Isn't 'fast && dangerous' unbalancing?

[Admiral]

Now for some Science!(tm)

I absolutely loved this.

Of course, if I were a developer at Arcen, I'd have to take your analysis to the next level, and build in a simulation engine where you could specify:

1) a game to load
2) a series of commands to issue immediately after load
3) a completion condition of some sort
4) statistics to output at the completion

And then you could tell it to run the simulation N times (say, 1,000) for each set of commands to see what really the ramifications for each one were.

That's the crux of the problem. X never invited me to be a part of the dev team.

*slinks off to the corner and pouts*

(I actually built something somewhat like this for Space Empires IV and V when I built the original Play by Web engine, but that was a TBS system and specifying commands was part of the whole play-by-email engine so it was pretty easy to do.)

[darke]

So having used Science! to prove beyond a shadow of a doubt that Real Men(tm) (and Real Women(tm)) use F&D to assault wormholes, we now try our second hypothesis: Is Normal harder when assaulting a group of guys surrounding a command post?

Stats are same as previous, as it's the same world. Ship distribution is the same, there's almost the same number of ships (266 this time, so a whole 5 less!) and this pack conveniently has an Ion Cannon III mixed in for extra hurty!

The ships are back! Once again almost 2000, of the same distribution, all nicely healed and replenished. We move the ships into the world, let them stabilise on the wormhole, save the game, and let our trials commence!

This episode's trial: Ships move from wormhole in group movement, to a point just slightly on the opposite edge of the pack of enemy ships surrounding a command post. This should ensure that the ships maintain their formation, rather then spread out slightly to attack some ships, and also either wipe the ships out, or get wiped out themselves.

F&D: the mobile ships moved as a group out to meet me, before I crushed them and obliterated their entire force save the ion cannon, with about 1850 ships left. The ships took about 5 or 6 losses to the ion cannon on the way there.
Normal: Things worked the same as with F&D, except I had a grand total of 1800 ships left. And the Ion Cannon took out about 13 to 15 on the way.

Let's test that one more time to see if we get the same pattern:

F&D: This time they didn't run out as quickly to greet me, so I hit both their towers, including at least 14 lighting turrets, along with the rest of their force. Ended up with only 1700 ships remaining.
Normal: AI moved their forces out, like the first set of tests. Ended with 1870 ships remaining.

Conclusion: Doing the most absolutely brain-damagedly stupid attack with both F&D and Normal (running into a mass of lighting turrets!? Dumb.), and the difference in ship losses is not only within a 5% margin of error, but appears to be completely on the whim of the PRNG. So this one is pretty much a tie.

Additional Conclusion: Assaulting a wormhole in normal (assuming I've done the math correctly), is approximately between 350% and 500% more deadly then assaulting the same pack of ships currently on a world. Assaulting a wormhole in F&D is at least 1300% more deadly, and considering the force got wiped out, mathematically infinitely more deadly!

Stay tuned for another episode of F&D vs Normal entitled, "Is F&D Unbalancing, or is Normal just for players who like easy-mode wormhole assaults, and have a fetish for so very, very slow cruisers?" Same Bat-Time, Same Bat-Channel!

(Sorry; couldn't resist hamming it up. )

[darke]

More Science! (Now with less snark! )

The last couple of the "dumb strategies" tests. Also decided to round all numbers to the nearest 50 since they're really random anyway.

This test: same as previous except the ships are set to lone mode:

F&D: 1200 left.
Normal: 1650 left.

Second run:

F&D: 1400 left.
Normal: 1600 left.

Next test: Ships are set to lone mode as before, except the target is next to the command post in the center, so rather then the fast ships zooming through the middle and piling up on the edge, they'll pile up in the middle so that can keep hitting anything in range.

F&D: 1050 left.
Normal: 1650 left.

Second run:

F&D: 1350 left.
Normal: 1600 left.

Conclusion Between these, and the previous tests, if you're going to play the combat side of the game with, shall we say, less tactical skill (since you may prefer to concentrate on the economics parts); it would be better to set the game to Normal and run it at Speed +1. You'll get all the benefits of the double-speed movement and the like in terms how how fast it feels to play, without the serious amount of ship attrition.

In short: If you haven't quite worked out how to use tactics, or don't want to use tactics, or you just want an easier combat side of the game, or just have a slow-cruiser fetish: play on EasyNormal mode.

I'll keep on going with the testing on this world with some smarter tactics and see what numbers we get different. Off the top of my head I'm inclined to test:

* Stopping before you get into range of the lightning turrets and let your carriers take them out. (Probably both group mode and long mode.)
* Gliding past on a tangent, so only your carriers, and maybe half your bombers will get a hit on the turrets. Since any closer and I'll get guys hit with the outer edge of lightning turrets. (Group only, since you don't have good defenses against the ships coming towards you otherwise, due to the way the AI attacks.)

Any suggestions for repeatable "smart" or "dumb" attacks to test will be appreciated; though probably ignored if they take waaaaaay too long to setup and execute. Doesn't mean you shouldn't suggest them though.

[Revenantus]

Your data do support the theory that F&D mode is more difficult than Normal mode.

The increased difficulty of assaulting the AI through wormholes can be attributed to Fiskbit's observation that damage is inflicted 4x as quickly, whilst movement is only 2x as fast, hence on F&D mode ships effectively approach the enemy at half the rate. This leads to more damage being done to your ships before reinforcements arrive through the wormhole. I'd be interested to see what would happen if you were to 'swirl' the fleet to a point before sending them ships through the wormhole on each mode.

If your intra-system combat tests are accurate, observe that if there are situations in which your fleet is disadvantaged compared to Normal mode, if you were to swap fleets with the AI then the AI would be similarly disadvantaged.

I am concerned by the huge variability in some of your tests, and while you have enacted all practical means to try and ensure that the only difference between the two games is the combat style, it is difficult to be certain if this is the case.

Also note that since you are accustomed to playing on F&D mode, your tactics may reflect this and players that are accustomed to Normal mode may employ different strategies, affecting the outcome. This point is actually very significant as each mode may require entirely different mixes of ships for optimum combat effectiveness, invalidating your tests. Some mixes may be disadvantaged on one mode, but highly effective on the other. Because ships effectively move towards targets at half the rate, physical attack ships such as cutlasses and vampires are prime candidates for tests such as these.

Sniper turrets are significantly more powerful on F&D mode (again, due to the effective movement rate change). How many Sniper turrets are present in your test system out of interest?

Also consider that if Normal mode is indeed easier, since it is (probably) not possible to defeat the AI on certain combinations of settings, a player could play at their maximum skill level on either mode.

As Admiral said, what we really need to test this is a simulation editor from which we can generate a huge data set to analyze.

[PhilRoi]

* PhilRoi is standing back cause someone is trying science....

[Revenantus]

If anyone is interested, Lanchester's Square Law can't be applied to Lightning Turrets because they damage all enemy ships simultaneously. Now, even just off the top of my head there are a large number of assumptions are required for this derivation to be valid, but, being a mathematician I can define a place where these problems don't exist, so I'm going to go ahead with this anyway.

Let's consider the situation in which a group of ships (A) with conventional (single-target) weapons are attacking some lightning turrets (B). This battle can be described by the equations;

dA/dt = -k_1AB
dB/dt = -k_2A

Without actually typing up the intermediate algebra, this solves to;

A - (k_1/2k_2)B^2 = (A_0)^2 - (k_1/2k_2)(B_0)^2

The interpretation of this is that conventional ships' fighting strength scales linearly with their number, whilst the lightning turrets' strength is proportional to the square of their number.

As stated above, too many assumptions have been made for this result to be practically useful, but I find it interesting all the same.

If I get bored I'll probably try and solve the equations for a battle with ships of both types on each side, they look a bit ugly however.

[Admiral]

The interpretation of this is that conventional ships' fighting strength scales linearly with their number, whilst the lightning turrets' strength is proportional to the square of their number.

I love how Rev applies DEs to gaming. And here I thought they were just for physics (and, apparently economics, biology, etc.)!

This back-of-the-napkin would be easy to validate (within a first order of magnitude presumably) when we have the simulation editor where:

1) we could place AI and player forces at will in a save game
2) we could pause the save game and issue orders which will be played out as soon as it is unpaused
3) we could define those "tactical zones"
4) we could give an ending condition based upon # of ships in one or more tactical zones

Now the simulation engine would just need a save game and it would execute it until the end condition (or a certain amount of game time occurs, c.f. halting problem in computer science which would be akin to Rev's incompleteness theorem in a large number of respects) is reached, and output the number of ships in each defined tactical zone. It would repeat the simulation the specified number of times, and also give a summary of the statistics on the tactical zone results (min, max, average, median, various percentiles, along with standard deviation would seem reasonable).

The good thing about this simulation engine is that #1 probably would not be hard to implement, #2 and #3 already exist (or will), and #4 should probably also be easy to implement. Then, the simulation engine would be quite straightforward, although it would not be terribly fast unless X implemented a "clone the game state in memory" command because he'd have to reload the state each time and, as anyone who has ever loaded a save game knows, that can take a few seconds.

Furthermore, you could specify that N simulation engines should be run simultaneously, so that you could take advantage of more available CPU, although that would most likely require spawning entirely different detached processes given X's mention earlier of various global variables that are used in the game engine (or converting those into a "global state object" that is passed around everywhere or associated with each thread in a TLS mechanism).

Sometimes I think I like designing and talking about the game more than playing it. LOL

Cheers!

[darke]

As far as tactics goes, there is none. I've used thick-as-a-brick "non-tactics" that any newbie would try, to try and maximize the reliability of things. I used maxed II/III/IV ships, because that was a nice stock late game mix and it was easy to explain. The large variability is due to the targeting, for example if most of the lightning turrets get taken out early on, there's going to be less killy, if they're not, there's going to be more.

I expect there to be considerably smaller variance between the test runs when playing with more intelligent tactics. I expect there to be considerably less damage overall using smarter tactics (which is why I qualified the "In Short"). However none of these change the fact that using brute force, dumb tactics, the effect on the ships matches what I said at the beginning, the hypothesis I was trying to prove or disprove.

"Smarter" tactics such as tossing lightning turrets through wormholes reduce the damage significantly, but they come at a cost, and they're available on normal mode as well. Knowing these tactics also comes with experience, which is why I noted that Normal mode is much more newbie friendly; which is logical given that Fast & Dangerous is not named Easy & Safe.

Conveniently enough, I figured I'd take a stab at playing what I suggested. Normal mode on +1 speed (and +1 frame skip to try and keep down the choppyness). And just for the extra challenge, the AI pair I have an absolute pain in the ass time with: one Special Forces Captain, and one Teleport Turtle. Playing with my usual F&D tactics of course, since I know nothing else.

Random thought log from when I was playing:

Securing the start world was insanely easy. Less then 10 minutes, and I'm done. Compared to 20+ I'm used to having to do against these AIs with F&D. This is possibly just luck though.

It seemed to take a much longer time for the first wave to hit me, happened after 23 game minutes, which felt like it took an age to arrive. Not sure if F&D makes them hit earlier, or if I'm usually just frantically trying to protect myself for the first while (which didn't happen this game), that the time passes quickly.

Taking down a Special Forces Command Post was snooze inducing. Given the massive amount of them in this game I think I'm going to have to rush for Tech II cruisers much quicker then usual.

Taking out a force field so I could take out a gate on an adjacent planet was surprisingly painless. I lost less then 10% of my force, usually I'm down at least a third.

First time I've ever seriously gone over the energy into negative. Took me, like, 15 minutes to work out why my repairers weren't doing much. Probably due to the fact I don't seem to be loosing as many ships, and the game is taking about 20% longer so far to clear out adjacent gates and take over my first world.

Enemy needs to have less of a fascination for my teleporting engineers. I can bamf them all over the map to keep the enemies out of the hair of my ships who are relaxing and repairing.

Speaking of which, my ships are repairing much faster then they're being damaged. I didn't even click that engineers are essentially a quarter as good in F&D mode since you're taking theorietically 4x the damage...

Taking out a command center then wiping out the ships as they fly towards you is way too easy in Normal. On the plus side I did have great fun dancing four sets of 50 ships around a floral shape in the middle of the map as the enemy ships were chasing after me being randomly wiped out, and my repair droids were happily running around in circles with me healing up any minor scratches and bumps. It just feels really cheap though; I mean, I've lost 40 ships out of 300, on taking out half of a world that had over 500 Tech I ships on it.

Like really, really despserate for energy. Three of each generator on my two worlds so far. Should have hit new worlds, then cleared out gates, not other way around. The ships move sooooo slooooooow in normal mode, even though I'm going at warp speed, obviously the clock is ticking over quicker.

Ahh, things are going much quicker now I've got the pace of things. Once I take out half the forces in the sector, I should be sending half my attack force over to the wormhole that leads to the next, whilst I use the rest of it to clear out the current sector. Once it's there and healed, I should be using it to pop in, and clear out the ships on the other side of the wormhole now I don't have to worry about loosing real amounts of ships, whilst I build up my force in this sector. Seems to take something less then 20 game minutes per world with this method.

Tech II Engineers are insane on auto heal when attacking a new world. Currently on my 6th world and I'm loosing at most 5% of my force per world conquering; insane!

Anyway, time for bed, I'll get around to playing further tomorrow after work. Hopefully it gets a little harder.

[x4000]

The time between waves seems the same, the frequency of specops raids seems the same (though I'm not 100% sure, can anyone verify?), my ships however move at 2x speed. Makes it so much easier to bring some reserves to a hot-spot. Or to wield a mobile reaction force.

Also ships spend most of their time moving. The frequency of Ai reinforcements is likely constant. Means I reach the juicy system with an adv. fac. and a not so juicy ion at 4:30h instead 6:00h. Means I prolly face less resistance.

Just wondering

Other people have already responded to this, but I think opinions have varied (based on a quick scan).  This is a great question, though, so I added a section about it in the wiki (scroll down to the second item on this page:  http://arcengames.com/mediawiki/index.php?title=AI_War_-_Combat_Styles)

[x4000]

I'll start by saying that I prefer to play with the Combat Style set to Normal.

I'm the opposite, but I added both modes because I know opinions vary.

The time between waves seems the same, the frequency of specops raids seems the same (though I'm not 100% sure, can anyone verify?), my ships however move at 2x speed. Makes it so much easier to bring some reserves to a hot-spot. Or to wield a mobile reaction force.

The specops raids are actually not on a timer at all, but are just dependant on when the AI happens to have enough ships gathered at the wormhole to bother sending them through.  Since the AI ships move faster, same as yours, this means that often they will reach the needed saturation point quicker on F&D mode.  But it's not exact.

It's true that AI raids occur at the same rate. This is my primary reason for preferring Normal mode because I feel Fast & Dangerous detracts from the value of effective ship positioning, an important aspect for me.

Yeah, that was why I originally had the entire game based on Normal.  I used to play solely like that.  But it just got too time-consuming after about 5 months of that, so I added F&D mode to placate my alpha group, which was getting a bit annoyed at how long it was taking to traverse planets.

Also ships spend most of their time moving. The frequency of Ai reinforcements is likely constant. Means I reach the juicy system with an adv. fac. and a not so juicy ion at 4:30h instead 6:00h. Means I prolly face less resistance.

Since the AI does not reinforce systems until that system is aware of the player's presence, the time required to reach a system isn't hugely relevant in that regard, although, if there is a significant time delay between the player making the AI aware of their presence and launching an attack on the system, then yes it could potentially, depending on AI reinforcement choices, be an important factor.

The AI reinforcements are made cyclically, though, like the AI offensive waves.  So that means that, given the same amount of time, you'll have fewer reinforcements made on F&D than on Normal.  These really aren't meant to be equivalent, but are two very different styles of play.

If you do play with the AI progress timer on however you will have to scale it appropriately between the two Combat Styles to maintain an equivalent level of difficulty.

Also true.  That's an easy way to offset the difference, if you are inclined.  I think that both modes are a similar level of difficulty, just in different ways.  The AI, when it is on the offensive in F&D mode, can really rip you a new one before you are ready.  Your ships don't build any faster in that mode, so you have a lot less time to react compared to Normal.

EDIT: I apologize, the AI does in fact reinforce at all of its planets, just preferentially at locations that you have made it aware of your presence. I think the AI Progress level and number of command posts dictates the maximum number of ships the AI can station at a given planet. I actually can't recall extremely long games ever resulting in significantly increased garrison sizes on AI worlds, but I could be wrong here.

Please see this:  http://arcengames.com/mediawiki/index.php?title=AI_War_-_AI_Reinforcements

Generally the AI is highly unlikely to reinforce a planet that is not on alert.  That would only happen if you had a lot of planets captured, and but were somehow only causing a few of their planets to be on alert (say, if you were on one of the really long corridors-style maps).  The AI progress level and the number of command posts do indeed dictate how many ships get added in each reinforcement.  I've added notes to the above link (bottom paragraph of the first section) on how that works, but you are basically correct on your notes there.

Is there a maximum number of ships the AI will station at a planet? If so, what are the factors affecting this number? Also, from their initial starting garrison sizes, how long does it take AI worlds to maximize their defensive forces? Is it likely that by playing extremely aggressively the player would encounter significantly fewer AI ships?

No, there is no limit.  There is no maximum.  I don't think I've ever seen an AI planet with more than 5,000 ships, though, simply because they tend to get spread around.  And once there are more than a certain number of ships in the game, the AI will start scrapping their lower-level ships in exchange for higher-level ships (a slight loss to the AI, but it keeps things moving along).  If you play really aggressively, yes you would encounter fewer ships for a while, but you wouldn't be likely to be able to sustain that.  By the same token, once you capture a lot of their planets with your early aggression, that will make them able to do even more reinforcements, so it would even out at some point in a game of a reasonable size (not 10 planets).

I'm not sure I can ever recall attacking a planet that had a surprisingly small/large garrison size, perhaps I'm being too cautious.

If you act fast with the early planets, you can often catch them with 50-150 ships at them.  On the later planets, you can see 5K or potentially even more.

[x4000]

Also: Lighting Turrets == Ow, ow ow, ow ow. (Recently lost the better part of 800 ships of various tech levels to an entrance wormhole guarded by 6 of these and a small force of ships). Sniper Turrets == irritating; Sniper Ships == don't expect to keep anything the AI thinks is irritating on an assault alive. This includes engineers, though thankfully they don't think science labs or colony ships are generally interesting.

Quite so, those become a lot more effective in that mode.

As far as timing goes, I've played almost exclusively with Fast & Dangerous and between all the above my game times are within the expected length for the number of planets.

Yeah, I think my estimates for most players playing at their appropriate difficulty level are really more appropriate for Fast & Dangerous rather than Normal.  It seems like most people on Normal in those circumstances go well over the expected range.

I find it amusing that I crank up the AI progress timer to stop me from turtling (and slowing the game down), whereas you're suggesting cranking it up *to* slow the game down.

I also find that amusing.  That's pretty cool that it works in both circumstances, though.

From memory of what Mr X has said previously, he populates the worlds, then if the game is getting too many ships in it he starts converting groups of lower-level ships into higher level ships of equivalent hurtyness.

Yeppers, so that keeps the numbers somewhat capped, but it's per-galaxy (around 60k ships), not per-planet.  And you still can get even more than 60k ships in a galaxy, but probably not in a 1-player game even on the largest maps.  In an 8-player game on higher difficulties, you'll start with more than that, though.

[x4000]

Heh, I also build harvesters in order to distract the AI from more valuable targets - seems a bit off doesn't it? . I too always play with the schizophrenic modifier as it makes things far more interesting. The question of how the combat style influences defense is actually contentious in that while you may have less time to construct new ships, making bolstering your defense forces more difficult, while in F&D mode your fleets can move towards wormholes more quickly, thereby significantly increasing the area in which a fleet can defend against the timed incoming AI waves.

Great point.  I've added a feature about this here:  http://arcengames.com/forums/index.php/topic,477.0.html

Assaults tend to be more deadly to structures because they don't fight back and their health does not scale with the increased damage, however, it does not inherently affect ship-to-ship combat as in theory everything should sustain damage 2x as fast, and the overall outcome should be the same.

Yes, the overall results of ship-to-ship combat will be the same, in general.  Although if low-powered ships get off shots that do 2x damage before they die, that can affect balance somewhat because then the stronger-but-lower-health ships that you control have a higher chance of dying.  If that makes sense.

True, lightning turrets are terrifying. I often require lightning missiles to establish a beach head on well fortified AI planets.

You can also send through some poor first ship, let it get whacked, and then send through your real forces.  That really can help.

This may be entirely because of the standard rate of reinforcement is highly consistent throughout the game. Maybe I'll load a save with a planet with a huge AI garrison and leave it for a while, to see if the number increases indefinitely.

Yes, that's what would happen.  If you left it for a day (and your stuff survived), you'd come back to find it incredibly entrenched.  It takes a while, so that you aren't under constant time pressure, but you can't dawdle forever.

[x4000]

I apologize if it's difficult to follow the derivation, it's tough to made it particularly legible using only ascii characters - x4000, can we have LaTeX support on this forum?

C'mon, he's a (really good) Windows dev... What are the odds he knows what LaTeX is?

(And, no, it's not something his wife wears... Unless he's lucky?)

I'll respect the privacy of my wife on that one...   But I am familiar with LaTeX from my time in a computer science curriculum.  If there's a lot of demand for LaTeX here, I can probably get something working with SMF, but the instructions I've seen are not uber quick to do and I'm not certain how much seriousness was behind that request.  Surely we won't be seeing a real proliferation of equations here on an ongoing basis!?

[x4000]

I'm not sure if you guys noted this, but F&D mode doubles movement speed, reload speed, and damage, so in the same amount of time, a ship will do four times the amount of damage it previously did. Because movement is only doubled, this also means that moving ships defensively is more difficult because this more-or-less halves the speed of the ships as they're moving toward their target (ships die four times as fast and your defenses get their twice as fast). During beta, you were allowed to switch between normal and F&D modes anytime as many times as you liked, so when being attacked or scouting, I found that normal mode was superior, while when attacking or moving around ships, I'd use F&D mode. This didn't sit well with me and it wasn't what Chris had in mind, so that was changed. I'm pretty sure the in-game toggle is there now so that you can go to F&D mode from normal as you get better at AI War rather than having to wait until you start a new game for it.

But anyway, the imbalance in speed vs ship offense increase makes for some definite differences between the two modes. Personally, I play F&D because I can't stand the slow-as-molasses ships like cruisers in normal mode.

Yeah, you and I had a variant of this whole conversation back when the game was still beta, I recall...

As far as the simulation editor that was mentioned above, I've created a future DLC topic for that: http://arcengames.com/forums/index.php/topic,478.0.html  I doubt it will be a real high priority for most people, but I agree that it would be cool to have at some point.

[x4000]

I think it's usually prudent to start with normal for the first couple of games though.

Those were my thoughts exactly -- I think it's a good way to ease into the game, and then decide which mode you prefer.  That's also why I made an in-game toggle, so people could start slower and then make it faster if they want.

I like both modes, I'm favouring F&D at the moment.  Agree with Fiskbit cruisers being that much faster is refreshing.

My thoughts exactly there.