Simulations of One-On-One Battles: Original Starfire Book 1

Home of the 1st and 2nd editions of STARFIRE

Moderators: SDS Members, SDS Owner

Forum rules
Forum rules
1. Nothing obscene.
2. No advertising or spamming.
3. No personal information. Mostly aimed at the posting of OTHER people's information.
4. No flame wars. We encourage debate, but it becomes a flame when insults fly and tempers flare.

Try to stick with the forum's topic. Threads that belong to another forum will be moved to that forum.

Re: Simulations of One-On-One Battles: Original Starfire Book 1

Postby Graywulffe on Thu 16 Jun 2016 16:43

Whitecold wrote:Just a question out of curiosity, how does W work? I take these are the box launchers, but they have been cut from the Solar version of the game.


W, or gun-missile launchers, can be used as either a projectile-firing gun that is useful at close range, or they can launch missiles that are more useful at far ranges (but not both at the same time). W systems take up as many hull spaces (3) as straight guns (G) or missile launchers (R), but cost more (20/5/10 MC respectively). They are considered more advanced.

I gather that guns in the newer versions of the game actually launch some kind of small missile, not a solid projectile?

-best
Graywulffe
Lieutenant JG
Lieutenant JG
 
Posts: 23
Joined: Sat 28 May 2016 08:35
Location: Vancouver, BC

Re: Simulations of One-On-One Battles: Original Starfire Book 1

Postby Cralis on Fri 17 Jun 2016 09:42

Graywulffe wrote:W, or gun-missile launchers, can be used as either a projectile-firing gun that is useful at close range, or they can launch missiles that are more useful at far ranges (but not both at the same time).


AAAAAAND it's important to remind everyone that Graywulffe is using First Edition Starfire. W as a gun/missile launcher disappeared in 4th and later editions of Starfire. The code W re-appeared in Ultra Starfire as the Box Launcher. It was removed from Solar Starfire because of balance issues that become a bleeding hemmorage at later ELs.

W systems take up as many hull spaces (3) as straight guns (G) or missile launchers (R), but cost more (20/5/10 MC respectively). They are considered more advanced.

I gather that guns in the newer versions of the game actually launch some kind of small missile, not a solid projectile.


G or "Sprint Missile Launchers" have been high-c, low guidance missiles since Third Edition.
Image
User avatar
Cralis
SDS Member
SDS Member
 
Posts: 10202
Joined: Tue 30 Jun 2009 19:27
Location: Oregon, USA

Re: Simulations of One-On-One Battles: Original Starfire Book 1

Postby aramis on Fri 17 Jun 2016 14:48

Graywulffe wrote:
Whitecold wrote:Just a question out of curiosity, how does W work? I take these are the box launchers, but they have been cut from the Solar version of the game.


W, or gun-missile launchers, can be used as either a projectile-firing gun that is useful at close range, or they can launch missiles that are more useful at far ranges (but not both at the same time). W systems take up as many hull spaces (3) as straight guns (G) or missile launchers (R), but cost more (20/5/10 MC respectively). They are considered more advanced.

I gather that guns in the newer versions of the game actually launch some kind of small missile, not a solid projectile?

-best

Per 2nd and 3rd, they are one launcher, using two different kinds of missile.
3rd both made Guns "slang" for "sprint mode missiles" and the ammo for W was able to be fired in either mode.

The initial idea, according to SV Cole, was inspired by missile fire from naval guns in the real world navies - certain naval guns had missiles which could be fired through them - less damage but longer range and more accuracy than shells (due to onboard guidance), and not requiring a separate box-launcher. Functionally, it's just a matter of a missile of a caliber suited to fit the gun, with a trigger the gun's initiator can hit, and short enough to go on the ramp. Open the breech, put it on the ramp, program it, ram it into the tube, close the breech, fire.
aramis
Rear Admiral
Rear Admiral
 
Posts: 261
Joined: Mon 01 Mar 2010 00:42
Location: Eagle River, Alaska

Re: Simulations of One-On-One Battles: Original Starfire Book 1

Postby Graywulffe on Sat 18 Jun 2016 09:58

Thanks for the clarifications and added info on guns and W in later versions of Starfire Cralis and Aramis.

Part IV: Superdreadnought on the Rampage

1. Interlude

Since I have been testing superdreadnaughts (SD), I decided to have a little fun and pit these heavy-duty ships against smaller hull types. To keep things quick, I kept the simulations to 100 games. All of the ships employed in this exercise were the same gun-missile launcher types that I have been using in recent experiments:

FG1 – (2) – SSSAAHIIWWDII (4)
DD1 – (2) – SSSSAAAHIIWWDWDII (4)
CL1 – (3) – SSSSSAAAAHIIWWWDWWDII (4)
CA1 – (3) – SSSSAAAAHXWHIWWIWWDWWWWDII (4)
BC1 – (3) – SSSSSSAAAAAAHXHXHWIHWIDWWWDWWWDWWWWDII (4)
BB1 – (4) – SSSSSSSAAAAAAAHXHXHDHIWHWWWIWWWDWWWDWWWDII (4)
SD1 – (4) – SSSSSSSSSSAAAAAAAAAAHXHXHWHWHWHWDWHWIHWHWWDIWWWWDWWWWDWWWWDII (4)

Key statistics for the outcomes are shown in what I loosely call a table below (I made multiple attempts to use the BBCode table tags, and they did not work.).

Column headers: Hull Type, % Wins, % Damage to SD, % of attacks that resulted in any damage to the SD, average number of turns per game.

    FG......... 0.0%, 0.2%, 14%, 2.5
    DD......... 0.0%, 0.8%, 50%, 3.0
    CL.......... 0.0%, 1.3%, 81%, 3.2
    CA.......... 0.0%, 4.4%, 100%, 3.8
    BC.......... 0.0%, 18.2%, 100%, 4.9
    2nd BC.... 0.0%, 27.7%, 100%, 5.3
    3rd BC.... 88.5%, 94.6%, 100%, 6.6
    2BC........ 61.0%, 85.3%, 100%, 11.5
    BB.......... 0.0%, 23.1%, 100%, 5.4

If you are the captain of a frigate (FG) and detect an incoming SD on the long-range scanners, the best course of action is to RUN. Out of 100 encounters, the SD destroyed the FG, and this without so much as getting a scratch 86% of the time. The typical game only lasted two or three turns! Destroyers (DD) and light cruisers (CL) only fared a bit better. It takes a CA or larger to at least slightly damage the SD during all encounters.

Battlecruisers (BC) hit harder than the CAs and managed to last about five turns before being reduced to scattered debris. A second BC encountering the damaged SD still did win in 100 games. It took a third such encounter to finally finish off the SD, but even with this, the reeling SD won 11.5% of the time. Two BCs working in tandem did better than single attacks in sequence, winning 61% of the time.

The battleship (BB) outcome, with no wins, was not surprising given that a BB is at least two hull types down from a SD. There is room for a dreadnaught (DN) with 100 hull spaces between a BB and SD. It is one of those intriguing mysteries of Original Starfire that either: 1) a DN was not used instead of a SD; or 2) a DN was not simply included among the available hull types in section 13.8. Returning to the BB, it managed to destroy some of the SD internal systems on occasion, about 0.7% in the average game (~one system module). Single BCs, by the way, managed 0.1% internal damage.

2. Thoughts Extreme Events

This exercise brought to mind something that is missing from the Original Starfire rules: Extreme event distributions. In basic terms, this is where you have a few very large events interspersed with many smaller ones. An example from history is the sinking of the HMS Hood on 24 May 1941, where it is believed that one of the Bismarck's shells penetrated a rear turret magazine, causing a catastrophic explosion. In Starfire there are many small events (e.g. gun, missile and force beam hits), but no true extreme events. In game parlance, there are no critical hits.

Given that the amount of energy required to move large starships is very high, it seems reasonable that a single hit could trigger a catastrophic explosion, even after accounting for the emplacement of safety systems. Plus some of these vessels would have gun/missile magazines. The fuel may also be explosive, depending on engine type.

To bring in an extreme event distribution, here is one idea: During the 2d6 roll used to determine a hit, one could make "snake-eyes", or a score of two a critical hit. Maybe limit this possibility to all non-shield strikes. A critical hit would then require a second roll. If a two is rolled again, the entire ship is destroyed in a Hood-style blast. This is approximately a one in 1,000 event. It might happen once in a large battle. If a three is rolled, the ship takes 2d6 damage per each damage point of the weapon being used (e.g. a force beam with a damage of three would destroy 3 * 2d6 system modules), with four resulting 1d6 damage per original damage point and five 1d3. Anything above a five would cause normal weapon damage. A critical hit method adds a slight twist to the game, one that might make a would-be ship builder reconsider concentrating everything into one large vessel.

-best
Graywulffe
Lieutenant JG
Lieutenant JG
 
Posts: 23
Joined: Sat 28 May 2016 08:35
Location: Vancouver, BC

Re: Simulations of One-On-One Battles: Original Starfire Book 1

Postby Graywulffe on Fri 24 Jun 2016 13:03

Part V: Fine Tuning System Allocations

1. Superdreadnoughts

The most interesting outcome of this exercise is that the percentages suggested by Cralis appear to be among the better distributions for allocating systems, though I will add that I did not test every possibility. So I suspect that the original numbers quoted by Cralis were arrived at through some amount of play testing.

To try to pinpoint the "golden allocation", I used superdreadnoughts (SD) armed with gun-missile launchers (W) for this set of tests. With 104 hull spaces after four engines are installed, this means that each hull space represented approximately 1% of the ship, allowing for more precision compared to using the 42 spaces of a cruiser. The allocation I started with was 20% shields and armor, 10% point defense, 60% weapons (all W) and 10% incidentals (H and X), or 20/10/60/10. In all cases I kept the incidentals at 10%, so I will be just reporting the first three allocation percentages.

Each test involved 100,000 game simulations. Most of these took 90-120 minutes, a bit too long for my liking. Thus, I did not test that many configurations before moving onto other things.

I ran the 20/10/60 SD against one with 25/15/50, and 25/15/50 came out ahead with 57.1% of the wins. I pitted the 20/10/60 against 15/5/70 and the former proved markedly more successful with 69.3% wins. A SD with a 30/20/40 configuration was bested by the both the 20/10/60 and 25/15/50 SDs, with 72.6% and 77.1% of the wins respectively.

By this point, the 25/15/50 configuration did the best in all of its trials. I then tried something between 20/10/60 and 25/15/50: 23/12/55. Against 20/10/60, the new configuration won 52.22% of the games, high enough to be well outside of random chance. A control run of the 20/10/60 SD pitted against itself returned 50.01%/49.99%, for example. Against 25/15/50, the 23/15/55 configuration succeeded 52.1% of the time. Note that in both tests the percentage of wins is very close, suggesting that 23/12/55 is very near the ideal, or akin to the golden allocation.

Among the many caveats to the above conclusion is that I only tested a limited number of configurations. Perhaps there is more than one ideal distribution—islands of stability in the chaos. Also, I only tested SDs armed with gun-missile launchers. Different weapons, and allocations with mixed weapon types, may alter outcomes. I plan to test this in a little while, though I might use simulations of 10,000 games instead of 100,000 to speed things up.

Also, I tested battleships (BB) using the 20/10/60 and 23/12/55 configurations. With these ships, 20/10/60 came out slightly ahead with 50.5% of the wins. I tried two other tests with BBs. I pitted the 20/10/60 ship against 25/15/50 and 20/10/60 ended up winning 55.5% of the time. Interesting. Against a BB with a 15/5/70 allocation, 20/10/60 won 76.7% of the games. It appears that BBs have a slightly different golden allocation compared to SDs, one right around 20/10/60. Perhaps BBs, with a "meager" 68 hull spaces after four engines are installed, are too small for very fine slicing as with the SD.

Of course, one big area of uncertainty is the fact that system placement is critical, as shown in Part I. Though I try to make the system configurations similar between ships, it is difficult to know for sure if the location of a specific system is not conveying at least a slight advantage to one ship over the other due to multiple factors explored in earlier posts (e.g. protecting weapons behind lots of systems generally has a positive benefit—and one allocation may provide more such protection in the form of lots of point defense systems). This leads to…

2. A Brief Return to Basic Manipulation of System Modules

When I first commented on the fact that simply switching the locations of two systems modules can have a dramatic effect on outcomes, I forgot to mention that I suspected the magnitude of the effect would be lower in larger ships. Testing of the SD bears this out, though it is interesting that the effect is still detectible in a ship with so many system modules (usually around 60 in my designs).

Using a SD of this configuration:

SSSSSSSSSSSSSAAAAAAAAAAAAAHXHXHDXWHDHIWHWDHWHWIDWWWDWWWDWWWDWWWDII

I simply switched the first D with the first W to make a second slightly different ship.

And what happened in a face off between these two ships? The SD with the systems unchanged, in essence with the W behind the D, or protected by the D and one X, had a slightly higher number of successes: 50.66% of the total (662 more wins out of 100,000), enough of a difference to be real. This is in contrast to moving systems around in smaller ships where the effect on outcomes was huge. The small effect shown here suggests that the exact system placement on the SDs and BBs is critical when trying to ferret out a golden allocation.

3. Primary Beam

I finally have a (mostly) working prototype of the primary beam that can soon be included in the simulator. Given that the primary beam largely breaks the basic "start destroying systems on the left and then progress to the right" model of the game, this took a fair amount of creativity. Implementing the primary beam in the simulator will also require some reprogramming of the other weapons as they will have to "see" any destroyed systems that may be randomly scattered through the ship—something that was not a concern before programming primary beams since all the other weapons progress from left to right. Even lasers after they skip the shields. I hope to have primaries fully working in a few more days. Energy beams will be next—and they have their own challenges.
Graywulffe
Lieutenant JG
Lieutenant JG
 
Posts: 23
Joined: Sat 28 May 2016 08:35
Location: Vancouver, BC

Re: Simulations of One-On-One Battles: Original Starfire Book 1

Postby southwestforests on Fri 24 Jun 2016 13:20

Man, you are really putting in some work here!
Fascinating results.
Screw the rivets, I build models for atmosphere, not detail
User avatar
southwestforests
Commander
Commander
 
Posts: 113
Joined: Mon 27 Aug 2012 21:01
Location: Boonville, Missouri - a little antebellum burg on the Missouri River

Re: Simulations of One-On-One Battles: Original Starfire Book 1

Postby Graywulffe on Sun 26 Jun 2016 09:38

Thanks, southwest forests. This project became much bigger than I imagined when I first started. I just keep plugging away when I find some time. I have many other tests in mind, so please stay tuned!

-best,
Graywulffe
Lieutenant JG
Lieutenant JG
 
Posts: 23
Joined: Sat 28 May 2016 08:35
Location: Vancouver, BC

Re: Simulations of One-On-One Battles: Original Starfire Book 1

Postby Cralis on Sun 26 Jun 2016 12:18

Fascinating.

You're right that a great deal of playtesting went into the recommendations we often give. However, a great number of people led by Marvin did playtesting and analysis before me and the defense ratios were part of that prior playtesting. So I cannot take credit for that work...
Image
User avatar
Cralis
SDS Member
SDS Member
 
Posts: 10202
Joined: Tue 30 Jun 2009 19:27
Location: Oregon, USA

Re: Simulations of One-On-One Battles: Original Starfire Book 1

Postby Graywulffe on Fri 15 Jul 2016 12:01

Part VI: Primary Beams

As of today, I have primary beams fully implemented in my starship face-off simulator. This required reworking all of the weapons so that they would "see" randomly distributed destroyed systems. In the process, I also refined all of the code and cut the entire program length in half even as I added the primary beams.

I changed the weapon-type fire order on each ship to this:

1) all P; 2) all F; 3) all L; 4) all G; 5) all R; and 6) all W.

The main reason for this arrangement is the rationale that a ship with beams and missile launchers would want to try to destroy the opponent's point defense first (P is good for this), before firing the missiles.

Aside from a few control tests, the only test I have run with the primary beams involves and age-old question that has been nagging Original Starfire players for decades ;) : If your ship is only going to be armed with just one weapon type, what is the better choice: primary beams or force beams?

I used these two destroyers for the experiment:

DD1 – Stinger – (2) – SSSAAAIIPIPIPDI (5)
DD2 – Ram – (2) – SSSAAAIIFIFIFDI (5)

The outcome of 100,000 games: The Ram, fully loaded with force beams, had a 13:1 advantage, winning 92.7% of the time. The primary beam ship, Stinger, did a little better than I expected. Likely, skipping the opposing ship's first six systems (shields and armor) was a benefit. They key event that needed to happen for Stinger to win was the early elimination of Ram's force beams. Note that the first two F's are within reach of the initial primary beam strikes.

Another thing to keep in mind is that with a starting speed of 5, the Ram generally does not get within 0 range of the Stinger on the first pass (the ships start at 12 hexes from 0), where force beams would do extreme damage. My next simulation will probably use similar ships but with a 4 movement as this would bring the ships to zero range on the first pass. However, I note that a captain solely using P beams would probably do everything possible to avoid a zero-range encounter with a ship loaded with F. In a real 2D game, this would create an interesting tension, with the Ram's captain seeking to close to zero distance at the right opportunity and the Stinger's captain constantly seeking a distance of a few hexes to maximize the hit probability while attempting to avoid a close encounter.

Edit 16 Jul 2016:

I tested the above ships after modifying them to have a movement of four. To do this, I changed the first "I" into an "H" (...AHIF...). The outcome was a bit of a surprise: The primary-beam heavy ship did better when moving into closer range, now getting 13.0% of the wins. It appears that the higher chances of a successful hit with the P beams improved the Stinger's chances of surviving the battle.

-best,
Graywulffe
Lieutenant JG
Lieutenant JG
 
Posts: 23
Joined: Sat 28 May 2016 08:35
Location: Vancouver, BC

Re: Simulations of One-On-One Battles: Original Starfire Book 1

Postby thebard on Sun 17 Jul 2016 04:25

One question to consider is that the 2 captains will try to control the range.
So - does range control matter?
One way to simulate this could be varying the starting distance, so that in different simulations , firing will happen at different distances.

Have you tried this?
thebard
Commander
Commander
 
Posts: 139
Joined: Mon 08 Sep 2014 08:47

PreviousNext

Return to Original Starfire

Who is online

Users browsing this forum: No registered users and 2 guests