AU mod: Cavalry

[Catt]

Originally posted by Dominae
I agree that it's not a huge deal, if you know how to do it. Also, I'm sure this is one of those slippery slope -type things; in a couple of months we'll realize we've created something as completely annoying to install and intrusive as the GOTM scenarios.

Catt

[nbarclay]

Dominae, I understand your concern about the "slippery slope" issue. On the other hand, when you start worrying too much about slippery slopes, you're on a slippery slope to never doing anything for fear of starting down a slippery slope. Seriously, the best course is often somewhere in between the extremes, but "slippery slope" arguments tend to push people so hard in trying to stay away from one extreme that they end up at the opposite extreme.

The reason the GOTMs are such a mess is that the folks who run them have a deliberate desire to create things that have nothing to do with the standard game - and to go on creating from one month to the next. The AU Mod's philosophy is far more conservative; no one really wants a new unit and graphic, but a new unit is the only way to implement a changing level of ability for a unit and a new unit would be confusing if it doesn't look different. I won't say it's impossible that we might someday want to incorporate another graphical difference for similar reasons, but given how extreme a situation it's taken for us to want to do it once, I'm not too worried.

Nathan

[lockstep]

Originally posted by alexman
we have to carefully weigh the benefit of easily identifying Light Cavalry on the map, against the potential problems caused by incorrect installation of the mod.

Originally posted by Dominae
but then a 4000BC savegame of the AU mod version of a scenario is unplayable unless you go through the installation procedure.

Okay, I get it now. At first we we were talking about possibly incorrect installation procedures (which I deem as unlikely), but the real concern is what happens to AU players that don't have the mod installed at all. I don't have a clue about the percentage (5%, 50%, or even 95%), but so far the last PtW version of the AU mod (v1.17) has been downloaded 416 times. If the panel majority nevertheless thinks that the ability to play AU saves without having the mod installed is crucial, then so be it.

P.S.: If AU courses (saves) can be created simultaneously in two versions (stock rules vs. AU rules), they can also be created in three versions (stock rules/AU rules/AU rules plus distinct unit graphics).

[Fosse]

I don't have C3C yet, so I might have NO idea... but wouldn't changing a unit graphic in the editor not require any file copying AT ALL?

I seem to recall that I could change which animation represented which unit on the units screen. Am I mistaken?

[Konquest02]

Originally posted by lockstep
If the panel majority nevertheless thinks that the ability to play AU saves without having the mod installed is crucial, then so be it.

I think it IS crucial, as it's the main point why many people do not play the CFC GOTM, they're too lazy to install the required files. I know, I'm one of those guys.

--Kon--

[pvzh]

Ok. Ok. Look what a mess we are in. 2 version of the Cavalry of the same graphics and rework of all associated UU's instead of carefuly looking at the problem and fixing the problem. In my opinion, you looking at it upsidedown.

1. Cavalry (6) indeed has 50% advantage over musketman (4), but lowering attack value to 5 will make it even worse than Knight vs Pikeman, as nbarclay pointed out. Besides, everybody forgets that at that stage of the game musketmen are defending mostly cities, i.e. +50% to defence bringing efective defence to 6. Not that poor.

2. Cavalry indeed has 3 moves, but that play a significant role only if you have overwhelming advantage in numbers, i.e. 40-50 for standart map: you move on your roads, take city use its roads take another etc. Taking 5+ cities a turn. However, you amassed OVERWHELMING advantage: you built 50 horsemen (standard map), paid all their maintains, paid 150 gold (75 gold with Leo) per horsemen to upgrade. You won this game by sheer fact that you were ABLE TO DO this. Would have you built library/university and diverted that money and maintaince cost to research, you would had a tech lead, gtp-drained AI's -- won game anyway you put it. My advice: move to the higher level: this problem will disappear.

However, I agree that taking cities with cavalry if you beelined to MT in a tech even game is fairly easy before rifleman. But why? Because, when you do so, only first 1 or 2 defenders are musketman, the rest are pikemen if not spearmen. Against them Cavalry indeed "overpowered" and "needs a nerf". The PROBLEM is MUSKETMEN and their PRICE (remember, upgrade costs are 50% up).

Thus:
Musketman -10 or -15 to cost, same stats.
Consider, dropping saltpepper requirement and/or increase in defensive bombardment.

Addition: Improving rifleman the same way (cost reduction) with even more subtle change: switch tech requirement to MT from Nationalism to reduce AI drive to research Nationalism and shorten Cavlry lifespan, but this might go to far from the stock game.

This approach
(1) requires no additional units to create (Cavalry Mk I, Cavalry Mk II) messing with respective UU, problems telling units apart and all that nasty stuff;
(2) does not change game mechanics: Cavalry vs Musketman still better than Knight vs Pikeman;
(3) fixes enefectiveness of Musketman as a whole.

All benefits without any trouble.

[Risa]

pvzh hits the nail right on the head.
30 shields for 1.3.1, 60 shields for 2.4.1 and 70 shields for 4.6.1 is just ridiculous. Either raise musketman to 3.5.1 (and musketeer to 3.6.1), or reduce musketmen's cost to 45 shields.

[Ision]

I agree with the arguments that point to the culprit as being the musketman and not the 6-3-3 Cavalry.

There is no need to go through the gymnastics of a '2 stage Cavalry' for the sake of 1 attack factor. I also believe that a no-cost to upgrade/ 2 stage unit/ with identical graphics - IS a great departure from the stock game, and this is exactly what the AU Mod seeks to avoid.

A simpler solution is a very slight cost reduction to the Musketman - combined perhaps with a slight increase in defense or attack.

Given these 2 options - a reworking of the musketman seems far more reasonable than the Cavalry changes.

Ision

[nbarclay]

I decided to do some experimenting with cavalry and see what would happen. In my experiment, I set up a scenario with 100 pseudo-veteran cavalry attacking 50 pseudo-veteran musketmen fortified in a size-8 grassland city. Rather than make them veterans (which I don't know how to do any way but one at a time), I added a hit point to regulars; that reduces the retreat odds for the cavalry a little and adds a little to the chace of promotion for the musketmen, but the results should still be reasonably useful. (Also note that the occasional 6hp elite caused some anomalies.) I ended up using Rome as my attack victim, so the promotion chance was a little better than if the victim had been non-militaristic. The results of five assaults are as follows:

1) Assault failed. 24 cavalry dead. 46 musketmen dead.

2) Assault succeeded. 31 cavalry dead. 50 musketmen dead. Four cavalry left that didn't attack.

3) Assault succeeded. 33 cavalry dead. 50 musketmen dead. One cavalry left that didn't attack.

4) Assault succeeded. 25 cavalry dead. 50 musketmen dead. Six cavalry left that didn't attack.

5) Assault failed. 29 cavalry dead. 48 musketmen dead.

Unit for unit and hit point for hit point, cavalry and musketmen fortified in flatland cities (size 7 and over) are actually fairly evenly matched. The reasons why cavalry assaults are so successful are (1) AIs built lots of regulars instead of veterens and (2) AI defenders are spread all over the empire, allowing attacking cavalry to achieve massive numerical superiority at the point of attack. Once healthy cavalry are attacking injured musketmen, and especially musketmen with just one or two hit points left, it's pretty much over.

I next reduced the attack value of cavalry to five. In the one battle fought with 100 cavalry attacking 50 musketmen, the attack failed with 39 cavalry lost and only 35 musketmen killed (although the survivers were down to one and two hit points). I then added another 50 cavalry (making 150 total) so the attacks would be at 3:1 odds favoring the attacker, with the following results:

1) Attack succeeded. 35 cavalry dead. 50 musketmen dead. 31 cavalry left that didn't attack.

2) Attack succeeded. 28 cavalry dead. 50 musketmen dead. 43 cavalry left that didn't attack.

3) Attack succeeded. 47 cavalry dead. 50 musketmen dead. 22 cavalry left that didn't attack.

4) Attack succeeded. 38 cavalry dead. 50 musketmen dead. 41 cavalry left that didn't attack.

5) Attack succeeded. 49 cavalry dead. 50 musketmen dead. 20 cavalry left that didn't attack.

Note that actual losses would be a tad less under normal rules than they were in my test scenario. The pseudo-veterans had a bit less chance of retreating than genuine veterans had, which made them more likely to die but also a tiny bit more likely to win. Also, the slightly higher promotion chance for the defenders and especially the occasional 6-hit-point elite gave the defenders a bit of an edge.

We're probably looking at somewhere in the neighborhod of one third higher death rates among attacking cavalry if we adopt the "light cavalry" modification. That may be just about right to make cavalry blitzes less overwhelmingly powerful while still leaving them clearly useful. My worst fears that the power of light cavalry might be hurt too much seem to have been unfounded.

[nbarclay]

Regarding the possibility of dealing with the cavalry problem by strengthening musketmen, keep in mind that strengthening musketmen also has the side effect of weakening knights, MedInfs, and longbowmen. Personally, I use kights so rarely that I have no real idea what the impact would be, so I'll leave analyzing the possible impact to those who have more relevant experience.

Nathan

[Dominae]

Excellent, thanks for the tests, Nathan!

[alexman]

pvzh made an interesting suggestion, which is worth considering. However, I'm not convinced that it will completely solve the problem we are trying to tackle.

A big reason that Cavalry is so powerful is the fact that you can beeline for MT and catch the AI with pikemen. The 3 moves of the unit makes it possible to finish off your first target quickly, leaving you time to roll over your next neighbor before he too gets Gunpowder. A cheaper Musketeer will reduce the time it takes the AI to upgrade its defenders after it gets Gunpowder, but the problem of Cavalry versus Pikemen will still be there. Note that if you get MT before the AI gets Gunpowder, it doesn't necessarily mean that you are ahead in the tech race, so it can happen even at your 'normal' difficulty level.

[vulture]

To summarize alexman's summary of the problem:

It's too easy to get into a cavalry vs pike war even against the more powerful of your AI neighbours, and this gives the human player a big advantage over the AI that we want to reduce.

Cavalry vs muskets isn't so much of a problem - it is not that hard to win a cavalry vs musket war, but you won't roll over them in the same way you can with pikes.

Possible solutions (most of which have been proposed here already - combinations of these may also be viable). Some of these are almost certainly outside the realm of not changing stock C3C too much, but are thrown in for completeness and maybe to spawn other ideas:

* Nerf cavalry (and cavalry UUs) until nationalism, to give the AI more time to get decent defenders in place.
* Lower the shield cost of muskts, so the AI can upgrade more pikes straight away upon getting to gunpowder (the relative shield cost / defence of pikes, muskets and riflemen is crazy anyway IMHO)
* Use flavours to encourage the AI to beeline to gunpowder (I've noticed the AI do this - I had a tech lead in the ealy middle ages, and went straight for MT - this lowered the cost of the MT branch techs enough that the other AIs all found them more attractive, so they all did the same)
* Improve the musket units (doesn't really solve the problem)
* Improve the musket units to 5 defence, and add in a new 4 defence unit between pikes and muskets (at Invention, presumably). Might be seen as a very big change, but we could just say that it is the pike becoming more effective (zero cost upgrade?)
* Prevent the beeline to MT - make it require a Theology branch tech as a pre-requisite, so it is almost impossible to get to MT before gunpowder has been around for some time.

[lockstep]

Originally posted by alexman
pvzh made an interesting suggestion, which is worth considering. However, I'm not convinced that it will completely solve the problem we are trying to tackle.

Yes, but if it alleviates the problem and does not require a new unit, I'm all for it.

A big reason that Cavalry is so powerful is the fact that you can beeline for MT and catch the AI with pikemen.

While I'm not sure if this is true, I'm very sure that the argument comes up for the first time. Until now, it was only 'Cavalry vs. Musketman is unbalanced'. Is the Light Cavalry proposal already sort of a 'beloved feature'?

Note that if you get MT before the AI gets Gunpowder, it doesn't necessarily mean that you are ahead in the tech race, so it can happen even at your 'normal' difficulty level.

If I get Military Tradition before the AI gets Gunpowder, this means at least a 4-tech lead in the Engineering branch. Unless the AI has a 4-tech lead in the Monotheism branch, I am ahead in tech.

IMO, pvzh's suggestion to reduce the shield costs of Musketmen is very interesting.

[Risa]

Originally posted by vulture
* Prevent the beeline to MT - make it require a Theology branch tech as a pre-requisite, so it is almost impossible to get to MT before gunpowder has been around for some time.

Interesting. Perhaps Education to Chemistry? There doesn't seem to be strong relation between any two of other techs.