If armies are auto-refitted during winter and costs deducted then this is another major reason towards finding a realistic way to have armies disbanded when taking serious losses and being forced to retreat to enemy territory.
I think that is the exact moment the check should take place because the retreat itself costs losses to an army already on the run with supply wagons either lost or unable to reach it (lines cut by the enemy). The army is entering enemy territory with garrisons, patrols, risks of ambushes, through a forced route and forced march with extreme difficulties in foraging (which is not the same situation for a winning army strong in spirit and numbers entering enemy territory).
This change would release important manpower and funds to the AI/players for the raising of a new army when taxes are collected and it would also address the fact that most armies are photocopies of pre-existing armies (due to either lack of funds after paying the mandatory units or manpower). We could have part of the manpower of the disbanded army return to the provinces, the upkeep would be saved and the winning player could release his "cat" army and keep his territories safe from pillaging. Both sides would benefit from this policy.
In regards with the Honjin, I only lost it twice but I never lost the battle when I did. Perhaps the % routed could be just slightly tweaked up from 3.33 to 5 but I can't really relate to that because as I said, I only lost it twice (and killed once, in the tutorial).
What I can say is I didn't see an appraisable change... still... when a sub-general is lost, all its group needs to pass checks and I'm wondering if the CiC is lost maybe all sub-generals should pass checks too so that the CiC's loss could be indirectly transferred to all the army.
What's problematic is the attitude of the AI which does attack the Honjin when it should not but I've never seen it defending it. There have been times I could have killed it but decided to use troops to fight and win the battle instead. With or without changes in this mechanic that's a cause of concern to me.
It's likely that if the problem of AI units flanking far and being out of the whole battle is solved and if the AI starts to use missile superiority and attack fortified units the proper way the player will actually never be in a position to kill the AI's Honjin... so I guess it's all secondary for the moment.
In regards with hidden statistics:
Gary Grisby's WBTS has a system (optional) where it takes some time for the stats to become visible to the player and he's always looking to assign the right general for the task.
Neither the player nor his opponent knows a statistic until it is discovered.
For the retraining of units it just takes time but for combat... i.e. offense, you need that general to attack a few times before his offensive capabilities are revealed (at that point you may choose to sack him or retain him). That is intriguing (You can opt for Historical, semi random and random) and of course both realistic and risky.
On one side your Johnston could be no match for Mc Dowell and you need to find out before 1st Bull Run happens or it would be an early catastrophe for CSA.
But on the other side this means every time you start a campaign you never know, so every campaign is different.
Add to this, the values are masked.
You can't go to the files and look at your Johnston or at your enemy's Mc Dowell: every time these 2 meet (actually every time anyone meets anyone else) it's a surprise... a possible discovery and a possible choice of sack/retain as counter-move in the next turn.
In SJ we have 2 stats, the CR and the DR.
A player who knows that particular general has a DR of 7 and who has played the other side, knows what the DR of the opposing generals is and he has a critical and unfair advantage... and since stats are not randomized, there's no point in hiding them because they are already known: this means that particular general will ALWAYS win against that particular opposing general.
Not all those who were known to be good generals actually performed well during the wars that were fought from the dawn of ages till nowadays. Ewell had never underperformed but he failed at Gettysburg (with disastrous consequences). Longstreet had to be replaced several times during the war due to nervous breakdowns (same with Sherman). Stuart was possibly the greatest cavalry officer that ever rode but he was out at Gettysburg. What about Grant... regarded as a drunkard and then emerging as the winner?
(I could carry on for ages, I'm an expert of SJ and also of the American CW periods)
Every situation is a new one, every subordinate adds or subtracts and every battle is an incognita. That incognita is the randomization but it can't be a real incognita without the hiding of the stats.
We can't hide the CR because it's needed to calculate the unit's range to the sub-general (you've got to see it on the battlefield) but we could use a MR (morale rating) which goes hidden (and possibly randomized by option) for a number of battles where the check to reveal it to the player becomes increasingly easier to pass. The same should go for the DR... this is not a duel in a temple courtyard but on the battlefield and anything may happen, unrelated to the general's skill (one of the generals might have been wounded PRIOR to the duel taking place or he might be more fatigued).
So, again, after a few duels, the player would know (if the general has not yet been killed in the meanwhile) if he's a good duellist. He would also possibly know if his MR is valuable and regardless of his DR he could choose to keep him safe).
But in order to add a system of this kind, the player should have access to the pool of generals and be able to review, select, assign and sack those he thinks proper. That would be the task of the Daimyo, the only one who can't be sacked... the one the player basically is impersonating in the campaign. I wouldn't rule out the possibility of a player to hire a ninja to target a particular general... provided he can afford it.
