Why Don’t Cheaters Repent?

[guerringuerrin]

48 minutes ago, Atrik said:

Thats cool data to have on the summary.

How is percentaje calculated? tabasco has 280 seconds idle time and burrito has 573 seconds. and yet both has 54%

Edited 16 hours ago by guerringuerrin

[Atrik]

13 minutes ago, guerringuerrin said:

How is percentaje calculated? tabasco has 280 seconds idle time and burrito has 573 seconds. and yet both has 54%

It's total idle time / total life time of buildings, while the time in seconds is just total idle time. So burrito had more buildings, or that were alive longer.

Basically it tells you the efficiency (or downtime) of your production buildings.

Edited 16 hours ago by Atrik

[Atrik]

The % are high, but you can see on the graphs that it's a lot of it that come from when the game is about to end when often some players are full pop and some have no resources at all to continue producing .

As well the buildings taken into consideration are CCs and Military buildings: so forges, workshops, forts are taken into consideration if they aren't researching or training. (towers etc DON'T have the Military class so it's only really the ones listed above).

Edited 16 hours ago by Atrik

[guerringuerrin]

14 minutes ago, Atrik said:

forges, workshops, forts

Maybe for a better perspective might be cool to exclude this

[Atrik]

16 minutes ago, guerringuerrin said:

Maybe for a better perspective might be cool to exclude this

You could. Basically you have a setting to customize the classes that are included. Since this use the trainer's functionalities, you can use its "Training Entity" setting in options.
I've just realized that means, for the experiment branch i pushed you need to have the trainer option on for it to work too.  (you can still toggle it off from the gui obviously and works when replaying, just need to have the option on for technical reasons).

Edited 16 hours ago by Atrik

[chrstgtr]

1 hour ago, Atrik said:

The % are high, but you can see on the graphs that it's a lot of it that come from when the game is about to end when often some players are full pop and some have no resources at all to continue producing .

 

Also probably makes sense to only clock idle time when a player is below max pop. 

Anyways, these are cool graphs.

[Atrik]

1 hour ago, chrstgtr said:

Also probably makes sense to only clock idle time when a player is below max pop. 

Anyways, these are cool graphs.

Yes there are some imperfections for sure. As well the graph smoothing make the first minutes inconclusive, since everybody will have a point at 100% and it takes until fcking min 5 for the smoothing to wear it away. 

[ffm2]

8 hours ago, chrstgtr said:

Also probably makes sense to only clock idle time when a player is below max pop. 

Anyways, these are cool graphs.

Also the building type (temple, house with FF researched). Context always mattered reading stats and are most useful to read while looking at the replay or in the autociv summary.

 

11 hours ago, Atrik said:

I found no correlation between usage of autotrain and idle building stats.

On 13/07/2025 at 12:25 AM, TheCJ said:

That the autotrainer provides any statistically significant advantage remains to be proven.

Problems with the autotrainer are also discussed long by a post Feldfeld. Besides the adjusted batching at the right time the meta changes that you don't need to check your base for production but can focus on raids.

On quickstart it's very easy to see that it's better, but we shouldn't care because the advantage is small. I told Atrik before he should at least make it slower and emulate a good player. The discussion was derailed, that he didn't create quickstart. Then the oddity detector came out and he slowed his local quickstart down by one turn just enough down not to get detected by it, but not as slow as a real player.