Alpha Centauri 2

Sid Meier's Alpha Centauri & Alien Crossfire => The Theory of Everything => Topic started by: Kirov on December 05, 2012, 11:59:55 PM

Title: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Kirov on December 05, 2012, 11:59:55 PM

As I had to check some numbers for our discussion with Yitzi, I decided to go for it and explore the issue of probabilities of probe team actions. To save your precious time, below I present my conclusions. As you know, the number on the left represents chances of success, the number on the right - chances of survival for the perpetrating probe. Feel free to comment and verify.

MORALE	RISK 1	RISK 2	RISK 3	RISK 4
Disciplined	100/50	50/0	-	-
Hardened	100/67	50/34	-	-
Veteran	100/75	67/50	34/25	-
Commando	100/80	67/60	34/40	-
Elite	100/84	75/67	50/50	25/34

With Algorithmic Enhancement:

MORALE	RISK 1	RISK 2	RISK 3	RISK 4
Disciplined	100/75	75/50	50/25	25/0
Hardened	100/84	75/67	50/50	25/34
Veteran	100/88	84/75	67/63	50/50
Commando	100/90	84/80	67/70	50/60
Elite	100/94	88/84	75/75	63/67

With Algorithmic Enhancement against HSA:

MORALE	RISK 1	RISK 2	RISK 3	RISK 4
Disciplined	50/25	25/0	-	-
Hardened	50/33	25/17	-	-
Veteran	50/37	33/25	17/12	-
Commando	50/40	33/30	17/20	-
Elite	50/42	37/33	25/25	12/17

RISK 1:
- Infiltrate Datalinks
- Procure Research Data
- Activate Sabotage Virus (“Just stir things up in there”)
- Drain Energy Reserves
- Incite Drone Riots
- Engage Mind Control Probe
- Introduce Genetic Plaque,
- also: subversion of a unit

RISK 2:
- Procure Research Data (if twice in the same base, i.e. “high security interlock”)
- Activate Sabotage Virus (“I have a specific target in mind”)
- Assassinate Prominent Researchers
- Attempt Total Thought Control
- Framing other factions for RISK 1 activities
- also: attempt untraceable capture of a unit

RISK 3:
- Activate Sabotage Virus and specifying any kind of target at the enemy’s HQ
- Activate Sabotage Virus and targeting Perimeter Defense or Tachyon Field in any enemy base;
- Framing other factions for RISK 2 activities

RISK 4:
- Framing other factions for RISK 3 activities


EDIT: added tables for Enhancement

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Kirov on December 06, 2012, 12:01:42 AM

Other useful info:

1) If a given probe team can’t undertake a RISK 3 or RISK 4 action, it simply does not get the relevant option. However, there is a bug here and when you specifically target some facilities under Activate Sabotage Virus, you get negative numbers. Disciplined probe teams get 0/-50 and -50/-100, while hardened ones get 0/0 and -50/-33, at RISK 3 and RISK 4 respectively. I’m pretty sure they just represent zero, but it needs checking.

2) To frame a third party, you cannot have Vendetta with the faction you’re just probing. I’m quite sure that if you do have Vendetta, the game decreases your chances as normal, but the frame attempt automatically fails, although that third party does not declare Vendetta upon you (that one last bit needs checking).

3) You can’t frame anybody for Infiltrating Datalinks, not that you would ever want to, as this is the only non-aggressive probe action and has no diplomatic effects.

4) You can’t frame anybody for introducing Genetic Plague. You always take the fall.

5) Targeting Perimeter Defense and Tachyon Field bears additional risk, but this risk is not further increased in HQ (so in HQ all facilities are protected at the same level).

6) I think I heard somewhere that targeting PB silos is also more risky, but my tests do not confirm that.

7) A mission once selected cannot be aborted, with the following exceptions:
- you’re not at Vendetta with the victim
- there is “high security interlock” in place
- at a mind control attempt
- you specify a target of Sabotage; in this case, aborting the mission is rolled as if you selected some target; your probe can die during that abort, and you can even frame somebody for the mission being aborted.

8 ) Mind Control of either a base or a unit is treated like an attack. If you didn’t have Vendetta against the victim, you lose Integrity (but get Vendetta in exchange).

9) Successful Total Thought Control means that you don’t lose Integrity for a sneak attack and you don’t automatically start Vendetta. Basically you take a base and the victim still has a Treaty with you. This option looks pretty cool. Also, if you fail at the Thought Control of a base, your probe gets eliminated, but the victim doesn’t even declare Vendetta for your failed action (this needs further testing). As far as I know, Total Thought Control has no effect if you’re already at Vendetta.

10) The same goes for thought control of a unit, i.e. “attempt untraceable capture”. If you’re successful, no Vendetta is declared, no hard feelings from your victim, for him the unit just disappeared. However, if you fail, the victim declares Vendetta. As above, the untraceable capture has probably no effects when at Vendetta already.

11) You cannot have very green and green probe teams, not even via the scenario editor.

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Kirov on December 06, 2012, 12:02:05 AM

Stuff that requires further research:

1) Mind control formula (I have some clues, so please let me know if you want to explore this issue)
2) Formula for amount of energy stolen in Drain Energy Reserves
3) check if stolen tech is selected completely at random (or maybe lower tech is easier to steal)
4) check if Perimeter Defense and Tachyon Field have the same chances of being targeted at random if you select “Just stir things up in there”.
5) Check under what circumstances the AI does not declare Vendetta even for hostile actions (“I do not condone industrial espionage” or something like that)

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Yitzi on December 06, 2012, 01:16:35 AM

There seems to be a calculator for unit and city subversion cost found here (http://apolyton.net/showthread.php/132830-How-to-calculate-the-cost-of-city-mind-control?p=3848207&viewfull=1#post3848207).  I'll see if I can turn the procedures there into explicit formulae.

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Yitzi on December 06, 2012, 02:09:26 AM

Ok, here's what I've figured out:
-Distance seems to be calculated with the "direct" distances (edge to edge, that's actually diagonal) counting as 1, and the corner-to-corner boundaries counting as 1.5 (round down.)  Perhaps the guys at Firaxis were D&D fans?
-The formula for the cost to subvert a unit (before applying the effects of the target's PROBE rating or algorithmic enhancement) is (target faction's EC+800)X(cost of the target unit in rows)/(4+2Xdistance to enemy HQ).  In particular, this means that if you're going to send your troops far away from home, you'd better either stack them or get a lot of energy to just sit in your bank.  On defense, subversion is a much smaller risk; conversely, it's a lot harder to wage a subversion war against an enemy who's fighting in their own territory.
-For bases, the formula is (target faction's EC+1200)X(population+military units in the base+past mind controls+past unit subversions/4 rounded down)/(4+distance to HQ).  The effective distance is halved by a children's creche, halved by a punishment sphere, halved by nerve stapling, and tripled by a genejack factory.  Drone riots halve the final cost, and a golden age doubles it.  Finally, the result is multiplied by 1 more than the number of bases captured (I presume this means via mind control) this turn.

EDIT: The calculator is missing the effect of past subversions, I'm adding it here.

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Petek on December 06, 2012, 02:30:50 AM

Although it's not documented in the manual, I believe that the cost to mind control a base is reduced for bases that previously belonged to you. I'll test this, perhaps, tomorrow.

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Kirov on December 06, 2012, 02:46:24 AM

Quote from: Yitzi on December 06, 2012, 02:09:26 AM

Ok, here's what I've figured out:
-Distance seems to be calculated with the "direct" distances (edge to edge, that's actually diagonal) counting as 1, and the corner-to-corner boundaries counting as 1.5 (round down.)  Perhaps the guys at Firaxis were D&D fans?

Ahh, now that would explain a lot, wouldn't it? Who thought it's a good idead to count corner-to-corner distances as bigger? What's the deal with D&D? :)

Quote

-The formula for the cost to subvert a unit (before applying the effects of the target's PROBE rating or algorithmic enhancement) is (target faction's EC+800)X(cost of the target unit in rows)/(4+2Xdistance to enemy HQ).  In particular, this means that if you're going to send your troops far away from home, you'd better either stack them or get a lot of energy to just sit in your bank.  On defense, subversion is a much smaller risk; conversely, it's a lot harder to wage a subversion war against an enemy who's fighting in their own territory.

I tested it in my random save and it's slightly off, but not that much. In one example the formula said 514,5, the actual cost was 513. In another it was 168 to 162. Close enough, I just hope the discrepancy doesn't increase at bigger distances or unit costs. Could you verify it in your game?

I remember it was almost the same formula as in Civ2, but now I can't find it. And yes, units venturing far by themselves are at high risk of being subverted for cheap.

Quote

-For bases, the formula is (target faction's EC+1200)X(population+military units in the base)/(4+distance to HQ).  The effective distance is halved by a children's creche, halved by a punishment sphere, halved by nerve stapling, and quadrupled by a genejack factory.  Drone riots halve the final cost, and a golden age doubles it.  Finally, the result is multiplied by 1 more than the number of bases captured (I presume this means via mind control) this turn.

I can't say this formula works for me. Could you run a test? It is also worth noting here the military units must be offensive, i.e. garrison doesn't increase the price (which is why I always believed that units don't affect the cost).

Thanks for the link, I know I haven't encountered any thread with the specific formula, but maybe I missed something, the search engine at Apolyton isn't of much help for me nowadays. That excel file uploaded would be a neat stuff, but it seems sort of unfinished business. The formulas don't work in my savegame, there's stuff I don't quite understand there (what's the deal with the DC 1 and DC2 sheets? ???), and the info in the probe successes sheet is definitely wrong except for the two first columns (which I labeled as risk 1 and risk 2).

And Petek is right, if the base was yours, it definitely affects the price.

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Petek on December 06, 2012, 03:20:53 AM

Quote

the search engine at Apolyton isn't of much help for me nowadays

Yes, the 'poly search engine hasn't worked for several years now. I now use Google, along with the string site:apolyton.net. If your search terms mostly come from SMAC phrases, you get good results. For example,

probe team morale site:apolyton.net

returns lots of relevant results.

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Yitzi on December 06, 2012, 03:40:28 AM

Quote from: Kirov on December 06, 2012, 02:46:24 AM

Who thought it's a good idead to count corner-to-corner distances as bigger?[

Pretty much everyone?  Imagine a normal grid (e.g. graph paper); the distance from (0,0) to (2,2) is further than the distance from (0,0) to (0,2); the former is 2 squares diagonally (corner-to-corner), the latter is 2 squares vertically (side-to-side).

The only reason this is different is that the squares are all rotated 45 degrees, so "side-to-side" ends up being diagonal, and "corner-to-corner" ends up being horizontal.

Quote

What's the deal with D&D? :)

When measuring distances between squares in D&D, you use the same rule: Vertical or horizontal counts as 1, diagonal counts as 1.5.  It's a fairly easy approximation to the actual formula (the Pythagorean theorem.)

Quote

I tested it in my random save and it's slightly off, but not that much. In one example the formula said 514,5, the actual cost was 513. In another it was 168 to 162. Close enough, I just hope the discrepancy doesn't increase at bigger distances or unit costs. Could you verify it in your game?

I already did some testing of my own, and it seems fairly good.  I'm not really interested in doing more testing, though if you tell me which cases gave you something that was off I might be able to figure out what's going on.  Note that his distance calculator seems to have a few bugs, so that might be what's going on.

Quote

I can't say this formula works for me. Could you run a test?

I'd rather not do a full test, but if you can set up a case where it doesn't work I can see if I can figure out what's going on.

Quote

Thanks for the link, I know I haven't encountered any thread with the specific formula, but maybe I missed something, the search engine at Apolyton isn't of much help for me nowadays.

I just do a general google search; most of the hits are Apolyton, and if there's a worthwhile site not at Apolyton that's good too.

Quote

there's stuff I don't quite understand there (what's the deal with the DC 1 and DC2 sheets? ???)

One references and is referenced by the unit cost sheet, the other is for the base cost sheet.

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: t_ras on December 06, 2012, 06:51:26 AM

Good info, thanks.

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: ete on December 06, 2012, 01:01:03 PM

Excellent work, few things to add for completeness:
How/does Polymorphic Encryption or Algorithmic Enhancement change the success rate?
How do different PROBE ratings for the target/user affect success rates?
How does Algorithmic Enhancement vs HSA affect success rate?

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Yitzi on December 06, 2012, 04:42:50 PM

I don't think Polymorphic Encryption or PROBE ratings affect the success rate (except for the effects of PROBE +3 on subversion attempts and morale effects of positive user PROBE rating.)
Algorthmic Enhancement IIRC (assuming that the action would be possible without Algorithmic Enhancement) halves the chance of failure.  (So a 50% chance of autosuccess, 50% roll normally.)  If the target has HSA or +3 PROBE, it gives half the usual chance of failure (so 50% autofail, 50% roll normally as though the target did not have HSA or only had +2 PROBE.)

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Kirov on December 07, 2012, 02:16:55 PM

Quote from: ete on December 06, 2012, 01:01:03 PM

Excellent work, few things to add for completeness:
How/does Polymorphic Encryption or Algorithmic Enhancement change the success rate?
How do different PROBE ratings for the target/user affect success rates?
How does Algorithmic Enhancement vs HSA affect success rate?

Thanks, I completely forgot to test Enhancement info. :) I just did and I updated the first post to make it nice in one place.

As for PROBE ratings - the user's PROBE only affects the chance indirectly, via additional morale bonus to attacker's probe team. There is no additional chance and no bonus effect past Elite.

The target's PROBE doesn't affect the chances of success or survival at all. The only thing it does is +50% to mind control price at +1 PROBE, +100% at +2 PROBE and immunity to mind control at +3 PROBE. This last thing then can be overriden by Algorithmic Enhancement units, which acts at chances like Enhancement vs. HSA, see the first post.

Polymorphic Encryption doesn't affect the chances, it only doubles the price of mind control. Note that the Encryption itself is an ability which comes at some cost in rows, so it actually more than doubles the price (i.e. the price is doubled if you'd exchange the Encryption to any other cost 1 special ability).


EDIT: added info on Encryption.

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Yitzi on December 20, 2012, 11:06:42 PM

By the way, did you also check whether a negative PROBE rating for the target affects the chances?
And what about those techs that are marked as increasing probe team success rate?

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Kirov on December 27, 2012, 06:03:07 PM

Quote from: Yitzi on December 20, 2012, 11:06:42 PM

By the way, did you also check whether a negative PROBE rating for the target affects the chances?
And what about those techs that are marked as increasing probe team success rate?

Negative PROBE rating only affects the price of mind controlled stuff, I just checked it to make sure. Interesting sidenote: it seems that the AI is less likely to go vendetta if you sabotage its facilities ("something something I don't condone industrial sabotage").

Those techs you mention do not increase the success rate, they increase morale. This is what you see when your midgame probes start as veteran. This works normally without any additional bonuses.

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Alpha Centauri Bear on January 08, 2019, 01:59:32 AM

Let me summarize it. Here is the text from PROBE effect description.

-2, -50% cost of enemy probe team actions; enemy success rate increased
-1, -25% cost of enemy probe team actions; enemy success rate increased
0, Normal security measures
1, +1 probe team morale; +50% cost of enemy probe team actions
2, +2 probe team morale; Doubles cost of enemy probe team actions!
3, +3 probe team morale; Bases and units cannot be subverted by standard Probe Teams!!

Based on the information in this post it looks like enemy success rate increased is false. Success rate does not directly depend on own or enemy PROBE rating. It directly depends only on probe team morale which, in turn, depends on own positive PROBE rating and some other things.

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Kirov on February 14, 2019, 11:27:07 PM

I'm quite sure I was thorough in testing that, however at the moment I have no recollection of the relevant tests. But yeah, my gameplay memories don't suggest that the enemies of Zak have it easier. Just cheaper.

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: Alpha Centauri Bear on February 15, 2019, 03:14:13 PM

Yep. Cheaper only for paying options like subversion. Other actions are unaffected.
It would be nice if anyone modded this so success rate depends on difference between your PROBE and enemy's. I.e. your rating increases it (maybe just slightly) and enemy rating decreases it.
Then PROBE would be indeed some important statistics!

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: magic9mushroom on June 07, 2023, 02:25:23 PM

Quote from: Yitzi on December 06, 2012, 02:09:26 AM

The effective distance is halved by a children's creche, halved by a punishment sphere, halved by nerve stapling, and tripled by a genejack factory.

The Crèche sounds okay to me, but the Genejack Factory does not.

With +0 PROBE and 10 credit balance for target and no previous probe actions:

Mind-controlling an empty size-1 base at 2 squares from HQ costs 201 credits normally, 242 with Crèche, 114 with Genejack Factory, 151 with both.
Mind-controlling an empty size-1 base at 12 squares from HQ costs 75 credits normally, 121 with Crèche, 33 with Genejack Factory,  57 with both.

It's not a tripling of the distance - indeed, it can't be any multiplier of the distance, because no number x would fit both 2x = ~6.6 and 12x = ~32. It also can't be a straight multiplier of the cost, because 114/201 and 33/75 are completely different.

Not sure what's going on here, but I wanted to flag it.

Title: Re: Breakdown of chances for probe team actions, per morale and type of mission
Post by: magic9mushroom on July 15, 2023, 07:32:39 AM

Figured it out. GJF doubles distance from HQ, but then also multiplies overall mind-control cost by 3/4 (or something mathematically equivalent; the rounding's different so I suspect it's either "subtract 1/4 of cost" or "multiply (distance + 4) by 4/3 prior to doing the division").