Scripting a RAM Garrison? Probably not, but check this out.

[RangerK]

Player 8 was yours truly.
Player 1 was Zekromus.

1) garrison of the unit happens the very TURN AFTER the attack.  Too fast for human reaction, right?  Because I think a turn is 0.2 seconds.  Can anyone confirm? 

turn 6766 200

...

cmd 8 {"type":"attack","entities":[11066,11525,11526,11529,11740,11741,11742,11744,12119,12389,12973,13259,13261,13264,13265,13266,13269,13270,13271,13276,13437,13438,13439,13440,13495,13517,13518,13519,13625,13627,13653,13654,13655,13656,13710,13711,13712,13750,13771,13841,13842,13843,13971,13973,14059,14060,14061,14620,14621,14632,14633,14654,14714],"target":12240,"queued":false,"pushFront":false,"allowCapture":false,"formation":"special/formations/null"}

end

hash-quick d0874eabce12f2f4dd5b62dd37b0d99b

turn 6767 200

cmd 1 {"type":"garrison","entities":[12240],"target":10148,"queued":false,"pushFront":false,"formation":"special/formations/null"}

end

2) garrison of the unit happens 3 turns after the attack on the unit. 0.6 seconds.

turn 6799 200

cmd 8 {"type":"attack","entities":[11066,11526,11529,11740,11741,11742,11744,12119,12389,12973,13259,13264,13265,13266,13269,13270,13271,13276,13437,13438,13439,13440,13495,13517,13519,13625,13627,13653,13654,13655,13656,13710,13711,13712,13750,13771,13841,13842,13971,13973,14059,14060,14061,14620,14621,14632,14633,14654,14714],"target":12240,"queued":false,"pushFront":false,"allowCapture":false,"formation":"special/formations/null"}

end

hash 144bce974a7c9b445307dceabbc89631

...

turn 6802 200

cmd 1 {"type":"garrison","entities":[12240],"target":10148,"queued":false,"pushFront":false,"formation":"special/formations/null"}

 

3) but then there's also this.  Garrison of the unit happens in the SAME TURN as the attack on the unit.

 

turn 6784 200

cmd 1 {"type":"garrison","entities":[12240],"target":10148,"queued":false,"pushFront":false,"formation":"special/formations/null"}

...

cmd 8 {"type":"attack","entities":[11066,11525,11526,11529,11740,11741,11742,11744,12119,12389,12973,13259,13261,13264,13265,13266,13269,13270,13271,13276,13437,13438,13439,13440,13495,13517,13519,13625,13627,13653,13654,13655,13656,13710,13711,13712,13750,13771,13841,13842,13971,13973,14059,14060,14061,14620,14621,14632,14633,14654,14714],"target":12240,

=======================

Probably #3 precludes the possibility of a script.  Amirite?  If so, then it was a coincidence - three times in rapid succession, the moment I attacked, he made the independent decision to disappear?

One of you experts please help me with this.

I'm 100% certain about all this, but just in case my cheating allegation was false, here you go:

 

Edited yesterday at 00:13 by RangerK
typo

[RangerK]

• Turn 6766 → 22:33.2   Garrisons 0.2 seconds after attack.

• Turn 6784 → 22:36.8   Garrisons 0 seconds after attack (at the same time).

• Turn 6799 → 22:39.8   Garrisons 0.6 seconds after attack.
   

[RangerK]

His average time per move for this part of the game was 1.43 seconds, which seems completely human, further pointing to his innocence.

Average time per move for the entire game (except the end) for all players:

Zekromus (1628)         0.98 sec
King_Okocha (1498)    2.38 sec
jimo (jimo)                    3.49 sec
Decger (1623)             1.52 sec
Atrik_III                         0.91 sec
jbc (1532)                     0.93 sec
acaradeperro (1274)    2.36 sec
RangerK (1588)            1.92 sec


These stats are so interesting.  I've never dug into a game file before.  It would be cool to show avg time per command in the stats at the end of a game.

[real_tabasco_sauce]

A turn in a 4v4 can be quite long, so there may have been more time than you expect for these inputs to be entered.

however it is a bit fishy. Maybe screen record part of the replay to see what this looks like? Or maybe give the replay so we can watch it.

[RangerK]

metadata.json commands.txt

[Atrik]

A lot of players do this trick : 
Garrison rams in a building (fort most likely but possible to make it from workshop)
Ungarrison them on approaching siege, then just after they land a hit, garrison them to safety.

It's annoying because you have no way to react fast enough to kill the trickster ram. You can still :

1. Surround your own rams with enough soldiers so that the opponent rams won't be able to unload too close to them.
2. Surround your own rams with melee on stand-ground stance (need less then for 1 but also less safe for you own rams, you might be able to kill enemy rams though).
3. Give up on ramming the fort and find an alternative (catapults, capture the fort, ignore the fort..).

The odds that your opponent use anything else then the default garrison hotkey to do this are extremely low.

Edited yesterday at 03:40 by Atrik

[ffm2]

The replay does not start from turn 0 and can't be replayed sadly. This happens on late joins.
As real_tabasco_sauce said, the turn length can vary highly.

In the oddity_detector the duration of the turn is computed, but not logged for replays (which could help a lot in those cases). Although it could vary for spectators again in comparison to the players. They can delay or speed up, so the players have less lag.

A indicator for lag (or pause) is when more players than one have lots of commands during a turn.

The reaction to garrison the ram is one a normal player would anticipate and focus on, so a fast reaction seems plausible. Same turn could also be for fast ungarrison and garrison. This is the case here. There is also some variance to see:

turn 6776 is a ungarrison and 8 turns (1.6 in game seconds) later a garrison command is send. Ram attack interval is 1.5 seconds. 

Then turn 6793 is ungarrison and 6802 is garrison (9 turns).

Turn 6776 sends the same command twice, which a program wouldn't (if it's not extra programmed to mix in slight variances to hide itself).

Does someone else have the complete replay? With this doubt in mind you could check other games of Zekromus.

I plead for not guilty.

[real_tabasco_sauce]

@Atrik i think he's talking about garrisoning into a ram, not onto a ram to kill it.

[wowgetoffyourcellphone]

We really should show troops jumping onto enemy covered rams, hacking away at it. lol

[Seleucids]

11 hours ago, RangerK said:

Too fast for human reaction, right?  Because I think a turn is 0.2 seconds.  Can anyone confirm? 

The turn period is actually scaled by network latency and computation time. By default, without latency (against AI), a turn is truly 0.2 seconds. However, as soon as the game is networked with real humans in different countries, it is much longer. This is not even excluding small freezes and pauses which prolong the turn time. During latency, a turn can be as long as 0.4 or 0.5 seconds and that is enough for a focused human to issue a command. There are European players and American players in your game, so the network latency is huge there. 

If the game was paused or frozen by lag, it's possible to queue up more than 1 command in one turn. In the laggy A27.0, I was able to queue up 11 commands into one turn, so 3 actions in one turn is not a problem.