Fortress Test Server

For things that have to do with those crazy test servers... and yeah. By request of z-man, and, of course, you gotta obey...

Moderator: Z-Man

Post Reply
User avatar
Rain
Round Winner
Posts: 300
Joined: Sat Apr 15, 2006 2:59 pm
Location: a random empty server playing with bots

Post by Rain »

worked!
END OF LINE
User avatar
Rain
Round Winner
Posts: 300
Joined: Sat Apr 15, 2006 2:59 pm
Location: a random empty server playing with bots

Post by Rain »

epsy is working on a script that will allow me to read server feedbacks easily.
so who has a comment could just say something like

Code: Select all

!FEEDBACK the rubber sucks!
and i'll read it

meantime you can use this topic of course
END OF LINE
User avatar
Rain
Round Winner
Posts: 300
Joined: Sat Apr 15, 2006 2:59 pm
Location: a random empty server playing with bots

Post by Rain »

the script is ready, even if epsy is still working on it for some other features.
if you say in game "!feedback [insert comment, suggestion, complain here]"
i'll read it via rss.

kisses
END OF LINE
User avatar
wrtlprnft
Reverse Outside Corner Grinder
Posts: 1679
Joined: Wed Jan 04, 2006 4:42 am
Location: 0x08048000
Contact:

Post by wrtlprnft »

Now that epsy removed the IPs from the feed I think it's OK to publish its URI. It's also announced in #armagetron on freenode by armabot (almost) in real time.
There's no place like ::1
User avatar
Rain
Round Winner
Posts: 300
Joined: Sat Apr 15, 2006 2:59 pm
Location: a random empty server playing with bots

Post by Rain »

new feature:
it is sumo from 0 to 3 players and fortress with 4+ players.

working on !help and !config
END OF LINE
blondie
Core Dumper
Posts: 160
Joined: Sat Jun 21, 2014 2:57 pm

Re: Fortress Test Server

Post by blondie »

it's back up

currently running Ladle settings except for the following changes
  • 5v5
  • Teams must win by 10 points
Why 5v5?

1) Strategic diversity
With six players, fortress is highly symmetrical. The standard and almost universally used tactic is two have three defensive players and three offensive players. Each base is a 3v3 fight. As a result, the game is very stable, it might even be a Nash equilibrium. Teams that deviate from standard 3 and 3 set up have not historically fared any better than teams who stick to the standard. (It's unclear if any team has exceeded expectations because of an advantage derived from straying from the standard 3 and 3 set up)

With five, that equilibrium is broken. Teams have to either match where their opponent sends the 4th man, creating a 2v2 at one zone and a 3v3 at the other, or mirror their opponent, creating 2v3s at each zone. While in 6v6 fort almost every round is 3v3 and 3v3, there are four different configurations a "standard" set-up could go.
  • Both defenses are 2v3
  • Both defenses are 3v2
  • Gold defense is 2v2, Blue defense is 3v3
  • Gold defense is 3v3, Blue defense is 2v2
This variety of set-ups could create a more varied and interesting opening strategy, where teams have to decide how to set up their five players. Every round a decision can be made.


2) Each player is more valuable, has more impact
This is pretty straight-forward, but potentially very consequential. In a team of five, each player is worth 20% of the team's overall strength. In a 6-player team he is worth 16.67% of the team's strength. His importance to the team is therefore about 20% greater in a 5-man team compared to a 6-man team (20/16.67=1.20).

The impact of having someone very skilled or very weak is therefore multiplied by about 20% by shrinking the team size. This gives every player more impact on the game and more control over the outcome, which can make the game more fun. The connection between individual actions and team performance is strengthened. However…

3) Five is the minimum number to preserve "true" fortress

This comes both from the crowdedness of the grid, but also from the grind. Below a 5-man team, the grind becomes very vulnerable, and setting up becomes chancy. Fortress isn't game that should always come down to a race to get in before people set up. The sweeper-attacker-defender dynamic is a lot of fun. Below 5v5, that is mostly lost.

In 4v4, if one person dies, we are already in an endgame scenario. In 4v4, you either have one sweeper or one attacker, so that teamwork is lost. Holing is basically invalid, because you won't have people to gank.

Of course, no one is really advocating 4v4. The reason to think about how bad 4v4 is compared to 5v5 is that at 5v5 we see the minimum required numbers for all the dynamics we enjoy in fortress. That suggests that 5v5 is perhaps the purest form. We have shrunk the players to bare minimum required for the real game to function. Trimmed all the fat.

4) Grinding tactics galore
A 5-player team has a symmetrical grind, but it's really an asymmetrical grind. The defender can play either 4 or 5. There are lots of questions about how to grind. Does the defender grind up? If so, which side does he play? Do both wings double-grind? What position does your sweeper play?

In 6v6, when we discount the defender, we have a 5-player grind, which is high symmetrical and stock-standard: every team does it the same way. In 5v5, the defender being part of the grind or not creates this instability which creates tactical questions for teams to answer.

There is real flexibility, and because both teams have flexibility, that means you get an active metagame of reacting to how your opponent is grinding.

5) Easier team organization
Although at first glance, we would think a 5-player team is proportionally easier to organize, in a competitive setting, it really is exponentially easier. There are a couple reasons why.

Most competitive teams are not centered around one person, but rather a core group of people. Sometimes one person decides to make a team all by himself, and looks for all the players himself, but usually a group of two or three people have been talking and decide to make a team, whether a long-term clan or a one-off casual thing. Anyone who has organized a team knows how tough it can be to find that six member. Teams begin by filling the most important or central roles first, without answering those important questions most organizers don't begin. Competitive teams don't find their best player last. Almost always, the fifth and sixth players they ask to play are weaker than their earlier recruits. Many times, that 6th player is just filling out the roster with someone solid and friendly. The reason teams do this is your sixth member doesn't need to be anything special, because in 6v6 fort, that sixth player is really somewhat fat, as I argued earlier. By reducing fort to the minimum, there's no longer that burden of filling out the roster. Now every player teams get is someone they really need to play the game, not just numbers to avoid 5v6.

Secondly, in a 6-player team, every player has some probability of needing a substitute, let's call it x. Just as with player impact, the sub requirement has been reduced from 6x to 5x. A lot of the time this is the difference between one sub and two subs, meaning the true size of a team is shrinking not from 6 to 5, but from 8 to 6. That has some great side-effects: more Ladle teams, less substitutes, and therefore more playing time on the grid overall.

Why win by 10?
In the past, I've suggested getting rid of kill points, but they're not going anywhere. This is an alternative solution to the problem of what happens when an opponent has 98 points. Teams are forced to play conservatively in close matches. By making a match win by 10, close games won't be decided by a single kill, which considering the existence of lag is really a shame. With a scoreline of 98-98, the match doesn't end instantly after someone dies, a team has to gain a 10 point advantage. In Ladle, we only play Best-of-threes, meaning the difference between 2-1 win and 1-2 loss can be a small as a single kill, a single slide, random lag. By making it so teams have to establish a more significant score difference, match results become more meaningful and less dependent on lag and other random factors.

With current settings, the closest matches have tense but stagnant conclusions, as neither team is willing to risk their lives. Teams still have to be cautious of giving their enemy a win with win-by-10, but at least an errant death can still be saved by heroic play of teammates. Win-by-10 creates opportunities for exciting match finishes, or suspenseful overtimes.

10 seems like a good starting place for the margin of victory, although it might be better a bit higher (12-16).
Durf
Match Winner
Posts: 426
Joined: Mon Jul 30, 2012 10:35 pm

Re: Fortress Test Server

Post by Durf »

Neat!
I'll give it a try sometime.
Definitely interested to see how some of my own strategies do in this setup.

I agree with the 12-16 thing:
In a game like table tennis, there's a similar rule where you need to be 2 points ahead to win.
1 point can be attained per 'round', so a win is only counted as a win if you're 2 rounds ahead.

Likewise, we can take the maximum points a team can earn in one round, and double it for a limit.
(I might be fuzzy on the details...help me out if I got something wrong:)
6 points for zone capture?
4 points for surviving the round?
2 points per kill? (x5 = 10 points)

total = 20 points in one round.
Doubled is 40 points.

It may seem like a lot, but consider when a game is very VERY close...a single mistake (or lucky break) on either side could change the outcome of the match.
If the required point lead was 40 or more, this would ensure that there are no "close games" and any game that is that close will be played until a winning team is more apparent.

Too high is too much (some games will never end).
Too low and it's not enough (but no change from regular fort really).

Overall, I think I'm leaning closer to 16 point lead (not likely to be earned with just one round - but it can be)


. . . . .
Again, very neat, I'd like to try this sometime.
User avatar
/dev/null
Shutout Match Winner
Posts: 819
Joined: Sat Sep 04, 2004 6:28 pm
Location: Chicago-ish

Re: Fortress Test Server

Post by /dev/null »

why is fortress still in testing?
User avatar
Phytotron
Formerly Oscilloscope
Posts: 5041
Joined: Thu Jun 09, 2005 10:06 pm
Location: A site or situation, especially considered in regard to its surroundings.
Contact:

Re: Fortress Test Server

Post by Phytotron »

It isn't, really. What happened is, years ago when Fortress was first introduced as a game mode, Z-Man had a server or two that he used for testing various elements of Fortress—and occasionally some new settings unrelated to Fortress*. All these dummies then came to take the settings of Z-Man's original Fortress test server as some sort of canonical/codified representation of what Fortress is "supposed to be."

All Fortress (and Sumo) really refers to is the zone behavior. There's no reason whatsoever that it needs to have a particular set of physics (speed, acceleration, arena size, rubber settings, etc.). The ones you usually see are just the ones Z-Man happened to choose when he set up the first test server(s)**, but that particular set has become standardized, and it's traced to this. And I personally find the stupidity of that to be absurdly funny as hell.


* e.g., cycle width, which resulted in a huge, irrational, and hilarious shit storm when these idiot kids believed that might become part of Fortress forever, because in their minds Fortress and the associated physics are inextricable.

** Which, not coincidentally (in terms of their origin being Z-Man), almost entirely mirror the default local game physics, which of course has nothing in particular to do with Fortress; they're just basic settings Z-Man liked. But when added into the above, it actually leads some of these new kids to erroneously believe that the local settings are meant to relate to Fortress, like to prep for it. All that then reinforces this stupid notion that Armagetron is defined by Fortress, like that's the primary intended game mode. So then when discussions of a tutorial come up, it's always about Fortress this, Fortress that. Blarg. Garbage.
User avatar
madmax
Round Winner
Posts: 306
Joined: Thu Mar 02, 2006 11:14 pm

Re: Fortress Test Server

Post by madmax »

Phytotron wrote:All Fortress (and Sumo) really refers to is the zone behavior. There's no reason whatsoever that it needs to have a particular set of physics (speed, acceleration, arena size, rubber settings, etc.). The ones you usually see are just the ones Z-Man happened to choose when he set up the first test server(s), but that particular set has become standardized, and it's traced to this. And I personally find the stupidity of that to be absurdly funny as hell.
The fact that, to this day, a fortress_complete.cfg is still included with those same settings could have something to do with that particular set becoming standardized...

Image
Winner of the How Many Pages Before The Lock® competition and a grand total of 18,93 euros in Euromillions.
User avatar
Phytotron
Formerly Oscilloscope
Posts: 5041
Joined: Thu Jun 09, 2005 10:06 pm
Location: A site or situation, especially considered in regard to its surroundings.
Contact:

Re: Fortress Test Server

Post by Phytotron »

Right. That's meant to be an example, like all the others included. It does add to the misconception I describe, and I don't really see their purpose or necessity otherwise. There was greater variety before that, and there weren't any "special game modes" to artificially boost that variety. They should be removed, in my view, or at least replaced with some narrower examples (like the 'single use turbo').

But, again, it's because players saw these settings as they played that they thought they must be related to that game mode—and because many of them were new and unfamiliar with the variety of actual and potential server settings available. Same way that a few other "game modes" have erected largely arbitrarily standardized sets of setting as well, like "high rubber" (how that even qualifies as a "game mode" is beyond me), "dogfight" (unwritten rules) or Sumo and CTF (zone-based). It's nonsense, and part of the reason I detest all these "special game modes." It's an illusion of variety (like the illusion of choice in the marketplace).

And then you add to the mix the fact that around the same time Fortress was being introduced and popularized, commercial server rentals gained popularity. Before that, just about every server in Armagetron was hosted on someone's home PC, so there was a certain level of entry there. When these commercial services became available and accessible, just about anyone with some money (or a parent's credit card) could throw one up (also part of why we have so many useless vanity clan servers). So, yeah, this newer crop of Admins who don't bother looking at the actual configs just go "lol wut shud i do?" and lazily just put in the Fortress include, and there you have it. From what I understand, some of these Armagetron-specific services even do that for them.


Also, way to come out of the woodwork there, madmax.
Durf
Match Winner
Posts: 426
Joined: Mon Jul 30, 2012 10:35 pm

Re: Fortress Test Server

Post by Durf »

I agree with Phytotron. Even though it's a rant, it makes many valid points.

Though I think you're confusing "standardized settings" when you think that it's that related to (rather, that dependent on) fortress.
When you think about the default physics (local), it's basically the same thing just with no rubber instead of 5.
By default, everyone plays any video game with default settings. (sounds redundant, roll with it)
Online (with lag) 0 rubber is a little ridiculous, so to me it makes perfect sense why the fortress settings are standardized in the community (5 rubber to account for lag - while balancing the difficulty of hitting walls and making perfect turns for optimal game play. In this context, 5 rubber is HIGH rubber. But in the context of HR servers, to rubber tactics, this is low).
So in the end, I think the "standardized settings" have nothing to do with fortress. Perhaps if the file was named differently, that might help (standard_physics.cfg?).
At least, not all players make these stupid associations you mentioned.

Perhaps it's just a large lack of creativity that cause the same settings everywhere.
Perhaps people enjoy those settings because they're actually fun (more than others).
Overall, I agree with you, on many of the problems (and causes of them) you mentioned.
(Too many copy-servers, "vanity clan servers", and overall servers that don't add anything new/more)
User avatar
sinewav
Graphic Artist
Posts: 6413
Joined: Wed Jan 23, 2008 3:37 am
Contact:

Re: Fortress Test Server

Post by sinewav »

MAAAAAAAAAAAAAAAX
User avatar
ppotter
Match Winner
Posts: 451
Joined: Sun Sep 28, 2008 12:45 am

Re: Fortress Test Server

Post by ppotter »

Quite an interesting read, that thread. Curious that many of the players who posted in opposition to the change no longer played when I started, whereas the players that were more receptive were still going strong, despite the change not becoming the standard.

I think one of the main reasons fortress settings have become so standardised is because a lot of the appeal from the game is the tactical/teamwork aspect. I compare fortress in my head with football (soccer) a lot. Neither game has changed significantly over the years, there was no need to change any rules(/physics) while there was still scope to develop tactically. If you don't have to adapt to new settings in each server, you can focus on your strategy. Clearly, this is not your preference, however it seems that the vast majority of fortress players prefer this way, and if they don't, they don't care enough to try and force a change, they don't attempt to make their own server with new settings. There have been some experiments with settings, but they've never stuck, and I doubt they ever will with the ladle ingraining those default settings into every fort player.
---
Going back to that thread, interesting how seemingly frowned upon centre attacks, double grinding and holing were. It would be interesting to see a match of the best players from that era against a team today. I suspect even an average modern team would win comfortably. Even if the older players were technically superior (no real way of knowing either way), judging by that thread, tactics were very basic back then. Even if they resorted to 'dirty' tactics like holing and centre attacks, it takes some time to become efficient at them, most players these days know how to counter such techniques also. As much as you can counter holing anyway...
User avatar
/dev/null
Shutout Match Winner
Posts: 819
Joined: Sat Sep 04, 2004 6:28 pm
Location: Chicago-ish

Re: Fortress Test Server

Post by /dev/null »

To play devil's advocate (harhar) why do zone based servers have such terrible rubber compared to the same server without those patches?
Post Reply