Author Topic: One does not simply win 10/10 AIs  (Read 9604 times)

Offline Kahuna

  • Core Member
  • *****
  • Posts: 2,222
  • Kahuna Matata!
One does not simply win 10/10 AIs
« on: July 08, 2012, 09:43:33 am »
So I started a game vs 2 10Doom AIs a while ago and got overrun by 2 consecutive waves. The first one had ~1k ships and the second ~2k. I had ~850 ships, turrets/defenses and 2 forts I think. The first wave destroyed the defenses and some of my ships and I couldn't repair everything fast enough because I didn't have enough resources. So I had no defenses when the second wave came (~1min after the first one). The ships that were inside the carriers were all MarkIII ships. The other AI ships were MarkII. AIP was ~100 and I had 4 planets which is 22000 knowledge.

What should I research vs 10/10 and in which order? Are there any "must" technologies? Should I just get all MarkIII ships asap? What about Starships, fortresses, force fields, OCCs (Warp Jammer and others), Zenith SpaceTime Manipulators and MarkII and III turrets?? MarkIII Scouts to find golems and advanced research stations?

I'm not yet like hardcore trying to defeat 10/10. Not quite sure yet how I should do it. Testing some stuff first.
I tried this research order and it seemed to work quite well. I only played till I had 5 planets though. At least it worked in the early game.

Technology - cost - - - - - - - - - total knowledge needed - - planets needed
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Metal Harverster MarkII - 2k K - - - - 2000  Knowledge
Crystal Harvester MarkII - 2k K - - - -4000  Knowledge
Riot Control Starship MarkII - 4k K - -8000  Knowledge   Starting knowledge
Riot Control Starship MarkIII - 5k K- 13000 Knowledge  Home planet researched
Bomber MarkII - 2,5k K - - - - - - - - - 15500 Knowledge   2 planets researched
Sniper MarkII - 2,5k K - - - - - - - - - - 18000 Knowledge   3 planets researched
Flasgship - 2k K - - - - - - - - - - - - - - 20000 Knowledge   4 planets researched
OCC - Warp Jammer - 5k K- - - - - - - 25000 Knowledge   5 planets researched

I was able to stop +2k ship waves without losing much. See the screenshots. I tried this "build" again and it worked very well. It - or something like it - would probably work in the end game too.

How do you guys play vs 10/10, do you use the Warp Jammer Command Centers and what's the max AIP you can handle?
set /A diff=10
if %diff%==max (
   set /A me=:)
) else (
   set /A me=SadPanda
)
echo Check out my AI War strategy guide and find your inner Super Cat!
echo 2592 hours of AI War and counting!
echo Kahuna matata!

Offline Mánagarmr

  • Core Member Mark V
  • *****
  • Posts: 4,272
  • if (isInRange(target)) { kill(target); }
Re: One does not simply win 10/10 AIs
« Reply #1 on: July 08, 2012, 10:00:11 am »
This is a topic you should take up with Wanderer or GUDare. They're basically the testing-hamsters when it comes to 10/10 and Keith is using them like some sort of "crash dummies" when balancing 10/10. :P The point is, if you win 10/10, that is a BUG.
« Last Edit: July 08, 2012, 10:08:01 am by Moonshine Fox »
Click here to get started with Mantis for Suggestions and Bug Reports.

Thank you for contributing to making the game better!

Offline Varone

  • Jr. Member Mark III
  • **
  • Posts: 96
Re: One does not simply win 10/10 AIs
« Reply #2 on: July 08, 2012, 10:11:06 am »
i play 9/9 at the moment and looking at your screenshot i wouldn't have AI progress so high at that time in the game. Identify those data centers and take them out, try and keep AI progress in the single digits or low twenties for as long as you can.

Also having Hard spirecraft and hard golems is probably going to be suicidal on 10/10, i would only play with one or the other at 9/9 and that is pushing it.

Offline Minotaar

  • Sr. Member
  • ****
  • Posts: 272
Re: One does not simply win 10/10 AIs
« Reply #3 on: July 08, 2012, 10:23:29 am »
This is a topic you should take up with Wanderer or GUDare. They're basically the testing-hamsters when it comes to 10/10 and Keith is using them like some sort of "crash dummies" when balancing 10/10. :P

Yeah, and they're also the same person. Weird, huh?  :D

Offline Kahuna

  • Core Member
  • *****
  • Posts: 2,222
  • Kahuna Matata!
Re: One does not simply win 10/10 AIs
« Reply #4 on: July 08, 2012, 10:24:30 am »
This is a topic you should take up with Wanderer or GUDare. They're basically the testing-hamsters when it comes to 10/10 and Keith is using them like some sort of "crash dummies" when balancing 10/10. :P The point is, if you win 10/10, that is a BUG.

Well lets hope they find their way to this topic.
set /A diff=10
if %diff%==max (
   set /A me=:)
) else (
   set /A me=SadPanda
)
echo Check out my AI War strategy guide and find your inner Super Cat!
echo 2592 hours of AI War and counting!
echo Kahuna matata!

Offline Kahuna

  • Core Member
  • *****
  • Posts: 2,222
  • Kahuna Matata!
Re: One does not simply win 10/10 AIs
« Reply #5 on: July 08, 2012, 10:26:43 am »
i play 9/9 at the moment and looking at your screenshot i wouldn't have AI progress so high at that time in the game. Identify those data centers and take them out, try and keep AI progress in the single digits or low twenties for as long as you can.

Also having Hard spirecraft and hard golems is probably going to be suicidal on 10/10, i would only play with one or the other at 9/9 and that is pushing it.

Yep. Haven't found anything to reduce the AIP yet. Also I've already won 9/9 AI's and it was "semi easy" untill ~200 AIP. So I think I can do 10/10 but it will be hard. Oh.. and I forgot to enable Beacheads for this game =( oh well.. next time :P
set /A diff=10
if %diff%==max (
   set /A me=:)
) else (
   set /A me=SadPanda
)
echo Check out my AI War strategy guide and find your inner Super Cat!
echo 2592 hours of AI War and counting!
echo Kahuna matata!

Offline Mánagarmr

  • Core Member Mark V
  • *****
  • Posts: 4,272
  • if (isInRange(target)) { kill(target); }
Re: One does not simply win 10/10 AIs
« Reply #6 on: July 08, 2012, 10:27:24 am »
Are you playing with the bleeding edge betas? Because the higher difficulties have gotten a few significant buffs lately.
Click here to get started with Mantis for Suggestions and Bug Reports.

Thank you for contributing to making the game better!

Offline Kahuna

  • Core Member
  • *****
  • Posts: 2,222
  • Kahuna Matata!
Re: One does not simply win 10/10 AIs
« Reply #7 on: July 08, 2012, 11:10:01 am »
Are you playing with the bleeding edge betas? Because the higher difficulties have gotten a few significant buffs lately.

Um yes I think I am. I started that game today morning.
set /A diff=10
if %diff%==max (
   set /A me=:)
) else (
   set /A me=SadPanda
)
echo Check out my AI War strategy guide and find your inner Super Cat!
echo 2592 hours of AI War and counting!
echo Kahuna matata!

Offline Mánagarmr

  • Core Member Mark V
  • *****
  • Posts: 4,272
  • if (isInRange(target)) { kill(target); }
Re: One does not simply win 10/10 AIs
« Reply #8 on: July 08, 2012, 11:11:56 am »
In that case, the AI might be a bit on the overly angry side. As I said, Keith is currently in a balancing streak on that.
Click here to get started with Mantis for Suggestions and Bug Reports.

Thank you for contributing to making the game better!

Offline Wanderer

  • Master Member Mark II
  • *****
  • Posts: 1,579
  • If you're not drunk you're doing it wrong.
Re: One does not simply win 10/10 AIs
« Reply #9 on: July 08, 2012, 12:04:32 pm »
This is a topic you should take up with Wanderer or GUDare. They're basically the testing-hamsters when it comes to 10/10 and Keith is using them like some sort of "crash dummies" when balancing 10/10. :P The point is, if you win 10/10, that is a BUG.

For the record (don't worry, OP, I'll get to you, just my brain is still assimilating caffeine... :) ) Yep, GUDare = Wanderer.  I also volunteered for the 10/10 brawls.  Finally, there IS no balance to 10/10, that's kind of the point. 9.8, MAYBE... 10?  No.  10 is meant to treat you like a nekkid roman in the colluseum against the lions.
... and then we'll have cake.

Offline Wanderer

  • Master Member Mark II
  • *****
  • Posts: 1,579
  • If you're not drunk you're doing it wrong.
Re: One does not simply win 10/10 AIs
« Reply #10 on: July 08, 2012, 12:07:24 pm »
Yep. Haven't found anything to reduce the AIP yet. Also I've already won 9/9 AI's and it was "semi easy" untill ~200 AIP. So I think I can do 10/10 but it will be hard. Oh.. and I forgot to enable Beacheads for this game =( oh well.. next time :P

9 to 9.3 is a pretty significant boost, though with some tweaks lately in the beta builds it got knocked down a few pegs as being a HUGE increment.  Wave size boosts alone increment faster through the 9.x series, amongst other things.  Also, as was mentioned, some pretty massive buffs just went in.  You REALLY want to read the patch notes for .036.  37/38 were mostly bug patches, but they all came in within a day or two of each other.

You're also going to be VERY glad you don't have beachheads on.
... and then we'll have cake.

Offline Wanderer

  • Master Member Mark II
  • *****
  • Posts: 1,579
  • If you're not drunk you're doing it wrong.
Re: One does not simply win 10/10 AIs
« Reply #11 on: July 08, 2012, 01:08:17 pm »
So I started a game vs 2 10Doom AIs a while ago and got overrun by 2 consecutive waves.
Doom... doom... oh, right, AI 10 Diff is called Doom.  What AIs did you use?  AI choice is incredibly important when riding on the edge... or over it, in this case.

Quote
The first one had ~1k ships and the second ~2k. I had ~850 ships, turrets/defenses and 2 forts I think. The first wave destroyed the defenses and some of my ships and I couldn't repair everything fast enough because I didn't have enough resources.
Yeah, I'm starting to call that the econ spiral, in deference to the tantrum spiral from DF.  It's in essence though how you end up losing most RTSs.  Economic attrition.  Also, that much spread in waves means you were facing an AI that most likely had wave boosters... or a ton of tiny ships.

Quote
So I had no defenses when the second wave came (~1min after the first one). The ships that were inside the carriers were all MarkIII ships. The other AI ships were MarkII. AIP was ~100 and I had 4 planets which is 22000 knowledge.
Yeah, waves seem to be LOVING back to backs right now.  Not sure when that fired up, but it's getting to be a pretty heavy duty hammer.  One of the recent changes that went in was an exponential increase to AIP affects at AI Diff 8+.  The more AIP you have, the more it affects you.  20 late AIP is no longer the same as 20 early AIP.  How selective have you been about your planets?  If a planet doesn't give you direct gain, you need to leave it behind.  Most of the time the worlds off my homeworld are literally ignored... or simply turned into neutered highways.

Quote
MarkIII Scouts to find golems and advanced research stations?
What level of Golem are you playing?  If it's medium with the auto-AIP increases you're pretty much screwed.  I'll discuss research below.

Quote
I'm not yet like hardcore trying to defeat 10/10. Not quite sure yet how I should do it. Testing some stuff first.
Out of curiousity, have you read my AARs on the nine through 10 run?  I admit, THEY'RE LONG.  Very long.  But I pretty much walk through the strategy at each level (along with my responses to a ton of balancing) at each gradient.

If you want to see the WIN against 10/10, it starts at this link:
http://www.arcengames.com/forums/index.php/topic,9869.msg106642.html#msg106642

If you want to read about me getting bounced off the walls like the nerd by a bully in some teen show, check out the AAR so far called Keith's Revenge.

Quote
I tried this research order and it seemed to work quite well. I only played till I had 5 planets though. At least it worked in the early game.

Technology - cost - - - - - - - - - total knowledge needed - - planets needed
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Metal Harverster MarkII - 2k K - - - - 2000  Knowledge
Crystal Harvester MarkII - 2k K - - - -4000  Knowledge
Riot Control Starship MarkII - 4k K - -8000  Knowledge   Starting knowledge
Riot Control Starship MarkIII - 5k K- 13000 Knowledge  Home planet researched
Bomber MarkII - 2,5k K - - - - - - - - - 15500 Knowledge   2 planets researched
Sniper MarkII - 2,5k K - - - - - - - - - - 18000 Knowledge   3 planets researched
Flasgship - 2k K - - - - - - - - - - - - - - 20000 Knowledge   4 planets researched
OCC - Warp Jammer - 5k K- - - - - - - 25000 Knowledge   5 planets researched
Alright, let's talk about your choices here.

Harvesters are pretty much a gimme in the opening moments.  At least IIs.  I currently usually go to IIIs but that may be changing in the near future.  Warp Sensors help tremendously in the very early game until you can get yourself situated as well.  I personally don't use snipers at this level.  The damage output from the cap vs. their fragility just doesn't make them worth it to me.

Riot IIs are pretty much a must have by the third planet, either that or EMP mines.  EMP mines recently sucked down a nerf though since I abused the hell out of them.  The problem with Riot IIs now though is they're going to die in the Alpha Strike unless you've HEAVILY forcefielded the entry port.  Those Carriers?  Yeah, they will now chew you to pieces.  They just got a serious buff.

Warp Jammers that late are nearly pointless.  You've either got control of most of your systems or you don't. You're looking at (at MOST) maybe 10 planets captured in a 10/10 game, or the waves will simply chew your feet off.  The rest of the systems you're taking will probably be abandonables.  Also, wave timings recently got themselves some emergent smarts.  The less access points you have, the more the AI will build to sledgehammer levels of time in the waves.  Reducing inbound points to a single choke is no longer the best option.  Two chokes is actually best, from what I've been able to tell.  See, more than 2 chokes you can still soak max-time waves with the same upper cap, but it's more likely that you'll get some lighter hits.  That's all well and good if you've got a response force hanging around in the backfield.  It sucks badly if you don't when a max-time hits.

With that said, Riot IIIs are fun little toys but what EXACTLY do they get you that you need?  Every inch of your research needs to be carefully chosen for a particular result.  Riot IIs get the tazer.  After IIs you get... well... nothing special enough to pull you out of a jam.  MOAR tazers doesn't help, they're capped on firing rates, 2 shots lasting 3 seconds each every 8 seconds.  This was to make sure stun locks could no longer happen, particularly in MP (Multi-player/multi-planet, pick your poison) games.

Also, Flagships.  This is... debateable.  I won't deny that upping firepower during attacks isn't helpful, and the Light Starship does an iffy job at best at it.  But for nearly the same price you can open up a cap of MK II ships.  With Flagships affecting ~200 ships (for the cap at normal) for 1.7x of their damage, you can double the damage of a single cap and get more ships (armor)... while the light starships are still giving you a 1.4x for those same 200 ships.  The research boost just isn't there and the firepower for the ship, again, is weak.  Flagship used to be a stepping stone to the old Zenith/Spire starships, real powerhouses in their own right that ALSO gave you fleet boosts.  Now they're 'meh'.

I don't have a good build for the new buffs yet.  My old one was along these lines:
<Harvester III/III, Warp Sensor, Tachyon/Warp Turret, Bomber II> the opening research
Scout IIs usually needed on second planet take, leaving me 2750.  This USUALLY sat quietly until the third planet.
<Riot IIs, HBC, Basic II> 0 research left
<Frigate II or Special II> 500 left
Variable from there.

But yeah, with no econ you're going to lose simply because game pacing is so high.

Quote
How do you guys play vs 10/10, do you use the Warp Jammer Command Centers and what's the max AIP you can handle?
Most people don't play 10/10.  It's not meant to be handled.  Anything over 9 (including the 9.x series) the developers are specifically incrementing to the point that it's just messing with you.  It's not balanced, it's not anything.  It's crushing.  That's the point.  9.6/9.8 is supposed to be 'top level' play.  10/10's supposed to wreck you.  EVERY. TIME.  Usually the only way 10 should ever be beaten is when there's a loophole somewhere that can be exploited, or with significant cheese (double Turtles, for example).

« Last Edit: July 08, 2012, 01:11:50 pm by Wanderer »
... and then we'll have cake.

Offline Mánagarmr

  • Core Member Mark V
  • *****
  • Posts: 4,272
  • if (isInRange(target)) { kill(target); }
Re: One does not simply win 10/10 AIs
« Reply #12 on: July 08, 2012, 01:30:01 pm »
For the record (don't worry, OP, I'll get to you, just my brain is still assimilating caffeine... :) ) Yep, GUDare = Wanderer. 
... that certainly explains a lot. *dummyhat*
Click here to get started with Mantis for Suggestions and Bug Reports.

Thank you for contributing to making the game better!

Offline Kahuna

  • Core Member
  • *****
  • Posts: 2,222
  • Kahuna Matata!
Re: One does not simply win 10/10 AIs
« Reply #13 on: July 09, 2012, 03:01:00 am »
What AIs did you use?  AI choice is incredibly important when riding on the edge... or over it, in this case.
Attritioner and Entrenched Homeworlder when I got overrun by the 2 waves. Shield Ninny and Grav Driller when I tested the Riot Starships. And I just started a new game vs Shield Ninny and The Tank.

How selective have you been about your planets?
Maybe not selective enough. I often take all planets adjacent to my home world and then some planets with 4-8 resource asteroids around them. Then I scout for ARSs, Golems and Advanced Factories and capture them. I always try to keep my planets connected to each other.

What level of Golem are you playing?
Always hard.

Out of curiousity, have you read my AARs on the nine through 10 run?  I admit, THEY'RE LONG.  Very long.  But I pretty much walk through the strategy at each level (along with my responses to a ton of balancing) at each gradient.
I haven't read it but I think I will. Maybe. Dunno.

Warp Jammers that late are nearly pointless.
So do you just destroy the warp gates? Wouldn't that increase the AIP too much?
« Last Edit: July 09, 2012, 03:13:21 am by Kahuna »
set /A diff=10
if %diff%==max (
   set /A me=:)
) else (
   set /A me=SadPanda
)
echo Check out my AI War strategy guide and find your inner Super Cat!
echo 2592 hours of AI War and counting!
echo Kahuna matata!

Offline Wanderer

  • Master Member Mark II
  • *****
  • Posts: 1,579
  • If you're not drunk you're doing it wrong.
Re: One does not simply win 10/10 AIs
« Reply #14 on: July 09, 2012, 03:15:35 am »
What AIs did you use?  AI choice is incredibly important when riding on the edge... or over it, in this case.
Attritioner and Entrenched Homeworlder when I got overrun by the 2 waves. Shield Ninny and Grav Driller when I tested the Riot Starships. And I just started a new game vs Shield Ninny and The Tank.
Shield Ninny doesn't drop waves. Attritioner has a wave multiplier of 1.25. Tank's a 1.0 waver but he has some heavy armor on his troops.  Grav's just annoying as all heck.  Entrenched is just a PITA because of the Human Settlements.

Just so you know, defeating a 10/10 with a non-waver won't earn you a bug report unless the other AI is something like Extreme Raider or Scorched Earth.  ;)

Quote
Maybe not selective enough. I often take all planets adjacent to my home world and then some planets with 4-8 resource asteroids around them. Then I scout for ARSs, Golems and Advanced Factories and capture them. I always try to keep my planets connected to each other.
I've been 10/10'ing vanilla basically, though I was abusing the Dyson to the point where I believe I coined the term Dyson Ball until it got itself nerfed.  I'm 50/50 about exos in 10/10.  Depends on your play style I guess.  You'll have to be careful about taking econ planets, though.  Eventually the econ gained cannot compete with the resultant AIP increases.  Particularly when you hit the homeworlds now that the maps spawn with a guaranteed 2 "Brutal" guardposts each at 10/10.  We call these Raid Engines, CPA Engines, and AI Eyes... on the homeworld.


Quote
I haven't read it but I think I will. Maybe. Dunno.
Fair enough, but if you're ending up struggling you'll probably find enough bits and pieces (and discussions with others) to be able to piece together the holes in your tactics, or at least spark a few ideas.
... and then we'll have cake.