How to use spy as France

[agonistes]

@Victoria is there a good way to suicide a spy with horrible promotions?

Hmmm, away from home so I can't test, but can you just delete them?

edit: I would not mind you keeping a daily log/report updating us with your spies' activities... its fun to read!

edit2: people who complain the mid and late game get boring maybe aren't playing the espionage game. I just zapped China for over 900 gold (and got partisan upgrade... blah!)

edit3: what conditions must be met for recruit partisans to pop up as a possible mission?

 

[psi_overtake]

what conditions must be met for recruit partisans to pop up as a possible mission?

The city has to have a neighborhood district.

 

[Victoria]

I just zapped China for over 900 gold (

Yep late game gold I am getting tonnes off India, I am trying to break them but they earn 1100 a turn. I think I am seeing them disband the odd jet fighter but am not sure ATM.

The city has to have a neighborhood district.

A great reason never to build a neighbourhood

This is interesting stuff from the randomcalls.csv in the log directory... suggestions?

ESPIONAGE_BONUS_GAIN_SOURCES = 2
ESPIONAGE_ESCAPE_BASE_CHANCE = 10
ESPIONAGE_ESCAPE_COUNTERSPY_LEVEL_MODIFIER = -1
ESPIONAGE_ESCAPE_LEVEL_BOOST = 1
ESPIONAGE_ESCAPE_POLICE_CORRECT_MODIFIER = -4
ESPIONAGE_GAIN_SOURCES_DURATION_MULTIPLIER = 3
ESPIONAGE_MAX_LEVEL = 4

 

[Victoria]

@Victoria is there a good way to suicide a spy with horrible promotions?

... not sure what horrible promotions are, the fact you are no longer at level 1 is a great bonus

 

[TheMarshmallowBear]

I didn't knew the "Spying on us" stacks.

Also, while I like what they have done with Spies this time around (as opposed to before...) I do wish that there were more options to do stuff.

 

[agonistes]

... not sure what horrible promotions are, the fact you are no longer at level 1 is a great bonus

View attachment 483998

Considering you have a limited number of spies, to me a horrible promotion is one you don't want.

 

[Victoria]

So I think I have sussed 2/3 of how it works but cannot get the last bit.
Here is a quick rundown briefly, I’ll type up a guide

Your spy gets a chance of espionage based on its level and this is displayed to you when you accept the mission.
This will at best be 90% success, 7% failure, 3% captured or killed.

It sounds simple but there is a second bit I cannot Identify how it works which is called detection.
This means a mission in fact has 5 outcomes not 3
Success in mission undetected
Success in mission detected, you must try to escape
Failed mission undetected
Failed mission and detected, you just try to escape
Captured/killed ... I cannot work out what the odds are of each, seems to be 50/50

So if you have to try to escape you get a base chance of 10%
Each level of your spy gives 10%
Each level of a counterspy -10%
There is also a police modifier of -40%
+10-40% depending on if you go by foot, car, train or plane
+10% for card modifiers or quartermaster
+20% for the right promotion at the time.

If you escape great, if you do not you are either captured or killed. THIS explains why so many spies die early. There are in fact 2 chances of being killed. Basically escaping by foot means no minuses, try to escape by car or train when you are level1 is pretty much suicide.

Also I have discovered 2 quartermaster values stack so 2 will give all offensive actions (including escapes) +2 levels or 20%

Base cost for a spy is 225 which increases by 75 gold for each spy you have currently completed building. You really do need to use the machiavellianism card and cryptography is also mighty handy. The AI does use both.

Diplomatic penalties for spying stack so spying on allies is useful.
There are logs that show where the AI is positioning spies which is a bit cheaty.
I need more validation on all of this but it seems to be about right.

If you read the messages it clearly shows detection, the second message below is a great example

 

[darkace77450]

Considering you have a limited number of spies, to me a horrible promotion is one you don't want.

A poorly upgraded bird in the hand is worth two optimally upgraded birds in the bush. A spy with undesired promotions is worth more than a perfectly promoted one if it's in the field sooner than later.

 

[myclan]

So I think I have sussed 2/3 of how it works but cannot get the last bit.
Here is a quick rundown briefly, I’ll type up a guide

Your spy gets a chance of espionage based on its level and this is displayed to you when you accept the mission.
This will at best be 90% success, 7% failure, 3% captured or killed.

It sounds simple but there is a second bit I cannot Identify how it works which is called detection.
This means a mission in fact has 5 outcomes not 3
Success in mission undetected
Success in mission detected, you must try to escape
Failed mission undetected
Failed mission and detected, you just try to escape
Captured/killed ... I cannot work out what the odds are of each, seems to be 50/50

So if you have to try to escape you get a base chance of 10%
Each level of your spy gives 10%
Each level of a counterspy -10%
There is also a police modifier of -40%
+10-40% depending on if you go by foot, car, train or plane
+10% for card modifiers or quartermaster
+20% for the right promotion at the time.

If you escape great, if you do not you are either captured or killed. THIS explains why so many spies die early. There are in fact 2 chances of being killed. Basically escaping by foot means no minuses, try to escape by car or train when you are level1 is pretty much suicide.

Also I have discovered 2 quartermaster values stack so 2 will give all offensive actions (including escapes) +2 levels or 20%

Base cost for a spy is 225 which increases by 75 gold for each spy you have currently completed building. You really do need to use the machiavellianism card and cryptography is also mighty handy. The AI does use both.

Diplomatic penalties for spying stack so spying on allies is useful.
There are logs that show where the AI is positioning spies which is a bit cheaty.
I need more validation on all of this but it seems to be about right.

If you read the messages it clearly shows detection, the second message below is a great example
View attachment 484108

The chance of successful escape seems to be much lower than I expected. It seems that even a Master Spy should only try to escape by foot. What does +20% for the right promotion at the time means?

 

[Victoria]

So you have a level 3 spy with +2 for great works heist and a quartermaster but against you is a level2 spy

You succeed in your great work heist but are spotted so need to escape
So you are
1 base
3 for level
2 for great work in use (needs more testing)
1 for quartermaster
4 for escaping on foot
That’s a 110% chance of escape but

Against you are
The police -4
The spy -2

So you have a 50% chance of escape
You could increase it with the Cryptography card or another quartermaster or +4 for the ace driver promotion.

From my anecdotal evidence I would say there is less chance of being spotted if successful.
This all still needs more testing and I am getting to a busy period in work.

In the rand_calls.csv log file I see one roll when not detected indicating detection is all part of the same roll which is weirdly out of 216 rather than 100.
When detected it shows modifiers and another 216 random call for the escape.

 

[Victoria]

Just as an update for anyone interested. I have been doing some tests on stealing gold with a level 1 recruit without modifiers which is 50% 34% 16% and against a civ with 0 spies or defensive cards.
I have recorded 46 attempts ... looking at the randomiser is uses a value of 0-216 for some reason. Below are 2 things of interest

First point, as many people complain about the amount of spies being killed. Below in red are the deaths... 25% and most deaths from successfully stealing gold but being detected.
As a side point, when a detected spy fails to escape, so far they have been killed 100% of the time.

Secondly, the results are indicating what was suspected, to random roll for success failure also seems to be used for detected/undetected as this scatter line graph indicates. So you may successfully steal but probably 50% of the time you will be detected and it seems at least 50% of that time your spy is caught and killed. The advertised 50/34/16 currently looks more like 39/18/43 but its still a low number of results.

I guess the lesson is... never trust a spy

... also.. never trust a website - http://civilization.wikia.com/wiki/Espionage_(Civ6)
"At any turn during or at the conclusion of the mission, whether succeeded or failed, your Spy may be discovered by the police of the target civilization."
I only have ever experienced detection at the end of a mission.

 

[cvb]

One interesting thing is the espionage Random call is a value 0-216 not 0-100 for some reason.

Well, espionage is hardly ever seen on Lets play videos, so I know very little about the mechanics.

Nevertheless your picture at 24 Dec 2017, 03:02 caught my eyes.
E.g. the dots on the 150 line look so quantized, seem to allow "to tell all the discrete numbers" right away. Ask myself whether it is just 2.16*RN, rounded (RN € 1-100) or genuine 216, maybe from 6³. Which different numbers are around 150?

6³ brought me to 3 or 6 ranges, read on, could be 'success undetected, success detected, failure undetected, failure detected, caught, killed'. But it is not 6 equal ranges with 36 values each.

And oddly enough, the highest range DOES include the delimiter 160. As the highest valid value is 216, I would have expected the ranges starting with an odd number and ending with an even; ie. 109-160, 161-216 ... but it is not. ?!?

edit: is 216 really a valid value or is it the upper limit, which is never reached? (ie. Is the range 0-215 or 1-216 or really 0-216, which means 6³+1 ... perhaps you can feel, how much I dislike this value)

The ranges 'success undetected, success detected, failure undetected, failure detected, caught, killed' could be 215-160, 159-108 (half of 216), 107-80 (half of 160), 79-36(?), 35(?)-?, ?-0.

 

[Victoria]

is 216 really a valid value or is it the upper limit, which is never reached? ....The ranges..could be...

@cvb I do need maths minded peole like yourself to help with these things so thank you, its about teamwork. I'm just a donkey.
I will do 100 tests on the 50/34/16 split up to 10 and then validate against another range
The problem with breaking the ranges into groups is there area few of them, so here's a full if not too colorful table of all the values involved. Numbers on left equate to final evel after adjustments for cards/counter-espionage etc.... 216 must be a number they chose for some reason.

 

[agonistes]

@Victoria

1) If I put a spy in City A's market, and City B's market is next to City A, will my City A spy also protect City B's market?
2) If I put a spy in City A's market, and then 2 turns later a different spy gets Seduction, so I put her in City A's industrial zone, next to City A's market, what happens when an enemy spy tries to steal funds from City A's market?

 

[Victoria]

@Victoria

1) If I put a spy in City A's market, and City B's market is next to City A, will my City A spy also protect City B's market?
2) If I put a spy in City A's market, and then 2 turns later a different spy gets Seduction, so I put her in City A's industrial zone, next to City A's market, what happens when an enemy spy tries to steal funds from City A's market?

1. Yes... but not another civs as far as I am aware.
2. Jolly good question... if I am Japan and have 4 districts with 4 spies all adjacent...

I will add them to my list.
I did some more testing and have results but the trouble is collating spy results is a pain.
It looks like about 50% of successful steal yield a detection and about 40% of fails.
I have only tested a 50% gold steal with 0 modifiers.

The escape is much more complex than I thought. There is an addition RND in there and weirdly it’s roughly
0 districts is 1 RND
CH is +2 RND
Harbour seems to be +4 RND
Airport about 4
But not properly tested, way too busy with work.

 

[Trengilly]

I do wish that there were more options to do stuff.

- - - your wish is being granted in R&F. Subvert Loyalty and Sideline Governors are at least two new options.

This all still needs more testing and I am getting to a busy period in work.

Wait you actually have a job and work? I just assumed you were full time dedicated to providing us with Civilization 6 analysis.
Just wanted to say thank you for all your hard work - you've enriched the Civ playing experience for me and I'm sure many other people here at Civfanatics.

 

[Victoria]

Wait you actually have a job and work?

I have been working 18 hour days this week, I had to stop to do the ironing tonight. I blame Americans, they are supposed to be good at computers arn’t they?
Thanks for the thanks but we all do what we enjoy in this life, given half the chance.
.... I’d better post something of use.

 

[Victoria]

So I managed to manuall perform 204 missions against enemies with no spies or defence cards slotted and with my spies staying at level 1 (no promo)
This gives a safer spread than before as follows

A little over 40% chance of being captured or killed as opposed to the quoted 16%

The 16% is not a lie... if we ignore the detection of spies... it validates that the quoted data is about the basic operation before detection. The unknown is a couple of missing values between the two bands. No biggie, just bad luck.

The trouble is whether you succeed or fail you have a chance of being detected
In the legend the U stands for Undetected and D stands for Detected

I do need to try a different % to see if the odds stay the same ... its just a lot of work.I think the detection will go down as level goes up as my experiences in games hint this is likely.
More importantly while we have some idea about the chances and detection the escape roll acts erratically and I still do not have enough numbers to get value
In the RandCalls.csv for espionage you can see on turns 322, 324 and 326 a random roll out of 216 is used and 322 succeeded undetected, 324 was captured and 326 was failed undetected turn 327 was a failed detected result (for a better spy on an easier task which is why the value 31 did not cause a straight capture).... So the first line of 327 is the normal call... when it decides you have been detected it asks for escape on foot, car, boat or plane (always choose foot) ... it decides to roll a randomn range of 7 (The range is 1 for none, 3 for just a CH and 7 for CH & Harbour) and somehow applies this to the next roll

So I had to just take results for just a Commercial hub and always escape on foot .. a RND of 3 (which is 0-2 in value)
The trouble is the results overlap and whil ethe RND could explain this , the values just do not match correctly with the RND roll.
So I need to get lots more values. I would rather Firaxis just told me or hinted as I have spent quite a few hours getting these stats and I really need 6 times as many to get further.
The lines are not a timeline.. they are just showing the amount sorted to make the overlap clear.

If you have read this far you have done very well. The only other thing I have to add is about the logs and slotting cards

There is a log called AI_Espionage.csv that tells you where each AI spy is going and what task it attempts. It is possible to turn on location marks for tiles and know exactly what AI espionage is being done where but thats quite cheaty, interesting to know how the AI is behaving amnd the truth is the AI's are happily spying and sabotaging and stealing each others techs ... other files show lots of deals being done and lots of diplomacy... with bribes for joint wars as I always I suspected.

There is a log called ASI_GovtPolicies.csv which if filtered on slotted only will show what civics are used by each civ each turn of a game. I looked at my game last night and while there were spies aplenty there was not much slotting of Machiavellianism or Cryptography... both vital spy cards. And spied are damn expensive otherwise.

Anyway, it will be a while before I test again. I can provide the stats I got if someone finds them useful