Forged Alliance Forever

Re: Replays.

2012-08-20T08:23:54+02:00

2 replays with 3619 test mod. don't know how helpful these will be, as i was dead tired when I played these games (not balanced either). Something should be better than nothing though.

The shaders are beyond awesome, and T3 land definitely feels better at -20%

Statistics: Posted by Icy — 20 Aug 2012, 08:23

Re: Replays.

2012-07-21T20:24:26+02:00

I'd say to increase the dps a bit again. Otherwise they're going to be spammed even faster than now. Don't forget that the AA is still the same and the AA is decent.

Statistics: Posted by Plasma_Wolf — 21 Jul 2012, 20:24

Re: Replays.

2012-07-21T20:53:17+02:00

I'm concerned about the Seraphim t3 sub new DPS. It seems too weak now. I can't post the replay, because it was a Nomads replay, but I felt it was underpowered in a game where it should have made a difference to go t3 and it actually was a mistake because it was easily overwhelmed by t2 and t1.

According to the changelog, it lost a third of its DPS. If it's accurate, it's too much IMO. It loses badly to its cost in t1 subs and t2 ones. Of course it has a much longer range, but they can overrun it and the range itself doesn't justify such a high cost anymore.
I think it should either get some DPS back or stay as it is but at a cheaper price.

Statistics: Posted by pip — 21 Jul 2012, 19:13

Re: Replays.

2012-07-15T00:36:36+02:00

besides that you can't kill exps with t3 land adequately, now you can't use it vs ACUs as well (snipers are the only option)
t3 land is now shit

PS in that replay I could've killed 2 ACUs, but it seems like you need a t3 army in equality to exp mass wise? REALLY?? WTF YOU SERIOUS???
PSS yes I'm mad

Statistics: Posted by Raging_Squirrel — 15 Jul 2012, 00:36

Re: Replays.

2012-07-14T16:37:53+02:00

I Just lost my rights to come here because I'm not able to make a valuable post without trolling.

^ I didnt write that.

Statistics: Posted by Karottenrambo — 14 Jul 2012, 16:31

Re: Replays.

2012-07-14T16:11:53+02:00

Me and benny ran three tests:

First test, each player had a paragon, 50 T1 engies and 1 T3 engie to start. Benny started spamming T3 bots, I started spamming T4 bots. I have 3 megaliths by the time he has 10 T3 bots, and totally crush him.

Second test, same replay, we repeated the same conditions, but intentionally unbunched engineers. The result was similar, with my first mega produced when benny had 2 bricks, and me easily overwhelimg small numbers of bricks with many XPs.

Third test, same replay, we start with just a T3 factory and a paragon, no engineers. I was able to produce a megalith in the time it took him to produce 3 bricks, because I could build engineers with the factory while other engineers were producing the megalith, whereas benny could not build both attack units and megas at the same time.

Experimentals are clearly the most efficient way to build late-game attack power, by a large margin.

Statistics: Posted by FunkOff — 14 Jul 2012, 16:11

Re: Replays.

2012-07-14T12:43:55+02:00

At least it's some test, and at worst, prove that (some) T3 land produced with the same BR than a XP is very capable of destroying it (for those who doubt it).
Of course, loyalist are the best units for that (loyalists + firebeetle are a even better combination). And the GC one of the worst (because the claws will stun the GC every time they trigger). But it also apply to monkeylord and megalith (slow turnrate, easy to turn around it), and even ythota (you can avoid the big ball of death), and any other fast t3 bots.

I agree that it doesn't match any real game situation, but at least it's something.

If you don't agree with that procedure, post your replays and your own analysis.

We are not dumb, we don't need you to point out the flaws of any type of test. This is not the other balance forum, the people who watch these tests are not complete morons.

So :

1. Post replay
2. Post analysis
3. Stop posting sterile debates.

This will be the only warning.

Statistics: Posted by Ze_PilOt — 14 Jul 2012, 12:43

Re: Replays.

2012-07-14T09:11:09+02:00

FunkOff wrote:
uberge3k wrote:Here are tests of experimentals vs T3 land with the current balance.

Ah, more of Uber's "tests":

First "test": Spindrah uses a T3 engineer to build a GC unassisted as Uber uses a T3 factory assisted by a T3 engineer to build loyalists. Nobody plays this way, rest is irrelevant.

Second "test": Spindrah produced 1 GC with a T3 engineer and 100 T1 engineers. Uber produced loyalists with 1 T3 land fact, 1 T3 engineer and 100 T1 engineers. When spindrah's GC was complete, Uber has 6 loyalists. Spindrah stopped building, walked the GC to uber's base (on 3v3 canis) and by the time it arrived, uber had enough loyalists to kill it.

Terrible, horribly biased test performed by people who don't understand how tests work.

I think you are right. While this test is totally artificial it cannot provide any information about real game situations. The problem is, that you simply CANNOT really test in FA with predetermined games. The "knowing" factor is just too big. In FA you can only test in game situations, and thats the reason why it is so hard to balance units, that arent built often like Novax or something. But t3 is pretty much often, and I have to say, since the build power adjustments it is quite fine. You still can rush an experimental but it is also nearly possible to build enough t3 bots to defend against it.

Statistics: Posted by Kryo — 14 Jul 2012, 09:11

Re: Replays.

2012-07-14T04:49:20+02:00

loyalists vs GC's... Best matchup ever...
what about harbies, t3 tanks, or titans, where the GC does work with the tractor claws, and not gets hard countered by one unit from one faction?

Statistics: Posted by Cerberus — 14 Jul 2012, 04:49

Re: Replays.

2012-07-14T04:17:32+02:00

Also, note during the second test with 100 T1 engineers, uber spent under 111 mass per second on loyalists (his peak was 111, but he wasn't always building) whereas spindrah spend 647 mass per second on loyalists. It's almost as though it's possible to build experimentals SIX TIMES FASTER Than T3 units.

Statistics: Posted by FunkOff — 14 Jul 2012, 04:17

Re: Replays.

2012-07-14T04:12:21+02:00

uberge3k wrote:
Here are tests of experimentals vs T3 land with the current balance.

Ah, more of Uber's "tests":

First "test": Spindrah uses a T3 engineer to build a GC unassisted as Uber uses a T3 factory assisted by a T3 engineer to build loyalists. Nobody plays this way, rest is irrelevant.

Second "test": Spindrah produced 1 GC with a T3 engineer and 100 T1 engineers. Uber produced loyalists with 1 T3 land fact, 1 T3 engineer and 100 T1 engineers. When spindrah's GC was complete, Uber has 6 loyalists. Spindrah stopped building, walked the GC to uber's base (on 3v3 canis) and by the time it arrived, uber had enough loyalists to kill it.

Terrible, horribly biased test performed by people who don't understand how tests work.

Statistics: Posted by FunkOff — 14 Jul 2012, 04:12

Re: Replays.

2012-07-14T02:07:10+02:00

Here are tests of experimentals vs T3 land with the current balance.

The complaint about experimental balance seems to be, "it's too easy to rush an experimental because they build so much faster than T3 land! T3 land needs to be built faster in order for it to be competitive!".

To test this hypothesis, we removed as many variables as possible:

- Both players start with a T3 land fac and one T3 engineer.
- Both players have a Paragon.
- Fog of war is off; player with T3 land defends while player with T4 attacks.

Results are attached. Unsurprisingly, the experimental was smashed both times by a far superior number of (less expensive!) T3 bots.

In the second test, we both started with 100 engineers in addition to the t3 fac and t3 engie, and waited for the pathfinding to "settle" before beginning tests (to remove any randomness due to some engies assisting first). The results were nearly identical.

206500-TAG_UBER.fafreplay

206536-TAG_UBER.fafreplay

Note that if we wanted to be even more accurate, we should have the land player utilizing the remaining mass that they "save" compared to the experimental-rushing player to make more land or perhaps air to hurt the exp even more. This was not even necessary, however. And, perhaps most importantly, that your T3 land can be out raiding as soon as the first bot rolls off, while the exp is useless until it's finished.

Statistics: Posted by uberge3k — 14 Jul 2012, 02:07

Re: Replays.

2012-07-09T18:55:49+02:00

But well, this is not over, you can still (and very welcomed) to post replays + analysis about any aspect of the game !

(surprising as it sounds, because I'm a well known dictator, most if not all of the changes in 3615 are not originated by me).

Statistics: Posted by Ze_PilOt — 09 Jul 2012, 18:55

Re: Replays.

2012-07-09T18:49:08+02:00

It was on the facebook, twitter, welcome message of aeolus, and Uber made regular ads for balanced test games in the chat. We can't do more.

We surely can add a popup before and after every game you play, but if nobody read the previous ads, I bet they will also click okay to close them without reading.

About the "one test/week", I've tried that with bombers, trying to make it fun too, it was a big fail :
viewtopic.php?f=11&t=231&p=1952#p1952

Statistics: Posted by Ze_PilOt — 09 Jul 2012, 18:49

Re: Replays.

2012-07-09T18:31:22+02:00

Ze_PilOt wrote:
These were meant to be tweak with the help of players providing replays. Check the replay section to have an idea of the commitment. Except Uber and Rock, nobody bothered.

Maybe you should advertise the balance-test-mod more. I bet most players (including myself) have no clue on what is going on, find out too late, or don't know how/if they can/should help.
E.g. automatically ask some questions in the chat, or even have a popup: "Do you like change xyz? Try it out, and provide a replay! The decision on including the new feature is in x days."

Or do a "test week", one week forced balance testing for everyone, afterwards a revert and an open discussion about the changes.

Statistics: Posted by Benoker — 09 Jul 2012, 18:31