5.021 criticisms

[Wanderer]

I do want to get rid of the double-wave-per-player-on-diff-10 rule but basically I'd replace it with base wave sizes roughly 2x what they are now (on 10), so that wouldn't really change the challenge.

That'd work.  Really you'd end up only facing 1 wave a piece from them then (43 x2 to 86, capped to 88 bombers, for example).  That'll remove that issue in the early game but keep the fierceness for later.

[Hearteater]

I put together a spreadsheet of wave sizes just to see how they looked across various AIP and at each difficulty.  Both with and without Step 9.  The green box in the top left can be used to set the wave size calculation to MIN for minimum possible wave size (for really small waves that happen right after the last wave), AVG for average size waves, and MAX for the maximum possible wave if the last wave was a long time ago.  Note the Compare tab for the effect Step 9 has on the wave size.  The compare tab also shows you the difference from the previous difficult level so you can see how much harder each difficulty is from the previous.  Blank spots on this table indicate that the mark level increased between difficulties so a direct comparison on number of ships isn't very accurate.

The initial row with 10 AIP would be the first wave of the game, and I used increments of 25 since that is the basic AIP increase for taking another planet if you factor in things like Special Forces GPs and the other odd one and two point AIP increases that end up getting destroyed.

[Nic]

I'm not familiar with high-difficulty play, but multiple-waves at 9+ could be interesting since it provides something between regular waves and Schizophrenic waves.

[Hearteater]

I redid my spreadsheet with Keith's new wave sizes.  You can mess with how Step 9 works on the Data tab.  The green squares are those that are intended to be modified.  You'll notice that with half-cap in Step 9, only Difficulty 1 is affected.  I messed around with some AIP scaling for Step 9, but honestly, Step 9 can probably just be removed if it is just going to be 50% cap.

To achieve the wave sizes for 9.3+, I simply modified the Step 8 multiplier as follows:

Code: [Select]

9.3   2.75
9.6   3.0
9.8   3.825
 10   4.5Previously those multipliers were all 2.5.

[keith.lamothe]

I messed around with some AIP scaling for Step 9, but honestly, Step 9 can probably just be removed if it is just going to be 50% cap.

I'm thinking that step 9 can be left as-is, as the main "the first wave is deceptively harder than the second" problem was due to both AI players sending their first waves at the same time every single game on high difficulties.

The only other change I'm still thinking of doing is removing the "diff 10 gets 2x as many waves" rule and simply multiplying the base wave size by 2 on diff 10.  I'm not sure that would be a big improvement to how it feels (it's still a pretty big challenge-cliff between 9.8 and 10) but it wouldn't look as strange.  I'm not sure it's worth changing.

[Hearteater]

The Step 8 multipliers I listed actually look pretty solid for 9.3+ and would prevent the huge jump from 9.8 to 10.  I'm also very happy about the change to the first waves syncing up.  I think that makes the difficulty a lot more easy to balance.

[keith.lamothe]

The Step 8 multipliers I listed actually look pretty solid for 9.3+ and would prevent the huge jump from 9.8 to 10.  I'm also very happy about the change to the first waves syncing up.  I think that makes the difficulty a lot more easy to balance.

Ah, right.  I hadn't actually read the table in your post; those numbers do look like they'll work.  We'll try them for 5.023, with the double-waves-on-10 thing removed

[Wanderer]

I messed around with some AIP scaling for Step 9, but honestly, Step 9 can probably just be removed if it is just going to be 50% cap.

I'm thinking that step 9 can be left as-is, as the main "the first wave is deceptively harder than the second" problem was due to both AI players sending their first waves at the same time every single game on high difficulties.

The only other change I'm still thinking of doing is removing the "diff 10 gets 2x as many waves" rule and simply multiplying the base wave size by 2 on diff 10.  I'm not sure that would be a big improvement to how it feels (it's still a pretty big challenge-cliff between 9.8 and 10) but it wouldn't look as strange.  I'm not sure it's worth changing.

It might be.

In general, the splitting of the waves has made this reasonable.  I'm able to reasonably defend myself with a relatively tight build order without any significant problems.  Foul it up or be on a homeworld that's an 8/4 metal/crystal or something similar and you'll have problems, but you CAN work around it, generally.  That's without the buggy tractors.  I purposely didn't use ANY just to see survivability.

However!  Let's discuss now the Sledgehammer and Mad Bomber, shall we?  For giggles, I wanted to see the effect of these opening waves from heavy handed wavers.    I even took Beam Frigates to attempt to lower the issue.  Ended up on an 8/4 world which even my econ boosts couldn't get over to full fleet, but that wasn't the important part.

Out of curiousity, i cracked open MainThreadWAveComputation.  These lines made me curious:

Code: [Select]

Performing first CheckWave with size factor of 1 on wave at Game Time: 0:06:40

CheckWave: populating count of GrenadeLauncherII with base magnitude of 144
numberUnits = kv.Value * this.WaveSize : 144
after applying the ship-type-specific cap multiplier (which includes the unit-cap-scale multiplier), numberUnits : 72

Hm, okay, so it cut it in half.  Eventually you get to this:

Code: [Select]

after applying UsefulnessInAIWaveMultiplier if any, numberUnits : 72
after applying Mark-based multiplier if any, numberUnits : 64.79
after applying at-least-one rule, numberUnits : 64.79
after applying difficulty-based multiplier (if <= 3 then 1, if <= 4 then 1.5, if <= 5 then 1.75, if <= 6 then 2, if <= 7 then 2.25, anything higher gets 2.5), numberUnits : 161.98
numberUnitsInt = numberUnits.IntValue : 161

Which brings it back up to where it was supposed to be.  I'm not quite sure why that halving reduction occurred in the first place.

Now, yes, I know the heavy-wavers are SUPPOSED to be difficult.  I'm just wondering what the effect that rule change might have?

[Wanderer]

To confirm, I've tried against a number of normal 'waver' AIs, including a few mediums.  As long as you don't get yourself caught with your pants down and you use a reasonably good DPS/Mat cost unit, the split makes the opening game viable.

To wit, though, my normal build order:
Sciences:
Crystal/Metal Asteroid IIs
Warp Detector
Bonus Fleet Ship Mk II
Grav Turrets.

Build:
15x Engi Is
A few extra converters of whatever I'm short on (so you've got 25 of them) based on asteroid count.
Second Science station
Warp Detector
Set unit build to 10 Fighters/5 Bomber/5 Missile (you want the metal/crystal usage to bounce back and forth).
Ratio in Mk 1 and 2 bonus unit to build mix, preferably at 2:1 of the triangle ships.

Use your Scout IIs to go out and check the locals.  Find out what they're packing.  In particular, look for eyebots, autocannons, or anything else that cloaks.  If they're possible, get the Tachyon I up and when you build your gravs later get a small screen of these in place as well, both at the wormhole and near the station.

WAIT.

Around 6:30-6:45 you'll get a wave indicator.  Build off 3 Grav turrets in general direction of approach close to the home station. You want their faster ships to separate from the main fleet early.  Raids and friends in particular.

Around 7:00 you should be nearly complete building the Mk1 fleet.  Toss up a 30 or so LRM turrets and 10-20 basics near the home station.  Build these one set at a time so you don't overwhelm your econ and get none of them working.

Don't engage the enemy at the gate, let them spread out.  Tackle everything heading for the home station at about halfway in.  Priority targets are anything that ignores FFs.  Raid/Cutlass/etc.  Secondary is the light starships (sick your Missile Frigs on those) to get rid of their boost.  Then frag everything else.

Second wave will notify around 10:00-10:30.  Continue to improve turret defenses, get another layer of gravs around the command station, beef up LRM and Basic turrets.  Again, be gentle on your econ, don't overwhelm it.  You're also replacing fleet ships at this point.

Once you've survived those waves, you can start building off Scout Is and start looking for your first target.

[keith.lamothe]

Out of curiousity, i cracked open MainThreadWAveComputation.  These lines made me curious:

Code: [Select]

Performing first CheckWave with size factor of 1 on wave at Game Time: 0:06:40

CheckWave: populating count of GrenadeLauncherII with base magnitude of 144
numberUnits = kv.Value * this.WaveSize : 144
after applying the ship-type-specific cap multiplier (which includes the unit-cap-scale multiplier), numberUnits : 72

Hm, okay, so it cut it in half.

That's because you're playing on normal caps instead of high.  High is actually the default as far as the code is concerned, because it used to be that that's all there was.  If you were playing on low caps it would have quartered it.  This is what it means by "(which includes the unit-cap-scale multiplier)".

Also, if the ship type in question had a different-than-normal cap (like Zenith Beam Frigates with their 0.5x cap or Laser Gatlings with thier 2.8x cap), that multiplier would be factored in there.  This is what it means by "the ship-type-specific cap multiplier".

Now, yes, I know the heavy-wavers are SUPPOSED to be difficult.  I'm just wondering what the effect that rule change might have?

It shouldn't change much on diff 10.  The individual waves would be nearly 2x larger, but they wouldn't be sending 2x as many waves as they would on 9.8 (as it currently does).

FYI, the AI-type multiplier is applied in the AI-thread part of the computation, and thus in that log file.  It looks something like

Code: [Select]

Inside AdjustNumberShipsFromAIType, multiplier: (whatever)
after AdjustNumberShipsFromAIType call, numberShips :(whatever)