I keep on reading here that currently in Disharmony, just stacking melee kinetics is the most cost-effective. I haven't been able to find thuvian's proof of this. Also, I tend to find that beam weapons do the most damage. Missiles almost never hit and battles are usually decided before melee range (and before that melee kinetics have horrible accuracy).

I cant answer your question as I haven't played enough. But I usually go with Missiles and Beam because the computer usually goes with a melee kinetic configuring. I play one very easy and newbie. This allows me to take them out before melee round even gets there as you say. However, later in the game if you face some opp with a lot of flak, you're going to get roasted. I've only played two games through though so...

I use Kintetics heavy + a bit of beam @ MR... That seems to do the trick.

Yes, melee kinetics are still the best cost ratio, and missiles are still worthless

I've always thought that Thuvian is a bit daft.



Melee range Kinetics are better than the other weapons by something around 400%. However, if your opponent is building ships with 10 armor, 0 shields, and 0 flakk, then pure kinetic ships won't do very well. That's why I suggest building 3 types of ships in your fleet: the all melee range kinetic destroyer, the all medium range laser destroyer, and the all long range missile destroyer. Producing them at a rate of 3 kinetic to 2 laser to 1 missile is about right, although your mileage may vary. You can build ships that combine all three types of weapons, but they are more awkward to allocate and distribute, plus you'd need to fine tune the amounts based upon your opponents armor. I find it easier to ignore all that and just change my fleet composition to match the opponents weaknesses.



By mixing up your offense weapons, you keep your opponent honest. If he stays armor heavy ships, then your lasers will take him out. If he diversifies, your kinetics + lasers will take him out. You have missiles, but most likely they won't contribute that much, except when your opponent has lots of all of the defenses, in which case his offensive power is really diminished.



On the other hand, the next update is coming in a few days (edited: I lost track of time), so we'll see the new balance patch in action.

thuvian wrote:
On the other hand, the next update is coming in a few weeks, so we'll see the new balance patch in action.

You mean in 2 days? Or is there another announced patch that changes combat? The Nov 14th patch, from what I've seen, doesn't touch military aspects at all.

jktstance wrote:
You mean in 2 days? Or is there another announced patch that changes combat? The Nov 14th patch, from what I've seen, doesn't touch military aspects at all.

As far as I know, they are changing the ammount of % damage each weapon does in each certain battle phase in this next new dlc on Nov 14. I couldn't find the link but i know i read it on the devblog.

thuvian wrote:
I've always thought that Thuvian is a bit daft...

Thuvian is cool. Especially when he says he does stuff that all his spreadsheet calculations prove he shouldn't do.



A Couple experiments with AIs I built over a decade ago made it clear to me. If a spreadsheet can figure out the optimal strategy... then you have a simulation, not a game. But that puts a major strain on AI's, because the AIs have to learn what works and what doesn't.



And then that has other challenges because who should AI's learn from? Other AI's offers iteration, but in single player games, wouldn't it be best if the AI learned how to beat... ME? (for example... what battle card choices do I prefer? What choices do I prefer when I am stronger... weaker)



I gave up programming ai's until I spun up a neural net... which I haven't done yet.

Oops. No. There is only the 14th patch that I'm aware of. I just forgot what day it is.



Hmm. I'm not sure I agree on your distinction between game and simulation. A spreadsheet is just a visual depiction of an algorithm. If your claim is that optimal strategies can't be calculated, then I'm pretty sure that's wrong. In contrast if you are claiming that that it's complicated, that has more validity, but I still think you can do it. My favourite story is some tournament involving spaceships back during the 80s(?). A computer program wrote up a program to develop an optimal fleet composition for a table top space war game tournament. His program revealed that the ideal fleet was a horde of tiny ships with nothing but weapons (i.e., no engines). He won devastating against all of the other competitors. I find the story hilarious, but I couldn't find a link with accurate details about the event. If someone else remembers, that would be great.



Anyway, the model for the selection of the ideal ship design do support the above framework of fleet design. Melee kinetics are better than other weapons, but that's assuming the defenses are equally distributed. When your target ships have armor, the balance changes. If the target ships Effective Health for kinetic is >400% of the other Effective Healths (i.e., Missile & Laser), then kinetic is best. Otherwise you shift to the appropriate weapon. By using a fleet composition with varying weapon ratios, you can force your opponent to take heavy losses regardless of their own composition and fleet design. In multiplayer it works a bit differently because your opponents are going to be doing similar sorts of things to you, then you'll both have ships that follow the 4/2/1 defense pattern (or something similar depending on which exact values are in the current patch). In that case, the weapons are interchangeable, and then you get into the game states where you try to find cheaper ships that are equally combat effective. Regardless though, melee kinetics are STILL ~400% more damage efficient than the competitors.

The event that really shoved this point into my face was an early game loss due to pirates. I had just started a game, built a few scouts and then ran into some pirate ship events. I couldn't support a very large fleet and I couldn't even build ships very fast. Eventually in shifting my ships around 1 of my ships got caught by the pirates who were blockading my empire with 2 ship fleets. I did roughly 0 damage to his ships. He killed my ship in the long range phase using melee kinetics. That was because I didn't have very many guns on my ship, he had lots of them on his, and evasion, accuracy, and evasiondisorientation. Rebuilding ships wasn't sufficient. To kill the pirates, I had to create a group of ships that contained sufficient guns to breach the evasiondisorientation limit, because he already had and I couldn't gun down anything individually. So I did the natural thing, I rage quit and did something else.



If you want to see this sort of outcome yourself, it is easy to reproduce. Load a game, create a fleet of all melee kinetic destroyers, go looking for another player's fleet to attack. When you are 1 turn away from them, save your game. Then attack him with increasing numbers of ships. Start with 1, then 2, then 3, etc. Notice how the damage you inflict doesn't linearly increase (depending on the number of weapons each ship has. With fewer weapons this is more noticeable), but instead jumps. So while 1 ship might do 1 damage, 2 ships do 3 damage, and 3 ships do 8 damage (numbers not to scale, once it saturates evasiondisorientation it does become linear).

First the bit that we disagree on but I know I am right about... errrrr... pretty sure I can get you to agree with me on.

thuvian wrote:


Hmm. I'm not sure I agree on your distinction between game and simulation. A spreadsheet is just a visual depiction of an algorithm. If your claim is that optimal strategies can't be calculated, then I'm pretty sure that's wrong...

A common misconception is that building tough AI's is hard. It is easy... I have stumbled upon two unbeatable AI's in two different games I have built. But that meant the game design was flawed, not that the AI was great. (Side note... do not design said AIs for wife who likes to win... that was one 'upgrade that did not go over well) But lets leave my little guns vs butter two dimensional games I toyed with and talk about an obvious answer.



Would you like to play a game?

This time, its time to tell the Thuvian to play TicTacToe.



The ultimate in spreadsheet games is not 'winnable'... i.e ... fun in a competition sense.

thuvian wrote:


.... In multiplayer it works a bit differently because your opponents are going to be doing similar sorts of things to you, then you'll both have ships that follow the 4/2/1 defense pattern (or something similar depending on which exact values are in the current patch). In that case, the weapons are interchangeable, and then you get into the game states where you try to find cheaper ships that are equally combat effective. Regardless though, melee kinetics are STILL ~400% more damage efficient than the competitors.

... aaaaandddd, then you illustrate my points. The system in place has a 'most efficient' configuration... but not an optimal one because the efficient configuration is counterable especially once you factor in the cards. This is actually very difficult to accomplish. And very difficult to teach AI's to play with as they must learn their opponents tendencies in order to counter them. Interestingly... in these cases, a very strong AI is a random one. As the good ship designs and battle plans survive while the bad ones die off, and the AI only needs to learn from itself what is working and what is not. Of course then you have the user community complaining that the AI is brain dead and games aren't worth playing to the end because they are beating the AI in the learning from its mistakes portion.



That's why humans are the neural nets that play balance games... not spreadsheets.

ShuShu62 wrote:
That's why humans are the neural nets that play balance games... not spreadsheets.

Spreadsheets are awesome. Especially the really huge ones with all sorts of interdependent cells. Yay functions!

Full disclosure... my day job is in Business Intelligence... spreadsheets are my friends. And I do love Thuvian's analyses. The spreadsheet guys are the starting point for most insights, so I hope I am not coming off as antagonistic... I am shooting more for... playful.

Memes... Memes everywhere...

I wish that Swordfish scene resonated with me. But i would consider my job more in line with convincing a bunch of folks that Glass Cannon only fleets aren't the optimal solution, I don't care how many posts they found on the forums saying it is.