These forums have been archived and are now read-only.

The new forums are live and can be found at https://forums.eveonline.com/

Ships & Modules

 
  • Topic is locked indefinitely.
 

Titan changes - update

First post First post First post
Author
Previous Topic Next Topic
I'm Down
Macabre Votum
Northern Coalition.
#1061 - 2012-04-20 23:33:28 UTC
Kadesh Priestess wrote:
I'm Down wrote:
a would represent angular velocity, or transversal velocity
b would represent turret tracking
x would represent ship signature
y would represent gun resolution.

If you notice, signature has no means of approaching infinity in game.
Tracking, had a way to approach 0.
If you just approach zero - nothing will break, you need to make the value negative.

Anyway, what does it change? Nothing, it's question of tools used to change any member of formula. I think i clearly said that TD had such tools while others didn't. Availability of tools to achieve negative tracking doesn't turn tracking speed iself into unique member of formula.



dude, plug any number / 0 on your calculator.
Kadesh Priestess
Descendance.
GoonSwarm.
#1062 - 2012-04-20 23:42:51 UTC  |  Edited by: Kadesh Priestess
I'm Down wrote:
dude, plug any number / 0 on your calculator.
Dude, show me TD which subtracts exactly 50% tracking (exactly 100% under c6 magnetar effect).

Just checked killboard, aharm guys used meta4 TDs which with spec @ V in c6 magnetar had effect of -20.1×1.25×2×2 = -100.5%, which turns any tracking speed into relatively small negative value. And my calculator is perfectly fine with it.

ps Well, probably it's possible that value was capped somewhere, and when ZeroDivisionException was caught - eve assumed chance to hit = 1, but then why ccp just didn't change the cap? Also, according to my experience CCP doesn't cap values from bottom side, they cap'em only from breaking through roof (lock range @ 249999, shield/armor resonances @ 1), so cap hypothesis sounds suspicious too. All our speculations are far from truth imo, both mine (because squared negative member is positive member) and yours. Plus, I heard several times that widely used turret damage formula may be not accurate, if this is the case - it's even more stupid to base speculations on top of it.
steave435
Perkone
Caldari State
#1063 - 2012-04-21 10:12:31 UTC
Yaay, you can't mathematically deduce anything from that bug. What probably happened was that the tracking variable was defined to have a value between 0 and X, so that a series of bits like 000000000000000000000000 would mean 0. When something manage to reduce the value below that anyway, things break and the formulas can't be relied on anymore.

You are just flat out wrong. Greyscale has access to the 100% definitely accurate formula, and he has math experts there to verify the numbers in them. Just stop talking.
Civ Zomas
Deep Space Engineering INC
#1064 - 2012-04-21 11:06:13 UTC
Fixing the tracking

  • Titans get new "role bonus" of minus 100% effectiveness to Tracking Speed Bonus from modules. Optimal Range and Fall-off bonuses apply as usual.
  • EWar immunity extended to the Tracking Speed Bonus of Tracking Links.

Done.

You could also exclude other local bonuses and assists through these methods. (*cough*scanresolutionbonus*cough*).

While we're at it, let's fix doomsday devices

  • Firing the DD invokes the "no assists" limitation for a period, like siege and triage.

I don't know exactly how long would be appropriate, but certainly more than the 30s immobility timer. I'm thinking a few minutes, but not the whole cool-down period.

Titan: "PPPPPPPPPEWWWW! EAT IT!"
Remaining dreads: "Damn, but now it's our turn. Let's see how good your local rep is." Twisted

Yeah, I'd like to see that.

Refitting in combat
Not broken. Leave it alone.
Kadesh Priestess
Descendance.
GoonSwarm.
#1065 - 2012-04-21 11:08:12 UTC  |  Edited by: Kadesh Priestess
steave435 wrote:
tracking variable was defined to have a value between 0 and X
You don't know how python's float works, such lower cap at 0 should be enforced manually, it's not part of data format. There's some extremely minor data loss implied by format itself, but it's definitely not enough to force td strength rounding to -100 / tracking speed rounding to 0.

And if there's some artificial cap on number, like you assume, then why ccp just didn't change cap from 0 to, say, 1E-10, and removed magnetar's ew bonuses instead?
Mechael
Tribal Liberation Distribution and Retail
#1066 - 2012-04-21 11:23:12 UTC  |  Edited by: Mechael
I'm Down might be rather (somewhat understandably) disagreeable, but he's right. Signature currently means fuckall when there are a lot of ships on the field. All it does is multiply the effect of tracking, and at medium/long ranges tracking is not an issue (therefore sig is not an issue.)

Big guns need to have their chance to hit little targets at medium/long ranges reduced ... not a damage reduction like missiles, a chance-to-hit reduction. Period. Modifying the sig resolution of the turret over distance is the best way to go about this. The gun is calibrated to have a certain sig res at its optimal range (400m sig res for a 1400mm Arty, for example.) Closer than that, the sig res effectively decreases (making it actually easier to hit small targets, while tracking conversely makes it more difficult.) Farther than that, sig res effectively increases (making it harder to hit small targets, despite tracking, as that tiny little cruiser starts to look more and more like a speck in the distance.) The smaller a target is, and the farther away, the more precisely the barrel of your gun must be angled after all. Essentially this mechanic can replace falloff, and ships/modules/etc that give bonuses to falloff can instead give bonuses to turret sig resolution.

If it's not that exact change that's due for the long term, it's something very similar to it. Artificially adjusting damage is ridiculous, unrealistic, immersion breaking, and just plain not cool. Adjusting chance to hit based on size and distance in addition to tracking on the other hand ...

Sig res needs to be a standalone modifier for your chance to hit and not a modifier of a modifier (tracking,) any which way you slice it.

EDIT: Basically, it should look something like this: chance to hit = (effective tracking modifier) + (effective sig modifier) + (effective range modifier), where each modifier can be set up however you like it. Alternatively you can have sig affect both tracking and range as I suggested (but certainly not only one of the two, as it is now.)

Whether or not you win the game matters not.  It's if you bought it.
steave435
Perkone
Caldari State
#1067 - 2012-04-21 11:30:51 UTC
Kadesh Priestess wrote:
steave435 wrote:
tracking variable was defined to have a value between 0 and X
You don't know how python's float works, such lower cap at 0 should be enforced manually, it's not part of data format. There's some extremely minor data loss implied by format itself, but it's definitely not enough to force td strength rounding to -100 / tracking speed rounding to 0.

And if there's some artificial cap on number, like you assume, then why ccp just didn't change cap from 0 to, say, 1E-10, and removed magnetar's ew bonuses instead?

Regardless, when variables hit values that should not be possible and were thus not accounted for, so it starts throwing errors etc and just generally means that it has no bearing on the normal mechanics.
IIRC it worked if you went with the optimal range script as well.
Kadesh Priestess
Descendance.
GoonSwarm.
#1068 - 2012-04-21 11:40:49 UTC
steave435 wrote:
Regardless, when variables hit values that should not be possible and were thus not accounted for, so it starts throwing errors etc
It seems you've got no coding experience, lol.

Float as data format has its own pecularities, it doesn't care about yours 'should not be possible'. To make it not possible - you have to limit it yourself. I can hardly imagine that someone placed lower cap on tracking in formula's scope equal to zero - really hard to make such mistake having formula just few lines above/below.

And Dogma itself doesn't place any lower caps on any attribute, it supports only upper cap restrictions, thus if tracking was capped - it was special handling for turret damage calculation.
ilammy
Amarr Empire
#1069 - 2012-04-21 14:44:22 UTC
Civ Zomas wrote:
While we're at it, let's fix doomsday devices

  • Firing the DD invokes the "no assists" limitation for a period, like siege and triage.

I don't know exactly how long would be appropriate, but certainly more than the 30s immobility timer. I'm thinking a few minutes, but not the whole cool-down period.

Titan: "PPPPPPPPPEWWWW! EAT IT!"
Remaining dreads: "Damn, but now it's our turn. Let's see how good your local rep is." Twisted

Yeah, I'd like to see that.
This. Usage of doomsday will me more tactical (use it nao if you aren't afraid of losing your Capital Spidertank of Immortality for some time for the sake of wrecking enemy's spidertank, or continue to slowly push though it), rather than only strategical (use it now if the circumstances don't allow you to lose the fight in 10 minutes).
Andy Landen
Deep Core Mining Inc.
Caldari State
#1070 - 2012-04-21 15:24:56 UTC  |  Edited by: Andy Landen
Is there much resistance to removing DD ability to target ships? If so, how about we give sub-caps their own similarly game breaking "I win" module. IF and ONLY IF, we keep the Titan DD, I propose that the Marauder (T2 BS) be allowed to fit a new module called the "Mini-Doomsday Device" which operates similar to the Titan's DD, but inflicts proportionally less damage: 200,000 hp of damage and can NOT be used in HS. Cycle time would be the same as for the DD and it would not count as a turret or a missile launcher.

Or does this mini-DD proposal cause you to reconsider the merits of the DD? What do you think?

"We cannot solve our problems with the same thinking we used when we created them." Albert Einstein 
NeVeH
Sebiestor Tribe
Minmatar Republic
#1071 - 2012-04-21 15:42:23 UTC
I'm Down wrote:
Actually Raivi, when it comes to super capital experience, not only do I have have the same anti fleet experience as both of them, I've also got quit a bit more super on super and super on capital experience than either. I mean honestly, they've been involved in 1-2 major super fights between them, not counting the fountain black screen. I've got 4-5, all historically the largest in game to date, with the one exception of the WN massacre in Venal, which was also a blackscreen of death fight if I'm not mistaken.


lmao... get out of my corp.
Bouh Revetoile
In Wreck we thrust
#1072 - 2012-04-21 19:57:05 UTC
Tracking formula discussion already happened 20 pages ago. Please, read.
pmchem
Brutor Tribe
Minmatar Republic
#1073 - 2012-04-21 20:43:01 UTC
I'm Down wrote:

The other thing your PHD wizz didn't account for is that tracking grows with range and therefore is not a static variable. While signature is a static variable that never changes over the course of range. What this means is that the part of the formula that accounts for tracking varies and has more weight over range than the part that accounts for signature and remains static.

Tracking can in fact grow by as much 25000% in that multiplier from 1km range to 250 km range. Over the same course of range, signature will not change.

When you introduce webs in at close range, the same effect occurs as they essentially create an artificial range dilation on angular velocity.

And to throw a PHD in my face like it means something is total rubbish. It's like saying George Bush went to Yale and therefore he's not an idiot. Your PHD only know's as much information as you feed him you clueless git, which is why I say to either listen to someone who's actually studied the eve mechanics and has somewhat of a grasp, or GTFO of the way so someone better can handle it.

Next time you throw a PHD in my face, you better ******* have a grasp on the content knowledge first bro.


I'm Down wrote:

Do you even know why the wormhole exploit worked for perfect hits before it was removed? The formula as currently coded has a variable in it that can sink to 0 creating an undefined slope on chance. This was recognized by the client and allowed infinite range, and perfect damage. The patch to get it out of game had to adjust the wormhole effect that created the undefined slope. This in fact had 0 to do with signature of ships. The way it was exploited was to **** with the tracking variables on ships.
...
The only way for both probabilities to have equal effect is to separate the terms. The only way to separate the terms in mathematics is through addition or subtraction. You can ask any 6th grader that question as it's in their current curriculum.


Just, wow. Your previous posts about signature not being realistically handled at range were kinda nice, but you've went off the deep end in these posts and your replies to Kadesh. First, you totally misuse the term 'tracking' in the ChanceToHit formula:
http://wiki.eveonline.com/en/wiki/Turret_damage
You use 'tracking' to describe the entire transversal/(range-to-target*tracking) term instead of specifically limiting your rant to tracking speed. 'Tracking' doesn't grow with range. All other things being equal, when you increase range between you and the target, ONLY range-to-target increases, which means that the exponent for ChanceToHit decreases, which means you have a greater chance to hit. Yes, if you just teleported a ship away from you its transversal could be the same (but _angular_ velocity would decrease).

Why is that error important? Because your sperging about the wormhole exploit can only be described as wishful and nonsensical. If 'tracking' sunk to near zero, ChanceToHit would be near zero as well. This is true even if this near-zero value for tracking was negative, because the term is squared! A very large exponent in the ChanceToHit formula gives virtually zero ChanceToHit, full stop. If the value for tracking was exactly zero you'd get a NaN, exception, or garbage but fortunately this is very rare in floating point math... and the exploit, of course, did NOT produce garbage. It gave seemingly infinite tracking. This is because the value for tracking was almost certainly represented server-side as an unsigned variable, and when > -100% TD was applied the value underflowed and became a VERY LARGE positive value. Uber-tracking. It had nothing to do with something trying to 'sink to 0' or an 'undefined slope'. Seriously, 'undefined slope' to give infinite tracking? heh.

There's a wealth of literature on binary representation, unsigned vars, and over/underflows so here's a few links:
http://pages.cs.wisc.edu/~smoler/x86text/lect.notes/represent.html
http://en.wikipedia.org/wiki/Integer_overflow
http://en.wikipedia.org/wiki/Computer_number_format

p.s. I also have a Ph.D. and am pretty sure Greyscale's guy is better at math than you, wassssup
p.p.s. once you get past 6th grade math (way to put 6th graders on the spot there bro) you discover there are ways to separate terms in mathematics other than addition and subtraction. Operators, anyone?
http://www.wolframalpha.com/input/?i=derivative+of+x+times+derivative+of+x
http://www.wolframalpha.com/input/?i=derivative+of+%28x+squared%29

bored at a coffee shop on Saturday~~

https://twitter.com/pmchem/ || http://community.eveonline.com/news/dev-blogs/community-spotlight-garpa/ || Goonswarm Economic Warfare Cabal
I'm Down
Macabre Votum
Northern Coalition.
#1074 - 2012-04-21 21:09:13 UTC
pmchem wrote:
[quote=I'm Down] all that stuff
Kadesh Priestess
Descendance.
GoonSwarm.
#1075 - 2012-04-21 21:31:27 UTC  |  Edited by: Kadesh Priestess
pmchem wrote:
This is because the value for tracking was almost certainly represented server-side as an unsigned variable, and when > -100% TD was applied the value underflowed and became a VERY LARGE positive value.
It had nothing to do with any value of tracking imo, and your thoughts seem to be wrong - because aharm didn't get just infinite tracking, they also got infinite range (like medium acs hitting at 200 km). Plus it's unlikely to be caused by under/overflows, as afaik ccp uses floats for attribute values (python floats are c double-precision floats).

Most likely it was caused by some error raised in formula code + CtH code itself being wrapped into try-except block which caught all possible errors, provinding 'safe' fallback (if anything bad occurs) of CtH = 1.

Without knowing exact CtH calculation code, it's really hard to say what kind of error it was and in which part of formula it could occur.
Tehg Rhind
Atlantic Innovations
#1076 - 2012-04-21 22:09:00 UTC  |  Edited by: Tehg Rhind
ok I don't get teh argument against Grayscale on the signature resolution vs turret tracking thing. The formula is pretty straightforward. If the formula is actually correct, he's right, since the tracking term is in radians and every term is independent. This isn't "PhD" level math. This is basic ******* algebra. Doubling the top of a fraction is the same as halving the bottom.

Quote:
tracking grows with range and therefore is not a static variable


This is wrong and is the entire source of your problem. In the formula tracking is in radians, and is a static variable. Maybe you should...I dunno....spend 30 seconds actually looking at the formula before you take a dump on yourself in public next time.
pmchem
Brutor Tribe
Minmatar Republic
#1077 - 2012-04-21 22:30:54 UTC
Kadesh Priestess wrote:
Plus it's unlikely to be caused by under/overflows, as afaik ccp uses floats for attribute values (python floats are c double-precision floats).


True, the result would depend on the way attributes are stored server-side. The behavior looks a hell of a lot like an underflow, although it could also be the result of a try & except with a poorly handled exception. As you say, it's impossible to tell for sure without the actual code. If anyone's gonna complain that python doesn't have unsigned variables which can overflow, I will simply point them at the ctypes module, which I know CCP uses in other parts of their codebase.

https://twitter.com/pmchem/ || http://community.eveonline.com/news/dev-blogs/community-spotlight-garpa/ || Goonswarm Economic Warfare Cabal
I'm Down
Macabre Votum
Northern Coalition.
#1078 - 2012-04-22 00:08:01 UTC
Tehg Rhind wrote:
ok I don't get teh argument against Grayscale on the signature resolution vs turret tracking thing. The formula is pretty straightforward. If the formula is actually correct, he's right, since the tracking term is in radians and every term is independent. This isn't "PhD" level math. This is basic ******* algebra. Doubling the top of a fraction is the same as halving the bottom.

Quote:
tracking grows with range and therefore is not a static variable


This is wrong and is the entire source of your problem. In the formula tracking is in radians, and is a static variable. Maybe you should...I dunno....spend 30 seconds actually looking at the formula before you take a dump on yourself in public next time.


Radians are used so that tracking is not static. If tracking were static such as I can track 500 m/s always, then your arguement would be correct.

But when you use radians, It means that a ships velocity has to increase for the tracking to have the same effect. For a ship that has 0.02 tracking, this means at 10km of range, he can track 200 m/s of speed. But at 20km range, he can track 400 m/s of speed. Therefore, it's not a static number.

It's been tested and proven time and again on both the live and test server.

As for the denominator being 0, there's a huge difference in a calculation between a denominator being close to 0, and a denominator being 0. When a denominator is 0, the computer can't physically compute the number as there is no limit on infinity.

And yes, it was 100% that made the difference:

http://massively.joystiq.com/2010/09/29/eve-wormhole-exploit-is-fixed/
Andy Landen
Deep Core Mining Inc.
Caldari State
#1079 - 2012-04-22 02:59:54 UTC  |  Edited by: Andy Landen
Because of the Titan/super balance issues, as a capital pilot, the only way that I would fly a non-super capital ship in k-space null or low sec is either with the threat of titan/MOM hot drop at zero, or with alliance paying for my non-super capital AND a very good chance of winning. Otherwise, the only real capital ship choice is a super, hence the current capital fleet comps. Can CCP point to one change scheduled for the next update that would allow non-super choices to be viable as well? Or are you content with everyone being forced to fly a super or no cap at all, unless they feel like burning ISK? The presence of an anti-cap role, as the Titan has been dubbed, has turned into the massive "I win" button of destroying most interest in caps and even k-space null and low sec. Do a survey of the number of players who would say "I am not at all interested in caps." I personally know hundreds of people who have said that very statement. In Titan territory, expect to lose everything unless you are a super yourself. Is that how CCP wants Eve to develop or do we want real changes to the Titan?

"We cannot solve our problems with the same thinking we used when we created them." Albert Einstein 
Tehg Rhind
Atlantic Innovations
#1080 - 2012-04-22 03:43:42 UTC
I'm Down wrote:
Tehg Rhind wrote:
ok I don't get teh argument against Grayscale on the signature resolution vs turret tracking thing. The formula is pretty straightforward. If the formula is actually correct, he's right, since the tracking term is in radians and every term is independent. This isn't "PhD" level math. This is basic ******* algebra. Doubling the top of a fraction is the same as halving the bottom.

Quote:
tracking grows with range and therefore is not a static variable


This is wrong and is the entire source of your problem. In the formula tracking is in radians, and is a static variable. Maybe you should...I dunno....spend 30 seconds actually looking at the formula before you take a dump on yourself in public next time.


Radians are used so that tracking is not static. If tracking were static such as I can track 500 m/s always, then your arguement would be correct.

But when you use radians, It means that a ships velocity has to increase for the tracking to have the same effect. For a ship that has 0.02 tracking, this means at 10km of range, he can track 200 m/s of speed. But at 20km range, he can track 400 m/s of speed. Therefore, it's not a static number.

It's been tested and proven time and again on both the live and test server.

As for the denominator being 0, there's a huge difference in a calculation between a denominator being close to 0, and a denominator being 0. When a denominator is 0, the computer can't physically compute the number as there is no limit on infinity.

And yes, it was 100% that made the difference:

http://massively.joystiq.com/2010/09/29/eve-wormhole-exploit-is-fixed/


Ok. Tracking, as represented in the equation, is in terms of radians. It is independant of range. Hence why range is a seperate variable.

All your evidence to the contrary is meaningless, as you are trying to refute deductive reasoning with inductive reasoning. The failure of inductive reasoning, and your argument, is that there are other possible explanations for the results. The discovery of quantum mechanics was not used to refute calculus, it wad used to refute classical dynamics.

If you cannot accept that there is a fundamental equation behind this, I don't know what to tell you. Honestly I don't know how someone as thick headed as yourself was able to lead the NC to victory. Oh. Wait.