AI unit upgrade...

[greedisgood]

I think the AI should upgrade every units it has as soon as possible... Its kinda lame when the AI has rifleman and knights attacking you...

Just a suggestion...

 

[r_rolo1]

The AI upgrades all the units that it can upgrade AFAIK ...

 

[Cybah]

what is better: lowering the research rate (like the AI does atm) or set the research rate to 0 to get as much gold as possible for upgrading? I think human players do the last thing if not hunting for a tech.

 

[greedisgood]

what is better: lowering the research rate (like the AI does atm) or set the research rate to 0 to get as much gold as possible for upgrading? I think human players do the last thing if not hunting for a tech.

Setting to 0 would be good... The civ I was talking about was the biggest civ or the civ with the most units therefore unit upgrade is not implemented as much... What about switching AI production to wealth until everything is upgraded??

 

[bob_x]

I only upgrade stuff when I'm at war, and even then, only the units likely to face battle.

 

[Jaybe]

I lower my research for upgrades only if needed and garrisons of my inner cities may remain at archer types throughout the game (huge hemisphere maps).

I almost NEVER attack with weak/obsolete units, however. Why give an enemy free experience without damaging him?

 

[Pfeffersack]

I lower my research for upgrades only if needed and garrisons of my inner cities may remain at archer types throughout the game (huge hemisphere maps).

I almost NEVER attack with weak/obsolete units, however. Why give an enemy free experience without damaging him?

Depends on how outdated a unit is, IMO. If the combat odds are below 5%, I agree that is is most of the time not worth. But above that, there is a decent chance to damage the enemy unit at least. Sometimes this softening beefs up the chances for your regular, new units you don't want to lose and allows you an attack at much better odds - and if you win with the next unit, your opponent will not benefit from the gained experience (except GG points). Circumstances and situations were this tactic works are more frequent in mods like Planetfall or FFH, though (in the unmodded game, the game is more focussed on mass/SoD warfare and so this tactic only works if you have enough outdated units, when facing a stack)

 

[BobTheTerrible]

Occasionally I'll throw obsoleted units at a stack/cities to soften them up before I attack with my regular forces. It saves the gold of having to upgrade/maintain them, and occasionally one will do decent enough damage to help the main forces.

I personally think uits should slowly upgrade themselves for free (obviously, beyond the scope of better AI). I'm one of those people who ends up with warriors and archers defending my central cities in the modern age, because I concentrate all of my defense on the border zones.

 

[Yakk]

If there is an obsolete unit that upgrades to a unit you are building:
A> Increase cost of the unit by 10% when building it, and
B> Generate a 'fair' chance that the hammer-difference will be spent on a spontaneous unit upgrade.

Ie, if you are building a 100 hammer unit, and it is an upgrade to a 50 hammer unit...

It costs 110 hammers to build the unit. When you build the unit, you have a 10/(100-50) = 20% chance to cause a random 50 hammer unit you own to upgrade to the 100 hammer unit.

This isn't the scope of better AI, and in fact breaks down in some mods (where you can upgrade a more expensive unit to a cheaper one, or the same cost, etc).

 

[PieceOfMind]

Obsolete units can have really good odds against superior enemy forces if that enemy is already damaged. A rifleman backed up by a longbowman will almost always beat just a single rifleman. When attacking cities, the units that you intend to use as garrison in the city can actually be used to kill the last few defenders if they are injured enough. This depends on how well the city needs to be defended on the turn of capture, of course.

 

[Roland Johansen]

I would like to remark that this thread is about what is best for the AI. That's really different from what is the best practice for the human player.

The AI already has trouble killing human units when it is on an equal footing because it is not as capable at using collateral damage, defensive bonuses, the right unit types, promotions and local superiority in forces. If the AI attacks with inferior obsolete forces, then it's military capabilities are appalling. Next to that, the cost for the AI for upgrading units is half that of what the human has to pay, making upgrading a very efficient means of changing gold into military power.
I'd rather have the AI use its gold for creating modern units instead of using it to upkeep large numbers of obsolete units.

 

[Cybah]

I think it would be recommended for the AI to cancel an attack and move with all units temporary inside cultural borders to upgrade under special circumstances (lower power total than opponent? iGold higher than X?). I also support the suggestion of switching to wealth.

 

[Cybah]

bump...

 

[Caboose]

The AI already has trouble killing human units when it is on an equal footing because it is not as capable at using collateral damage, defensive bonuses, the right unit types, promotions and local superiority in forces.

Isn't that what Better AI is trying to improve?

I agree that the AI should prioritize gold for upgrades when at war and have many obsolete units.

 

[Roland Johansen]

Isn't that what Better AI is trying to improve?

Of course. I used the argument as a reason for the AI to upgrade its troops so that it will not waste many hammers worth of obsolete units but upgrade them to better units that have a chance to accomplish something. Humans also perform better with upgraded units and we would all upgrade our units if the upgrade costs were as low as they are for the AI.

I agree that the AI should prioritize gold for upgrades when at war and have many obsolete units.

Ok, we agree.