pirate Bob]] 15:55, 10 September 2012 (UTC)