Our defense after marines

[vulture]

I've been doing a bit of number-crunching for defense analysis (also relevant to the attack on Lego). The scenario is a simplistic one - an attack on a civ with 6 cities: 3 coastal, 3 inland. The defender has 30 infantry. I assume that all coastal cities get the same number of defenders, and all inland cities get the same no. (e.g. 4 in every inland city, 6 in every coastal one).

The first attacker has 40 marines and 20 tanks. Results in table below.

Column 1 (#coastal) is no. of infantry per coastal city.
Column 2 (#inland) is no. of infantry per inland city.
Column 3 (marine) is % chance of marines taking one coastal city
Column 4 (n) is average no. of inland cities taken by the tanks if the marines land.
Column 5 (N) is the average number of cities taken in the attack (n + 1 (for coastal city) multiplied by prob. of succesful landing). This last column is slightly controversial, so bear with me.

#coastal #inland marines     n      N
   10       0      59.1    3.00   2.36
    9       1      80.7    3.00   3.23
    8       2      93.9    2.99   3.75
    7       3      98.9    2.67   3.63
    6       4      99.9    1.95   2.95
    5       5      100.    1.43   2.43
    4       6      100.    1.08   2.08
    3       7      100.    0.97   1.97
    2       8      100.    0.87   1.87
    1       9      100.    0.68   1.68
    0      10      100.    0.43   1.43

Why is the last column controversial? you ask. Well, consider the last row. With 0 defenders in the coastal cities, there is a good chance that more than just the target city will fall. If tanks or marines can reach cities along the coast, those cities are toast. If one of the inner cities falls it may allow access to an undefended coastal city. Worst case scenario, all 3 coastal cities fall. But maybe they can't all be reached. So the N value for the final row is either 1.43, 1.43 or 3.43, depending on geography (i.e. it is very situation-dependent).

Case 1: no extra cities can be reached, and the table is accurate as it stands. Peversely, the best approach is to empty the coastal cities entirely, and stack the inland cities with defenders (really, you should count unthreatened coastal cities as inland cities for this purpose). Since you don't have enough defenders to stop the marines, realistically speaking, then your choice is essentially to let them use 40 marines to take a city killing 0 defenders, and then 20 tanks aganist all 30 infantry, or to let them use 40 marines to take a city killing say 3 defenders (without changing the odds of them taking the city), and then having 3 fewer defenders to defend against the tanks.

In case 1, the worst scenario is to put 7 defenders in each coastal city - getting the maximum number killed (with fewer to defend inland cities) without significantly affecting the odds of keeping that city anyway.

Case 2: all coastal cities can easily be reached. N for the last row is 3.43, and for the 1/9 row is probably close to that as well. The number at 5/5 is pretty much accurate - ignoring the effect of marines being able to assault a second city (we assume only tanks can do it). So we have minimum values at 10/0 and probably 4/6 (that's guesswork - it could be 5/5 or somewhere below that). We have maximum values at 8/2 and 0/10 (or 1/9 - it's hard to know).

[vulture]

Part 2: much the same as before, but now we have 30 marines and 30 tanks (rather than 40/20).

#coastal #inland marines     n      N
   10       0      00.8    3.00   0.33
    9       1      21.9    3.00   0.88
    8       2      45.2    3.00   1.81
    7       3      72.4    3.00   2.90
    6       4      91.7    2.90   3.58
    5       5      98.7    2.38   3.34
    4       6      99.9    1.94   2.94
    3       7      100.    1.59   2.59
    2       8      100.    1.22   2.22
    1       9      100.    1.04   2.04
    0      10      100.    0.99   1.99

The N values for the last few lines have the same problems as before: the last one could be 1.99 or 3.99.

It matters less in this case though, because the 10/0 strategy is so much better than the alternatives, subject to one caveat. The 10/0 startegy is a gamble - 92% chance of losing nothing, vs 8% of losing everything, whereas with the 5/5 strategy you're looking at a pretty reliable loss of 3 cities no matter which way the RNG goes. Which is better depends on game circumstances - which cities are most valuable to keep? Is losing 3 cities as bad as losing all of them in practical terms (doomed, but going down fighting for a bit longer)? Those kinds of questions can't be compensated for except in a specific scenario.

The conclusions I draw from this preliminary stuff are:

if you have the defensive manpower to have a good chance of stopping a marine attack on the beach, the 'shell' strategy is (probably) a good bet. That way you neutralise the tank part of his force, and don't lose a city.

If the attacker is going to have a good chance of taking a coastal city regardless of what you do, and if only one or two cities are under threat from marine attack, then a very light presence in those cities is the way to go, with the infantry defending the interior cities instead. If lots of cities are under threat, you probably want to defend them to a greater extent to avoid giving away too many cities for free.

Some more caveats and notes: infantry in these scenarios have a defence of 22.5 (+100% terrain / city bonus, plus fortified). In all the scenarios thanks only attack inland cities - allowing them to attack the coastal cities as well adds another layer of complexity to the simulations, which would take a while to sort out. Hence in the 10/0 scenario, 30 tanks may well be able to take another coastal city (95% chance, 0.01% chance of taking both) as an example. This screws up all of the above numbers of course, but hopefully a) I can sort something out to improve on this, and b) the conclusions might not be too sensitive to this change.

[DeepO]

okay, let's apply this to our current situation. Right now, there are no tanks (Lego won't have them before turn 245), but we face marines in 9 turns. Lego has what we assume, plenty of cavs, and enough marines to take out any city. So, at first, we go for a more shell like approach, with interior cities only defended with 1 or 2 defenders (just to not give them away to wandering cavs). Over the turns, Lego's marine potential increases, and we add more defenders to the interior, especially once we know they've got tanks?

Sounds good to me. What about the ration of defenders in cities, as to defenders on mountains / hills? are we going to leave flat land undefended?

The problem I'm seeing is that currently, we have no way of attacking any invasion force. We're good at defending, but not counting our cats, we only have offensive power of 4 in a couple of knights, and in the rifles... it would be devastating if Lego simply lands a bunch of infs on our coast, we can't defend against that! (Don't give them any ideas)

I'll repeat my request: it is time for a precise, and detailed defense plan, or there is no way we can implement it before Lego is able to do serious damage to us

(the simulations are nice, vulture, don't get me wrong!)

DeepO

[Cort Haus]

I'll PM Theseus.

Meanwhile, some thoughts :

We'll need Rubber from GoW not later than two turns before Lego get Amphib to upgrade the Rifles.

Could Destroyers be our best offensively-defensive weapon? Trouble is, you can't converge them on one point - which is what we're basing our attack aspirations on.

Artillery and Bombers are a good way of taking HP from Inf pre-Tanks, but we still need something to finish them off. Guerillas? They attack with six, don't they?

How long between when Lego could have Marines and we can have Tanks?

[Cort Haus]

- From Turn 225 onwards, we're facing Marines from Lego. With chaining, they can attack us that same turn. I doubt they will, but we can't take any risk... so our defense has to be in place at turn 225.

With our own picket line we could see a chaining op get set up by the first bunch of transports a turn or two before. Like us in our offensive mode, they'd have to take out a wide area of the picket to spread us thin, or indicate where the blow will fall.

Either advanced pickets and/or city snooping could detect a transport build up in danger zones.

For Lego to attack us they have to be either thinking of a pre-emptive strike to knock GS out before a GS-GoW alliance can hit them, or just some thorn-in-the-side attacks to stop us playing our own game and being forced to react to theirs. If they were to commit to an attempt to destroy GS they'd be leaving their arse somewhat vunerable.

[Cort Haus]

I've been staring at this thread for a few hours and a few beers tonight and I keep thinking how Intelligence could be key here. My defensive instincts are to get as many turns warning as possible and screw up the enemy's plan.

How many units will Lego send overseas if GoW move a bunch of transports within range and a few from ND too?

On the other hand, if we strike a very efficient looking defensive posture at Lego, including advance pickets, we might just give them ideas that they weren't going to do anyway and get our own naval buildup seen.

Could RP try any suss out who is active on Lego and whether they are the hard-core builder types or pragmatists like Vondrak?

[DeepO]

Originally posted by Cort Haus
We'll need Rubber from GoW not later than two turns before Lego get Amphib to upgrade the Rifles.

that was my idea as well, but if we can avoid that, and have 2 more turns, it would help me a lot in economical planning. Otherwise, we should mobilize earlier, as infs can't be efficiently built without it. (tanks can't either)

Could Destroyers be our best offensively-defensive weapon? Trouble is, you can't converge them on one point - which is what we're basing our attack aspirations on.

No indeed, they have disadvantages. Most importantly to me: if we show Lego we've got tons of destroyers, they will build their navy as well, making it more difficult for us to destroy their pickets when the time comes.

Artillery and Bombers are a good way of taking HP from Inf pre-Tanks, but we still need something to finish them off. Guerillas? They attack with six, don't they?

I think so. We could upgrade our MI, we've got more then a couple of them. However, are we going to do it as a precaution, or once we see Lego approaching our coast?

How long between when Lego could have Marines and we can have Tanks?

1 or 2 turns at most. I think they will appear at the same time.

Other replies later, got to run
DeepO

[DeepO]

One of the problems I see is that some of Lego's pickets can threaten our coast right now. Also, what are we going to do if Lego destroys our pickets? So far, I spotted 2 ironclads, which would easily destroy 2 galleons, leaving a big gap for them to navigate through.

Should we upgrade our galleons to transports now? Do we have to avoid that? How are we going to make the transition from our picket line to offensive mode, without warning Lego 3 or 4 turns in advance something is happening?

Also, some other things: I would rather not start naming alll our units, but wouldn't naming our transports help? It does igve info to the enemy, so we rather not overdo it, but as you can see from the turn report, it looks a mess.

I would really prefer named positions on a map (e.g. post a map, name some spot "picket1", indicate a destroyer should be present there), so I can use this while playing, and while reporting.

CH, more questions: I don't know how active Lego still is, but it's safe to assume that Vondrack holds the biggest influence in the team. He had before, and probably still does. They have never been true builders, and certainly aren't right now.
How much troops can they commit? Well, seeing that their productive power is 731 (Thanks Krill!), they are probably producing less then we are. We've got less waste, and are using almost all our shields efficiently. So they can wield about the same forces as us, with similar risk. However, they still need infrastructure: they will be finishing their factories now, and probably want coal plants as well. Plus, in a few turns, they will get Sanitation from Vox. I'm hoping that because of this, we will outproduce them in # of forces. Right now, we're vulnerable to them...

Upgrading is a big question: We've got plenty of units laying around, and have cash. However, are we going to upgrade now, and rise in power already, or wait for as long as possible, and only upgrade at the last possible moment?

DeepO

[Cort Haus]

Vulture's analysis from the Spartan Acadamy thread (text)

reposted - also on T217 thread

I've had a look at which of our towns connect to which in the event of a succesful amphibious invasion into one of our ports. The bad news is that with 3 move units, they can hit every one of our inland cities (and coastal cities for that matter, apart from Inchoff), starting from any port.

With 2 move units invading, taking any port south of Monsoon or Hurricans (including both cities) gives access to the entire south. Dissidentville, Santa Ana and the port next to SA (whose name I can neither remember nor read on the screenshots I have) from an isolated group across the middle - and invasion could take all three, but with two move units couldn't get any further. Everything from Inchoff upwards is isolated from the south with regard to two move units. EXCEPTION: if Dissidentville remains with its current lack of culture, that is a gateway from a northern invasion. It need to get to 100 culture to control D'ville 88, which is enough to protect it against a 2 move invasion from the north.

Invasions in RP territory at any port can chain through to Elipolis (they have to go through San Antonio with 2 move units), and can't get any further than Elipolis, Inchon and Winds of Change in a single turn.

So a pure tank invasion could take out all our good cities in one turn, landing in any of the southern ports, or take out RP and our three north-most cities with a northern invasion. If they are daft enough to invade the D-ville belt they get three towns. A cavalry / ansar / rider invasion can in principle take everything in one turn, and has no problems with artillery support for cavalry attacks (although the artillery will have to be stationed outside of towns).

Sticking to talking about 2 move units, there aren't reall any funnel points for fighting in the south, beyond the fact that the only route between the west coast towns and the southeast is through Sandstorm and then Eye of the Storm. EotS can only be reached from Hurricane, Cyclone or Sandstorm. Sandstorm can only be reached from EotS, Arashi and Tempest.

and here are the diags.

[Cort Haus]

Comment from DeepO moved from the turn thread :

Very roughly, does this mean we need 1 inf in every city, and a couple more defenders in the chokepoints for 2-move units? But still, how many units do we need on the coast and what ration between coast and interior cities?

How thick the shell? is the question. I'm not sure we can answer that quantitavely without an estimation of what could be fielded against us. We must occupy the forts on the high ground too. Seeing as Lego's productive capacity is not currently vast, and they can only prebuild as many Marines as they devote barracks cities to the task with prebuilds the turn before, they are unlikely to be able to deliver more than 10-12 on the first turn at the most. Factor in the unlikelihood of attacking a metro with these numbers, and we need the number of Inf in the other cities that would be requiredaccording to Vulture's table (perhaps on the Sparta thread).

[Cort Haus]

I'll post more thoughts on defence later - after playing some tennis and watching England v Portugal

[DeepO]

I agree that the first few turns, our defense can be minimal (but needs to exist nonetheless). But it's not that extra-ordinary to see Lego building 8 marines each turn. By 230, they can easily have a group of 40 marines ready, which would beat 10 infs in any coast city, or on any forted mountain.

One thing, though: around that time, we would have radio (5 or 6 turns after tanks), meaning we can build radar towers for the extra defense. It would be nice if a plan of those positions also exists by that time, so the moment we can, our complete territory is protected.

DeepO

[DeepO]

Oh, and good luck to England, CH!

DeepO

[asleepathewheel]

Originally posted by DeepO
One of the problems I see is that some of Lego's pickets can threaten our coast right now. Also, what are we going to do if Lego destroys our pickets? So far, I spotted 2 ironclads, which would easily destroy 2 galleons, leaving a big gap for them to navigate through.

Should we raise this as an issue with lego. Perhaps we could move a ship within landing range of their coast and then negotiate a withdrawal to neutral sea?

Or just let it go and hope for the best...

[Krill]

well, maybe we should do as Theseus last suggested. make a 7 tile border out to sea, and state that we will destroy any ships within the border. Make it public, make it look like Lego is an aggressor.