Restart eliminated players: How it works (and how to fight it)

[La Fayette]

Attention please: This thread is a spoiler for those currently playing GOTM36 at Civfanatics.



If the 'Don't restart eliminated players' command has not been clicked upon before starting a new game, the consequence is that every time one destroys the last city of an AI civ, 'respawning' happens: a new settler of the same colour as the civ that has just been destroyed appears somewhere on the map and a new civ is created with a rather good scientific knowledge (number of techs close to the average number of techs owned by the other civs).
This happens until 1500AD at Deity level and later than that at easier levels, and this means that most players (among whom my friend SlowThinker a few weeks ago ) consider that playing early conquest when 'Restart' is on brings only boredom.
But it is not true: it is in fact possible to achieve conquest well before 1500AD, and I shall try to explain why.

1) How it works:

The test I have just run is my GOTM36 game (the Game of the Month that is played each month at Civfanatics with the settings chosen by the moderator, Duke of Marlborough). The game was at Deity level with 'Restart' on and I played conquest after having built Marco, in order to see what would happen.

The AI civs were: Celts, Japanese, Vikings, Persians, Greeks and Sioux (I was Spanish). I conquered them all on the same turn in 920AD. General respawning occured and the 6 new civs that appeared were: Romans, Babylonians, Germans, Chinese, Carthaginians and Mongols.

From then on I decided not to care about destroying all civs on the same turn, since this seemed to have no influence at all upon respawning. I could spot the new civs respawning with help of Marco and destroy them as soon as one of my crusaders met a settler (pure luck) or managed to come close to a new city built by the respawning civ (thanks to Marco).

Here is the list (c = city, u = unit):

Babylonian (u) = Zulu
Chinese (c) = American
German (u) = French
Carthaginian (c) = English
Mongol (c) = Indian
Roman (c) = Russian
Zulu (c) = NO RESPAWN
Russian (u) = Celt
French (c) = NO RESPAWN
American (c) = NO RESPAWN
Indian (c) = Indian
English (c) = NO RESPAWN
Indian (u) = NO RESPAWN
Celt (c) = Russian
Russian (u) = NO RESPAWN

(to be continued next post)

[La Fayette]

2) How it works: Analysis of test results

When the 'Restart ' command is on (before 1500AD at Deity level), any AI civ that gets destroyed will respawn twice, until the 3 civs of the same colour have appeared (for example Celt = Roman = Russian in my test).

THEN:
- either 'No more respawn' (4 times out of 6 in my test)
- or 'One more respawn' (twice out of 6 in my test: Russian = Celt and also Indian = Indian)

THEN:
- either 'No more respawn' (once out of 2 times in my test)
- or 'One more respawn' (idem)

THEN:
'No more respawn'
'Your civ has conquered the entire world'

3) How to fight it:

Destroying all AI cities on the same turn has no influence at all.
Destroying a settler or a city are the 2 possibilities offered and have exactly the same consequence (AI civ destroyed).
You will at least have to face 2*(number of AI civs) respawnings ('basic respawning') + a few more ('residual respawning').

Therefore, the shortest way to global conquest requires building MPE (or having built it previously) + having a good transportation network (LH + Magellan, if archipelago map, or many roads, if continental map) + having strong units all over the world (single units are OK, since you will be destroying only settlers and size 1 cities).

[Old n Slow]

La Fayette,

I recall a post by Smash in an old GOTM game (between 1-2 years ago I guess) stating that he had to found a few more cities to nab the upstarts. I think that each respawn needs a certain amount of area (I recall another post showing three settlers of different flags unable to plant cities due to their proximity to each other.) AND be a certain distance from some other city.

One wonders if a player can "herd" respawns to an extent by locating outposts on major continents leaving few desirable grounds for planting cities.

[Elephant]

Not suprised about the first two respawns: they are the unused civs for each color. The number of repetitions after they are exhausted is the main question. Is it totally random, or is there a formula for how many extras, perhaps based on date before 1500AD? It would take a LOT of trial games to figure that out...

BTW, you should put "CFC GOTM 36 SPOILER" at the head of your first post...

[La Fayette]

OnS

I have also read a number of posts about the influence of building new cities or spreading new units here and there. None of them have explained to me how to stop respawning. Neither have I, and I think that respawning is unavoidable.

What I hope to have demonstrated is that respawning is very limited after the first 2 respawns corresponding to the 2 civs with the same colour as the one you destroyed first (what I called 'basic' respawning).

I agree with Elephant: it would be interesting to know whether what happens next (what I called 'residual' respawning) is purely random or not. But the practical value of that kind of knowledge would be limited IMO.

I hope to have demonstrated that, if there are n AI civs in your game, you need to destroy very probably more than 3n cities (or units) and very probably less than 4n cities (or units).

You are therefore sure to be able to win before 1500AD if you manage to know where to find them (= MPE) and how to kill them quickly ( = LH, Magellan, roads or even RR, ... and of course many strong units).

Elephant

Thank you for the advice: I have put a Spoiler Warning on top.

[Scouse Gits]

Now I am confused - I am certain that about 18 months ago Dr Fell demonstrated his evil genius by trouncing all opposition in an early conquest using the 'kill them all in the same turn' strategy -- where is the difference/distinction -

Stu

[duke o' york]

Could you link to that please SG? Don't spend too long looking for it though.

I have always understood that destroying them all on the same turn would give you a victory, but have just thought that maybe this would be affected by the colour of your civ.
I'm also convinced that the AI cannot respawn within a certain distance of one of your cities or units - it would be extremely unfair if they restarted within an attack's distance of one of your troops. This would mean that you can restrict their respawning to certain areas of the map, as the AI will always respawn somewhere away from your presence, but found somewhere with the possibility of growth. They'll just wander for years otherwise. That's why some games will have respawned settlers stuck upon the poles and they will never found a city, but you can't win until they've been "dealt with".

[Zenon]

Well done, La Fayette! Finally some new and useful knowledge in this tricky testing field. When I saw in my tests that not only the same colour but also the same civ would respawn I didn´t research further. You have done better!

Some remarks on other theories that "respawn" from time to time: killing all civs in the same turn does not prevent respawning. All tests indicate it. Respawned civs can appear right next to your troops and there is no indication that your units influence respawns. Respawns are likely to appear far from other cities, but I have seen them appear within city limits when the map was occupied by many cities.

Zenon

[La Fayette]

SG(1) and duke o'york,

Dave V and I also have achieved victory after destroying all civs on the same turn, so DrFell is not alone, but there are many tests and examples showing that this is not the rule.

duke o'york,

Would you agree with me if I state that 'sending the AI to the pole' is a very bad idea ? IMO it is much wiser to let the AI respawn close to your troops and kill them ASAP than chase them and make it more and more difficult to find them.

zenon,

Thank you for kind congratulations
(BTW I agree with all the points that you mention)

[Bloody Monk]

Good work, La Fayette!!

I have suspected for some time that there was some "method" to the respawn question. I always played with restarts off. The first time I seriously looked at this was June-ish, 2002, in a game sponsored by Campo. The "Campo Challenge" also had restarts on and was a conquest game.

What I remember from that game was that destroying all civs in one turn did NOT give the usual game ending victory graphics. We all "decided" on the fly that killing all civs in one turn would constitute a victory, no matter that the game respawned them next turn. Unfortunately, nobody thought to take the next steps as you just did.

Testing this would take many separate starts played to the end, to eliminate any one start anomalies; a very big undertaking. This may be a case where "knowing" something less than totally is enough.

You have provided a format for dealing with this problem. Thanks.

Monk

[Elephant]

Perhaps another "Conquest Respawn Challenge" with multiple players will get enough data points for us to figure out a pattern?

I'd also suggest that anyone playing Conquest with Restart On (or who has a recent game log that they can post or summarize for us) post the details of their own experiences here, so we can collectively build the data set to see if there is a general pattern after all the "initial respawns" are exhausted.

Primary data requested: Difficulty level of game, Human players color, Do you have MarcoPolo or UN?, Date and CivName of each AI civ destroyed, and Date of final victory.

And another trick worth trying: has anyone tried killing all 6 AI civs at the same time, three times in a row?

[Grigor]

This doesn't seem right to me. I was playing GOTM 43 at Civ Fanatics, and when I killed off the Sioux (the original purple civ) they did not respawn (lucky for me!).

I think I recall somebody suggesting that the mechanism might be a percentage chance for respawning which either remains constant or diminishes with repeated respawning. That seems more like the way the game engine likes to do things - such Samson's hut outcome analysis.

[La Fayette]

So Grigor was the one who knew where this thread was :
Thank you, Grigor!

Here are my recent *findings*:

1) No respawn (or one respawn) does happen now and again (exactly what Grigor just reported).
2) Marco allows contact with the new settler (this means you can spot the new civ before the first city is built).
3) Respawn of the same civ, or respawn #3 or #4, generally seem to occur close to the previous location.

Here is a new summary of my advice:

1) We don't know exactly how it works (and we probably won't know until samson comes back ... if he does ).

2) But this is not really annoying for someone making use of my advice:
- Build Marco (which should allow you to know where they are immediately after respawning).
- If you are short of techs, give money; if you are short of money, give units; they are so weak that you run no risk giving anything you own.
- Spread strong units all around the map, in order to get to them ASAP (one vet crusader, perhaps onboard a trireme, should be enough in most cases).

[solo]

I have had some time to run a few tests and have concluded that respawns are random events. More tests are needed to find out the probabilities.

Many report respawns in identical places, which suggests that the same algorithm is used by the game each time a newly respawned civ is placed. This would be good news for killing off respawning civs. Corner the last city of each civ and kill off just one. Find where it respawns (if it does), and kill it. If found quickly enough and if that color respawns again, it is likely to do so in the same place, where it can be re-destroyed until the that color no longer respawns. Then kill off another civ on a following turn and repeat the process until all 6 colors are gone. I believe that the spot chosen for respawns will stay the same as long as no major changes occur on the map.

[La Fayette]

Respawn at almost the same location is point #3 in my post above: I stronly agree