Starships in the sequel

[keith.lamothe]

Thing is there is large odds that 3 & 4 won't ever happen at higher difficulty level - and as I stated, there is a niche for suicidal strike force. That's about it.

If every single planet has sufficient local AI forces to match your entire fleet, either you've let them build up too much or there's something wrong with the numbers. That's not what I've seen at all, or thought I was hearing reported by the folks who enjoyed high difficulty.

Even then, you could probably achieve a #3 or #4 situation by a few short-lived fleet battles. That is, pulling back before you sustained heavy losses, but causing disproportionately more by your relative concentration of force against the things near the wormhole.

The AI has overwhelming galaxy-wide forces, sure, and its important planets are well-covered, sure, but most of the game is arranging strategic moments of local superiority for your forces. If you can never do so, you're up a creek without a warp nacelle.

What is necessary to know to start a game ? What's needed in the first 5 minutes of AI War ? A lot of easy to medium complex things to know.

Make a list and see.
- how to scout.
- how to use the 3 basic ships.
- how to use the 5 starship types and their 3 variants each => requires your to understand the armor types.
- how to manage energy ?
- how to manage fuel ?
- how to set-up defenses ?
- how to unlock technologies.
- how to set-up your factories properly.
- how to evaluate your attack forces to attack properly.
- (...).

That's a fair point.

One thing I was considering earlier was having the frigates be non-modular (just one "kind" of each line), and have the destroyers have only two weapon choices, with the cruiser and bb having all three. That might work differently with the Shield and Stealth lines, but I don't think the player would get the Stealth Frigate for free anyway. One problem in Classic is that you started with mkIs in so many different starship lines. It was nice to motivate experimentation, but ultimately it was counter to the new-player experience.

[kasnavada]

If every single planet has sufficient local AI forces to match your entire fleet, either you've let them build up too much or there's something wrong with the numbers. That's not what I've seen at all, or thought I was hearing reported by the folks who enjoyed high difficulty.

I remember seeing that from difficulty 8 onward actually, having planets (especially 3-4 planets) that could match your entire fleet. Thankfully, there was quite a few actions that enabled the player to beat that. I also remember lots of threatballs that the player had absolutely no way to take on offensively with its own fleet. One of them was to wake the planet up, which "freed" the enemy ships from defense, made them threat. And there was a few ways to get rid of even overwhelming threat. From what you say - and what I've seen from the design doc, this won't work the same anymore as the ships will go back to defend the worlds rather than being threat ? I'm not sure about that point.

The other thing that was mostly preventing this was the command station that prevented alertness on the AI worlds. Which I believe is gone ?

Thing is the high numbers of AI units in AI War I worked because the defenses were "more" static.

So... I'm basically worried about losing that "overwhelming forces" feeling, yeah - because the more responsive the AI is, the less it needs forces to actually be stronger than the player.

That said, as you stated, it's balancing numbers. It's going to be hard to theorycraft.

15 variants at start (...) There's enough to understand already in the game.

I think my point isn't understood, actually. I do get yours - it's not hard to know by itself. But my point ain't that. My point is that even very easy stuff to understand points make a complex mess, if there is a lot of them. What is necessary to know to start a game ? What's needed in the first 5 minutes of AI War ? A lot of easy to medium complex things to know.

Are you changing the point of disagreement now? I think Keith's reply was actually really good.

I'm not changing my point of view. I'm even stating that in itself, the starship is easy to get. Still I stated my point twice, even if the first time was less easy to understand. Also, while I'm pretty sure that the players will mostly choose one variant over others (it's very difficult, at best, to make 3 meaningful choices when doing variants), it can be balanced eventually.

One thing I was considering earlier was having the frigates be non-modular (just one "kind" of each line), and have the destroyers have only two weapon choices, with the cruiser and bb having all three. That might work differently with the Shield and Stealth lines, but I don't think the player would get the Stealth Frigate for free anyway. One problem in Classic is that you started with mkIs in so many different starship lines. It was nice to motivate experimentation, but ultimately it was counter to the new-player experience.

That could be nice - and it would have the benefit of enabling different human factions for example - on faction starts with different types of weapons (eventually reaching all 3 later). Possibly even more so if you could choose which of the weapons as the "level-up" procedure.

[keith.lamothe]

I remember seeing that from difficulty 8 onward actually, having planets (especially 3-4 planets) that could match your entire fleet.

Some planets, sure. Especially Mk4. Those are likely going to take multiple attacks unless you have a lategame fleet or major superweapons. But, iirc, if you tried to kite around a beefy MkIV planet you were going to get shredded by something because you'd wind up with too much distance to the wormhole. It was more a hit-and-run thing.

I also remember lots of threatballs that the player had absolutely no way to take on offensively with its own fleet.

Yea, if you don't stop threat from accumulating, and you have heavy defenses, the threatballs will grow large (because smaller ones couldn't break your defenses). But generally these won't be along your whole border unless you've only got one or two entry points. If you do have only one entry point, and especially if you've spent lots of K on turretry instead of mobile stuff, then the AI does do a bit of "returning the favor" of your having stonewall defenses.

One of them was to wake the planet up, which "freed" the enemy ships from defense, made them threat. And there was a few ways to get rid of even overwhelming threat. From what you say - and what I've seen from the design doc, this won't work the same anymore as the ships will go back to defend the worlds rather than being threat ? I'm not sure about that point.

Freed ships will stay free, as before. If anything it will be easier to free larger chunks of a planets defenders, since they come after you and once you shoot at them they're freed (as in Classic).

One thing to bear in mind is that in a #1 situation it doesn't send all the guardians after you. Just those within a certain guard radius of your ships. It's only if you come in with something that poses a threat to the planet as a whole that it concentrates its local forces. Partly this is to make the AI resistant to easy feints (while you send the main force in another way), and partly it's to make the AI overconfident enough that you can usually achieve some favorable kill-to-loss rate even on a doom planet.

The other thing that was mostly preventing this was the command station that prevented alertness on the AI worlds. Which I believe is gone ?

The station itself, yes. But most/all of the old command station functions will still be available through structures you can build on the planet. The Jammer will probably cost a fair bit of Power, which means fewer turrets or whatever on that specific planet (due to Power being per-planet), but it's a good option to include.

Thing is the high numbers of AI units in AI War I worked because the defenses were "more" static.

So... I'm basically worried about losing that "overwhelming forces" feeling, yeah - because the more responsive the AI is, the less it needs forces to actually be stronger than the player.

That said, as you stated, it's balancing numbers. It's going to be hard to theorycraft.

It will be a difficulty shift, yes, due to the qualitative boost to the AI's use of its forces. And the AI's overall forces will still be quantitatively overwhelming. But I have total confidence in the sneakiness of players, to still achieve victory against ridiculous odds It may mean a lot of people using lower difficulty settings than they did in Classic, at least for a while, but I don't think that's a bad thing.

[Atepa]

I am strongly in favor of using the hull types for capital ships - it distinguishes them further from fighters and makes them feel more "capital ship-ish" to me.  Plus I really want to be able to dramatically declare "Battlecruiser operational" over Skype and annoy my co-players. 

Okay, my vote was already for the hull types as opposed to marks, but this added 'perk' seals the deal.

I guess my point is, roles seem to only become important when stuff begins to break down or become ineffective. Unless the player is actively trying a gambit, those distinctions are likely to be overshadowed by 'increase the fleet's lifetime damage output'. So the question seems more, how to make the player need to try gambits, and then figure out how the differences between ships can give the player fodder to scheme with.

I think a big part of that is playing at the right skill level though too. If you are just using a fleet blob on the majority of the systems when you roll through them, then you're probably playing on too easy a setting.

That's one of the things that has always kind of boggled my mind is making a game that is hard enough to provide me a good challenge without just completely dominating my current skill level. Having to use tactics and such for more than a handful of systems is what has always helped me gauge that.

[chemical_art]

I myself have never been a fan of modular designs. It made it difficult to balance as it would seem one would be OP or worthless.

Perhaps, rather then trying to dice up 3 types modules for each ship, we make it so that each ship starts with one "set" module that defines the ship, or provides a bonus that compliments its role already. So a long range siege ship would get a long range weapon effective against a different class. Or a brawler that has a high RoF has a secondary module that provides a strong single shot weapon.

Then, after after unlocking the MK II ship there can be a "side tech" that would allow a second module that performs alongside and would more radically alter the ship class. It would be on all ships of that class, but perhaps a greater number of them as ship mark goes up These techs allow the ship to perform a greater number of roles but forces the player to "dig in" research on that ship branch, so for it to be worthwhile the player needs to really invest in the ship class more fully. As a further result, it would not be effective to perform too many of these side roles since they do not increase caps.

In this way each ship can be balanced without trying to tip toe with each module is stronger / weaker then the others. Since the order of the modules are set there is not that internal tension.

[WolfWhiteFire]

Personally my favorite ships are the modular ones in classic, not because they are op but because they allow customization. So I wouldn't really be a fan of replacing the modules. Also in any game some things will be better or worse than others in general, but some of the less used things could be extremely useful depending on the scenario.

[RabidSanity]

My vote is for hull types for the starships.  Just sounds better to me. 

Additionally if guardians are going to be the main defense of planets I am guessing they get reinforcements directly like the guard posts in AIWC.  If they are turned into threat fleet do they still get reinforced during the reinforcement cycle? 

[keith.lamothe]

Perhaps, rather then trying to dice up 3 types modules for each ship, we make it so that each ship starts with one "set" module that defines the ship, or provides a bonus that compliments its role already. So a long range siege ship would get a long range weapon effective against a different class. Or a brawler that has a high RoF has a secondary module that provides a strong single shot weapon.

Then, after after unlocking the MK II ship there can be a "side tech" that would allow a second module that performs alongside and would more radically alter the ship class. It would be on all ships of that class, but perhaps a greater number of them as ship mark goes up These techs allow the ship to perform a greater number of roles but forces the player to "dig in" research on that ship branch, so for it to be worthwhile the player needs to really invest in the ship class more fully. As a further result, it would not be effective to perform too many of these side roles since they do not increase caps.

In this way each ship can be balanced without trying to tip toe with each module is stronger / weaker then the others. Since the order of the modules are set there is not that internal tension.

Yea, something like that could work. Here's a similar idea I've been batting around:
(taking the Siege as an example, with no modular stuff)
- You start with the MkI (whatever we call it), and it only has the short-range plasma cannon.
- After unlocking MkII you can also unlock "Torpedo Generator"
- When that's unlocked, you can toggle any Siege (including the frigate) into or out of "alternate fire mode"; that causes the Siege to:
-- have to reload before firing
-- no longer be mobile
-- not fire its normal weapon
-- instead fire the long-range torpedos

Which might be too much micro, but it sounds interesting.

One UI thing that could help is a toggle for "giving a move order automatically cancels any mobility-removing mode", but if I'm already thinking about adding new toggles it's probably too complex

Additionally if guardians are going to be the main defense of planets I am guessing they get reinforcements directly like the guard posts in AIWC.  If they are turned into threat fleet do they still get reinforced during the reinforcement cycle? 

If a guardian is on guard duty, the reinforcement logic can put stuff inside them, yes. If the guardian leaves guard duty (for threat duty) it retains its subordinates but the reinforcement logic can no longer give it new ones.

[z99-_]

Honestly, I really liked modular ships and was happy with what you originally proposed. And we're going to have a dedicated tutorial that would cover basic stuff like modular ships, right? All that remains is balance, which would take a little time, but certainly isn't difficult. I don't see the problem with them.

[keith.lamothe]

Honestly, I really liked modular ships and was happy with what you originally proposed. And we're going to have a dedicated tutorial that would cover basic stuff like modular ships, right? All that remains is balance, which would take a little time, but certainly isn't difficult. I don't see the problem with them.

I'm still more likely to follow the original proposal (with the modification of having the modularity start at the Destroyer level instead of the Frigate level), but I wanted to throw the toggle-mode idea out there. Modularity is great for several-purpose ships like the Shield or Stealth, but for a combat line maybe it's really just a dual-purpose ship, and modularity is overkill for them. On the other hand, maybe toggles are worse UX than modules.

[Draco18s]

I think toggles are OK, as long as there's a visual distinction. Think the trebuchet from Age of Empires. No one has ever said "man, these are confusing!"

[kasnavada]

Honestly, I really liked modular ships and was happy with what you originally proposed. And we're going to have a dedicated tutorial that would cover basic stuff like modular ships, right? All that remains is balance, which would take a little time, but certainly isn't difficult. I don't see the problem with them.

Ohhh... I so need to sink those ships right now.

Whatever the tutorial is, the game start needs to be intuitive enough so the about half of people (INTERNET STATISTICS ARE ALWAYS RIGHT) that don't start games with tutorial don't leave disgusted.

More seriously, look at the achievements stats for AI War classic. 22% of not even playing an hour ? I'll stress that again, this should not happen in the sequel.


Second point, balance is very difficult and takes a lot of time and tweaking. The more possibilities, the more difficult there is. Given the number of units that are already in the game, and the AI types, and everything, AI War II is already stated to have about the same number of units that a MOBA has heroes. (/sacarsm) And as we all know, MOBAs are all completely balanced, there is no heroes that are basically worthless (/sarcasm). Thing is the more stuff there is in a game, the more combo (squared, or cubed possibilities) there is, and the higher chance of some OP or worthless "combo" being there.

Given the limited ressources that Arcen has to make this game, I'm pretty sure that we can soften some edges, and patch the glaring holes abused by the community - but that's about it. Remember that a company like Blizzard fails at this.

[MaxAstro]

Although that 22% number could be somewhat misleading, since it counts (for example) people who received the game as a gift but never actually tried it.  Consider that Borderlands 2 has a 27% rate of people who got the game but never played far enough to get their first gun (~5 minutes of gameplay); in that light, AI War doesn't look so bad. 

[keith.lamothe]

Whatever the tutorial is, the game start needs to be intuitive enough so the about half of people (INTERNET STATISTICS ARE ALWAYS RIGHT) that don't start games with tutorial don't leave disgusted.

The % of people ignoring the tutorial is undoubtedly higher than 50%. The tutorials in classic are only there for people who are invested in learning the game.

The alternative is somehow making it so the tutorial is always part of their first play experience. That's possible in some games but I dunno how it would work here without being obscenely obnoxious, especially when trying to cover advanced topics.

More seriously, look at the achievements stats for AI War classic. 22% of not even playing an hour ? I'll stress that again, this should not happen in the sequel.

Although that 22% number could be somewhat misleading, since it counts (for example) people who received the game as a gift but never actually tried it.

I'm pretty sure that doesn't count people who own it but have never installed. It probably doesn't even count people who've installed but not run it for the first time. If it did include all owners I'd expect the played-an-hour percent to be single digits. Tons of people own games on steam and don't even realize it.

Anyway, if the "ran the game, but bounced out before the first hour" rate is only 22% I'm surprised it's that low. They must've just wanted to listen to Pablo's title track several times through. I'd be less surprised if the rate of "stayed longer than an hour" was 22%. When someone bounces that early it's often a genre mismatch or other thing that's idiopsychological, so I doubt we'll be able to hit better on that metric.


Where I think we can improve drastically is "stayed longer than an hour, and still made it to hour 10", and so on.

[Tridus]

I'm pretty sure that doesn't count people who own it but have never installed. It probably doesn't even count people who've installed but not run it for the first time. If it did include all owners I'd expect the played-an-hour percent to be single digits. Tons of people own games on steam and don't even realize it.

Anyway, if the "ran the game, but bounced out before the first hour" rate is only 22% I'm surprised it's that low. They must've just wanted to listen to Pablo's title track several times through. I'd be less surprised if the rate of "stayed longer than an hour" was 22%. When someone bounces that early it's often a genre mismatch or other thing that's idiopsychological, so I doubt we'll be able to hit better on that metric.

I'm pretty sure it does, actually. Steamspy has imperfect, but more useful info: https://steamdb.info/app/40400/graphs/

The key ones are the average and median play times. The average is over 8 hours. The median is 1.3. Steamspy's data shows 13% played for around 5 minutes. That makes it pretty clear that a ton of people barely played, and the dedicated folks played massive amounts. There's also ~100k folks who own the game and have never played it (since March 2009, at least), which is more common than you might think for games on Steam.

Where I think we can improve drastically is "stayed longer than an hour, and still made it to hour 10", and so on.

Yep.